HBHB ®tfe ^, p, ^tll plirarg QH566 V96 .^.} i'ii> '«!, NORTH CAROLINA STATE UNIVERSITY LIBRARIES S02514748 V Date Due VOL. I. PI. ! % 7 OENOTHERA LAMARCKIANA, A Mutating Species. M^tinHood&LarkiniittLondon.WC. VOL. I. PL II. OENOTHERA GIGAS Originated in 1895. M:artin.Hood&Larkiniitli.London,WC. VOL. I. PL III. OENOTHERA ALBIDA, Produced Yearly by the Parent-Species. Martin, Ho ad &LarkinJ.ith,Loiidon.W.C. VOL. I. PL IV. A MUTATION IN A FAMILY OF OENOTHERA LATA. Origin of Oenothera albida. Martin.Hood&Larldn^Litt.London.W.C. VOL. I. PL V. I OENOTHERA SCI NTI LEANS. od &LarkinLitli,London.W.C . VOL. I. PI. VI. ' i J^^. OENOTHERA OBLONGA. Ptin.Hocii&l,arkiniith,LoiidDn,W.C. THE MUTATION THEORY EXPERIMENTS AND OBSERVATIONS ON THE ORIGIN OF SPECIES IN THE VEGETABLE KINGDOM BY HUGO DE VRIES PROFESSOR OF BOTANY AT AMSTERDAM TRANSLATED BY PROF. J. B. FARMER AND A. D. DARBISHIRE VOLUME I THE ORIGIN OF SPECIES BY MUTATION WITH NUMEROUS ILLUSTRATIONS AND FOUR COLORED PLATES CHICAGO THE OPEN COURT PUBLISHING COMPANY LONDON AGENTS KEGAN PAUL^ TRENCH, TRUBNER & CO., LTD. 1909 COPYRIGHT BY The Open Court Publishing Co. 1909 AUTHOR'S PREFACE TO THE TRANSLATION. The promulgation of the principle of unit-characters is the main theme of this work, as is emphasized in the first sentence of the Introduction. At the time of the publication of the first part of the German edition (1900) this principle was new and was also in evident opposition to the current belief in the slow and gradual evolution of the organic world. During the years that have since elapsed, it has gained almost universal acceptance, though there are still some authors, especially among zoologists, who are opposed to it. The evidence which supported this view was derived from three main sources. First a clearer understanding of the processes of selec- tion in agricultural plant-breeding. This conception has since been corroborated in the most convincing manner by the work of Nilsson and of Korschinsky^ and it points to the elementary species as the real material for artificial and natural selection. Secondly, the experimental evidence afforded by the evening primroses and some other groups of plants ; espe- cially the observed origin of Oenothera gigcts, which ap- peared suddenly in my cultures in the year 1895 and pos- sessed, at its first origin, all the attributes of a new species, including constancy and even a double number of chromo- somes in its nuclei. Thirdly, the new light, thrown by the principle of the unit-characters on the work of Mendel, neglected up to \'-ji y iv Author s Preface to the Translation. that time, which even Focke in his standard work on Plant-hybrids did not give the rank of a work of first-rate importance. The work of Bateson and of his school, of CuENOT, Webber and many others, but above all that of Davenport, have since brought the principle of unit-char- acters to its now prominent rank in the study of hybridi- zation. There can be no doubt, that this principle of the unit- characters opens up a wide range of questions and subsidiary theories, which may now be subjected to experimental in- quiry and critical study. The phenomena of inheritance and hybridization constitute a wide field, which had scarcely been explored hitherto. Half- and middle-races, with their apparently incomplete heredity, constant hybrids, correlated and associated characters, and many other most curious phe- nomena afiford plenty of scope for future investigation. Under these circumstances I feel deeply indebted to Pro- fessor Farmer and Mr. Darbishire for their painstaking work in preparing an English translation of this book, as well as to Dr. Paul Carus of the Open Court Publishing Co. for his liberality and full confidence in the scientific and practical value of the principles enunciated therein. Great support has already been accorded to my ideas by many English and American workers in this field, and it is con- fidently hoped, that this translation will secure that uni- versal cooperation without which no great scientific principle can attain its full measure of usefulness to mankind. No essential changes have been made in the translation, with the exception of those which have been made necessary by the work of Hj. Nilsson on the selection and the im- provement of cereals in Sweden. Corrections of minor points have been introduced wherever necessary. It is proposed to publish this translation in two inde- pendent volumes, the first dealing Avith the origin of spe- cies by mutation, the second with the origin of varieties and with the general considerations found at the end of the Author's Preface to the Translation. v German edition. Some chapters, especially among those on hybridization, which seem to be of too technical a nature for the general student, will be omitted from the second volume. It is proposed to publish their translation in a separate work. Amsterdam^ June 1908. Hugo de Vries. TRANSLATORS' PREFACE. The task of preparing' this translation has been made Hghter by the knowledge that the need for it is urgent. Professor De Vries's successful attempt to bring the pro- cess of specific differentiation within the sphere of experi- mental inquiry is now recognized as a landmark in the history of our knowledge of these phenomena. But those who take part in the discussion of evolutionary questions are rarely equipped with even a superficial familiarity with the broad features of Professor De Vries^s investigations and ideas, and, still less, with (what should be a minimum qualification for a participant in such discussions) a de- tailed knowledge of the contents of Die Mutationstheorie. It is hoped that this translation will help to remedy this state of affairs. This much, at any rate, is certain, that the evidence, collected by De Vries up to 1901, bearing on the question of the origin of species and varieties by mutation, is now for the first time available to the student of evolution who cannot read German. In the translation itself we have endeavored to convey the author's meaning as faithfully as possible rather than to provide a word for word translation of the German. And to this end we have, wherever it seemed necessary, split sentences into two, or run two into one, or made other such additions and omissions as seemed desirable. All of these alterations have been examined and approved by the author. Translators' Preface. vii We hope that our translation of this great work will help to bring us closer to an understanding of one of the most puzzling manifestations of vital activity — that of spe- cific diversity. J. B. Farater. A. D. Darbishire. PREFACE TO THE FIRST VOLUME. The origin of species has so far been the object of com- parative study only. It is generally believed that this highly important phenomenon does not lend itself to direct obser- vation, and, much less, to experimental investigation. This belief has its root in the prevalent form of the con- ception of species and in the opinion that the species of animals and plants have originated by imperceptible grada- tions. These changes are indeed believed to be so slow that the life of a man is not long enough to enable him to witness the origin of a new form. The object of the present book is to show that species arise by saltations and that the individual saltations are occurrences which can be observed like any other physio- logical process. Forms which arise by a single saltation are distinguishable from one another as sharply and in as many ways as most of the so-called small species and as many of the closely related species of the best systematists, including Linnaeus himself. In this way we may hope to realize the possibility of elucidating, by experiment, the laws to which the origin of new species conform. The results of these studies can then be compared with those which have been obtained with systematic, biological and particularly with palseonto- logical data. A most remarkable agreement will be found to exist between these and my new results. These saltations, or mutations, of which the so-called Preface to the First Volume. ix sports are the best known instances, constitute a distinct province in the study of variabiHty. They occur without transitional gradations and are rare ; whilst ordinary varia- tions are continuous and always present. The whole subject of variability, therefore, falls into two sections, one of which includes the ever present, indi- vidual or fluctuating variability, whilst the other embraces mutability. The former phenomena conform to the well- known laws of probability and are determined by general nutritional conditions ; they also afford the material for the production of many of the so-called improved races of agri- culture. Mutations give rise not only to species but also to varie- ties ; and, as has been recognized for a long time, they play an all-important role in horticulture. An exhaustive com- parative and experimental study of horticultural varieties is an indispensable preliminary to a complete treatment of the problem of the origin of new forms. It will be given in the second volume. The generalizations here outlined apply obviously to animals as well as to plants. Though as a botanist I have confined my attention to the latter, I am convinced that my results will be confirmed in the realm of the animal kingdom. Again a proper distinction between variability and mutability is of the greatest importance from the point of view of the application of the results of biological investi- gation to the solution of sociological problems. For, the question of the origin of species has really very little to do with these highly important problems ; whilst that of fiuc- tuating variability is intimately and fundamentally bound up with it. The contrast between these two groups of phenomena, variability (in the strict sense) and mutability, becomes ob- vious when we imagine that the properties of organisms are built up of perfectly distinct and independent units. The origin of a new unit is a mutation ; but the new unit X Preface to the First Volume. varies in the degree of its manifestation according to the same laws as those to which the elements of the species, already existing, conform. The properties of these units can be studied far more conveniently by means of experiments in hybridization than by merely observing or rather waiting for their origin. On the basis of this principle the most complicated phenomena of hybridization must be explained by means of the results of the simplest crosses. For, by a combination of these simpler processes we may expect to arrive at an elucidation of the laws to which the phenomena of hybridization con- form, and ultimately be in a position to predict the result in special cases. In this way the application of the theory of mutation to the elucidation of the phenomena of hybridi- zation will enable us to ascertain what conclusions relating to the origin of species the study of these processes may warrant. A knowledge of the laws of mutation must sooner or later lead to the possibility of inducing mutations at will and so of originating perfectly new characters in animals and plants. And just as the process of selection has en- abled us to produce improved races, greater in value and in beauty, so a control of the mutative process will, it is hoped, place in our hands the power of originating permanently improved species of animals and plants. Hugo de Vries. Amsterdam^ August 1901. CONTENTS. PAGE Introduction 3 Erratum. Figure 38, page 191 should read figure 37. 1. The Transmutation Theory Before Uakwix lu 2. Darwin's Selection Theory 28 3. Wallace's Selection Theory 39 4. The Various Forms of Variability 43 Polymorphism, 45 ; Individual or Fluctuating Varia- bility, 47 ; Spontaneous Variations, 53. 5. The Elements of the Species 56 The Variation of the Elementary Specific Characters, 60. 6. The Mutation Hypothesis 62 Historical Review, 62, Bateson, 63 ; Scott^ 66 ; Kor- SCHINSKY^ 68. III. Selection Alone Does Not Lead to the Origin of New Species .'. 71 7. Selection in Agriculture and Horticulture 71 Maize, 71; Results of Chance Crossings, yj \ Novel- ties, 78. 8. Selective Breeding Followed by Vegetative Propagation 82 Regression, 83. 9. On the Duration of the Process of Selection 85 10. Acclimatization 92 American Maize in Baden, 95; Degeneration of Races, 97. 11. Sugar Beets 99 X Preface to the First Voliune. varies in the degree of its manifestation according to the same laws as those to which the elements of the species, already existing, conform. The properties of these units can be studied far more conveniently by means of experiments in hybridization than by merely observing or rather waiting for their origin. On the basis of this principle the most complicated phenomena of hybridization must be exj^lained by means of the results of the simplest crosses. For, by a combination of these ^_j,.xx v^i cpv^x^ics luc sLuuy oi tnese processes may warrant. A knowledge of the laws of mutation must sooner or later lead to the possibility of inducing mutations at will and so of originating perfectly new characters in animals and plants. And just as the process of selection has en- abled us to produce improved races, greater in value and in beauty, so a control of the mutative process will, it is hoped, place in our hands the power of originating permanently improved species of animals and plants. Hugo de Vries. Amsterdam, August 1901. CONTENTS. PAGE Introduction 3 PART I. THE PRINCIPLES OF THE CURRENT THEORY OF SELECTION. A REVIEW OF THE FACTS. I. Selection and AIutation n II. Mutability and Individual Variation 16 1. The Transmutation Theory Before Darwin 16 2. Darwin's Selection Theory 28 3. Wallace's Selection Theory 39 4. The Various Forms of Variability 43 Polymorphism, 45 ; Individual or Fluctuating Varia- bility, 47 ; Spontaneous Variations, 53. 5. The Elements of the Species 56 The Variation of the Elementary Specific Characters, 60. 6. The Mutation Hypothesis 62 Historical Review, 62, Bateson, 63 ; Scott, 66 ; Kor- SCHINSKY,, 68. III. Selection Alone Does Not Lead to the Origin of New Species .'. , 71 7. Selection in Agriculture and Horticulture 71 Maize, 71; Results of Chance Crossings, 77; Novel- ties, 78. 8. Selective Breeding Followed by Vegetative Propagation 82 Regression, 83. 9. On the Duration of the Process of Selection 85 10. Acclimatization 92 American Maize in Baden, 95; Degeneration of Races, 97. 11. Sugar Beets 99 xii Contents. PAGE 12. Cereals io6 Pedigree and Elite, no; Nilsson's Results, 114. 13. The Limits to the Amount of Change that Can be Ef- fected by Selection 118 Linear Variation, 118; Regression, 120; Instability of Improved Races, 120; Adaptation, 121. 14. The Behavior of Improved Races After the Cessation of Selection 122 Progeny of the Original Seed, 126; Change of Seed, 126; Intermediate Generations, 128. IV. Controversial Questions 130 15. Acquired Characters and Variations Caused by Nutri- tion 130 16. On the Inheritance of Acquired Characters 135 Papavcr somnifcrum polycephalum, 138; Selection and Nutrition, 142. 17. On Partial Variability and Selection by Vegetative Methods of Propagation 143 Alpine Plants, 145; Sugar Cane, 148. 18. Variation and Adaptation 149 19. Variability in Man and Social Questions 154 20. Some Subjects for Future Investigation 159 Correlative Variation, 160; External Conditions, 161; Regression, 161 ; Retrogressive Selection, 163. V. The Origin of Species by Mutation 165 21. Species, Subspecies and Varieties 165 Varieties, 169; Elementary Species, 171. 22. Species in Nature 172 22). Species in Cultivation 176 Age of Races, 176; Cereals, 178; Apples, 179; Age of Garden Varieties, 183. 24. Species and Specific Characters 185 25. Mutations in Cultivation 187 Historical Accounts of the Origin of Species, 188; Sterile Forms, 195 ; Constancy from Seed, 196. 26. The Hypothesis of Indiscriminate Mutability 198 27. The Hypothesis of Periodic Mutability 205 The Migration Theory, 206. 28. The Phenomenon of Mutation Within the Limits of the Mutation Periods 207 Delboeuf's Law, 208. VI. Conclusion 211 Contents. xiii PART II. THE ORIGIN OF ELEMENTARY SPECIES IN THE GENUS OENOTHERA. PAGE The Pedigree Families 217 1. Oenothera Lamar ckiana, a Mutating Plant (Plate I) . . 217 2. The Lamarckiana-Family 221 Pedigree, 224. 3. The Mutations in the Lamarckiana-Family 226 O. gigas, 226; O. Albida, 229; O. ruhrinervis, 230; O. oblonga, 234; O. nanella, 235 ; O. lata, 239; O. scin- tillans, 243. 4. The Laws of Mutation 247 Sudden Appearance, 248; Constancy, 249; Elemen- tary Species, 251; Indefinite Direction of the Muta- tions, 255. 5. A Branch of the Lamarckiana-Family 259 Pedigree, 262. 6. The Laevifolia-Family 265 Original Locality, 266; Pedigree, 273. 7. Two Lata-Families ( Plate IV) 27S O. ruhrinervis and O. oblonga from O. lata, 283 ; Pedigrees, 285, 288. 8. Mutations in Other Families 289 Characters of the Leaves, 293-295 ; ]\Iutants from Crosses, 298-299. 9. Mutations in Nature 300 Beginning of the Mutation Period, 306. II. The Origin of Each New Species Cosidered Separately. 308 A. The Two Older Species 308 10. Oenothera laevifolia 308 Crumples on the Leaves, 310. 11. Oenothera brevistylis 315 B. The Constant New Species 318 12. Oenothera gigas (Plate II) 318 13. Oenothera rubrinervis 3^7 Units of the Characters, 328. 14. Oenothera oblonga (Plate VI) 337 Mutation Coefficients, 337. 15. Oenothera albida (Plates III and IV) 349 16. Oenothera leptocarpa 353 xiv Contents. PAGE 17. Oenothera semilata 358 18. Oenothera nanella (O. Laniarckiana nanella) 360 Conception of Variety, 360; Dwarf Forms, 361; Atavism, 2>^2) \ Constancy, 372 ; Compound Types, 375. C. The Inconstant Species 277 19. Oenothera scintillans (Plate V) ^^yy Unfit Types, 381. 20. Oenothera elUptica 393 21. Oenothera sublinearis 399 D. The Sterile Species 402 22. Oenothera lata 402 Units of the Characters, 402. 2S. Incipient Species 416 Sterility, 417; O. spatliulata, 419; O. fatua, 420; O. sitbovata, 420. III. The Systematic Value of the New Species 425 24. The Nature of the Boundaries Between Related Spe- cies 425 Transgressive Variability, 426. 25. Transgressive Variability 430 Of the Petals, 433; Of the Fruits, 436. 26. Oenothera Laniarckiana seringe 437 Seeds, 437; Species of Ouagra, 438; Mutation Period of O. biennis, 440; Diagnosis, 441 ; O. grandi flora, 441. 27. Synopsis of the Characters of the New Species 444 Fruits, 446-447; Analytical Tables, 448-454. 28. Comparison of the Characters of the Old and New Species 454 IV. On the Latent Capacity for Mutation 462 29. Repeated Mutations Are the Result of the Same Inner Causes 462 30. The Latent Inheritance of Other Characters in O. La- ■ marckiana 468 Pitchers, 470; Tricotyly, 474; Fasciation, 476; Varie- gation of Leaves, 480; Polymery, 481; Other Devia- tions, 484. 31. The Hypothesis of a Premutation Period 490 Oenothera biennis, 495. V. Conclusion 497 Ways to Look for Mutable Plants, 497; Cultures, 502; Intermediate Forms, 504; Species or Varieties, 506; Other Periods of Mutation, 510. Contents. XV PART III. NUTRITION AND SELECTION. ' PAGE I. Simultaneous Influence of Nutrition and Selection ON Various Characters 515 1. Variability as a Nutritional Phenomenon 515 Sensitive Period, 521 ; Nutrition of the Mother- Plant, 522 ; Selection and Nutrition, 523. 2. Methods of Investigation 523 11, The Length of the Fruit of Oenothera Lamarckiana. . . 528 3. Correlation Between Individual Strength and Length of Fruit 528 4. The Simultaneous Operation of Nutrition and Selec- tion 536 Long-fruited Race, 544; Short-fruited Race, 546. 5. Shifting of the Curves of Variability by Nutrition..., 551 III, Curves of Ray Florets of the Compositae and of the Rays of Umbels in the Umbelliferae 556 6. Obliteration of the Effect of Selection by Nutrition. . . 556 Anethum, 559. 7. Equilibrium Between the Effects of Selection and Nu- trition 562 Chrysanthemum, 563; Coreopsis, 566; Bidcns, 567. 8. Obliteration of the Effects of Nutrition by Selection. . . 569 Coriandnim, 569; Madia, 571. 9. Summary 573 Index 577 LITERATURE. EARLIER STUDIES AND PRELIMINARY NOTES OF THE AUTHOR. a. Intracellulare Pangenesis. Jena, 1899. b. Fluctuating Variability. Ueber halbe Galtoncurven. Ber. d. d. bot. Ges., 1894, Bd. XII, Heft 7. — Bot. Jaarboek, 1895, \TI, p. 74. — Archiv. Need., 1895, T. XXVIII, p. 442. Eenheid in Veranderlykheid. Album der Natuur, 1898. — Revue de I'uni- versite de Bruxelles, 1898, T. Ill, p. 5. — University Chronicle, 1898, I, p. 311. Alimentation et Selection. Vol. Jubil. Societe biol. Paris, 1899, p. 17. — Biol. Centralblatt, 1900, XX, No. 6. , Othonna crassifolia (L'Othon) Botan. Jaarboek, 1900, XII, p. 20. c. Mutability. Over steriele Maisplanten. Bot. Jaarboek, 1889, I, p. 141. Steriele Mais als erfelyk ras. Bot. Jaarboek, 1890, II. p. 109. Sur I'introduction de I'Oenothera Lamarckiana dans les Pays-Bas. Ned. Kruidk. Archief, 1895, VI, p. 4. Sur I'origine experimentale d'une nouvelle espece vegetale. Cps. rs. de I'Acad. de Paris, 1900. Sur la mutabilite de I'Oenothera Lamarckiana. Cps. rs. de I'Acad. de Paris, 1900. Recherches experimentales sur I'origine des especes. Revue generale de Botanique, 1901, T. XIII, p. i. Die Mutationen und die Mutationsperioden bei der Entstehung der Arten. Vortrag in der Naturf.-Vers., Hamburg, 1901. Leipsic, Veit & Co. d. After 1901. Species and Varieties; Their Origin by Mutation, ist ed., 1905; 2d ed., 1906; Chicago, The Open Court Publishing Co. Plant-Breeding. Comments on the Experiments of Nilsson and Burbank, 1907. Chicago, Xhe Open Court Publishing Co. Ueber die Dauer der Mutationsperiode bei Oenothera Lamarckiana. Ber. d. d. bot. Gesellsch., 1905, XXIII, Heft 8. Die Svalofer Methode zur ^'eredlung landwirthschaftlicher Kultur- gewachse und ihre Bedeutung fiir die Selektionstheorie. Archiv fiir Rassen- und Gesellschaftsbiologie. 3. Jahrg., Heft 3, 1906. Elementary Species in Agriculture. Proceedings American Philosophical Society, Vol. XLV, 1906, April 18. Aeltere und neuere Selektionsmethode. Biolog. Centralblatt, Bd. XXVI, Nos. 13-15, 1906. Die Neuziichtungen LuTHER Burbanks. Biolog. Centralbl., Bd. XXVI, No. 19, Sept. 1906. INTRODUCTION. INTRODUCTION. By the Mutation theory I mean the proposition that the attributes of organisms consist of distinct, separate and independent units. These units can be associated in groups and we find, in alhed species, the same units and groups of units. Transitions, such as we so fre- quently meet with in the external form both of animals and plants, are as completely al)sent between these units as they are between the molecules of the chemist. It is perhaps unnecessary to remark that these gener- alizations refer to the animal as well as to the vegetable kingdom. In this book, however, I shall confine myself to the latter, in the belief that the truth of the principle will be granted in the case of the former as soon as it has been shown to apply in that of plants. The adoption of this principle influences our attitude towards the theory of descent by suggesting to us that species have arisen from one another by a discontinuous, as opposed to a continuous, process. Each new unit, forming a fresh step in this process, sharply and com- pletely separates the new form as an independent species from that from which it sprang. The new species ap- pears all at once ; it originates from the parent species without any visible preparation, and without any obvious series of transitional forms. 4 Introduction. The Mutation theoiy affects not only our views on the origin of species but in my opinion bears strongly on the whole question of hybridization. For it shows us that the units with which we deal in hybridization are not the species themselves but the single characters which compose them — the so-called elements of the species. This principle leads to an entirely new method of hand- ling the subject, by enabling us to proceed gradually from the simpler to the more complicated phenomena instead of following the present custom which consists in dealing with the complex cases first. This work therefore falls into two main parts of which the first treats of the origin of species and varie- ties by Mutation, and the second of the principles of hybridization. The Mutation theory is opposed to that conception of the theory of selection which is now prevalent. Ac- cording to the latter view the material for the origin of new species is afforded by ordinary or so-called in- dividual variation. According to the Mutation theory individual variation has nothing to do with the origin of species. This form of variation, as I hope to show, cannot even by the most rigid and sustained selection lead to a genuine overstepping of the limits of the species and still less to the origin of new and constant char- acters. Of course every peculiarity of an organism arises from a previously existing one ; not however by ordinary variation, but by a sudden though minute change. It is perhaps appropriate to compare such a change with a chemical substitution. The name I propose to give to this "species-forming" variability is Mutability — a term in general use before Introduction. 5 Darwin's time. The changes brought about by it, the Mutations, are phenomena as to the exact nature of which we understand very httle so far. The best-known examples of such Mutations are the so-called spontaneous variations (the "single variations" of Darwin) by which new and distinct varieties arise. They are also termed, fitly enough, sports. In spite of the fact that they occur fairly often, they are usually not noticed until the new form has already appeared, when of course it is too late to study the phenomenon of its origin experimentally. These new forms can be sought for in cultivated species, which are seldom of pure origin ; as well as in Nature. But as yet we have no power of inducing them at will. It is my belief that all the simple characters of ani- mals and plants arise in this way. . The methods of artificial selection correspond to these two types of variability. Ordinary variation, which is also known as individual, fluctuating or gradual varia- tion, is always present ; and it can be described in terms of perfectly definite laws which have now been fairly completely formulated. It provides the breeder with material for his improved races. On the other hand he has to deal with Mutations which do not need repeated selection but, at the most, must be kept free from ad- mixture, and which almost always breed true from the first. Under the general term variation, then, are included two distinct phenomena : Mutability and fluctuation or ordinary variation. The latter forms a suitable object for statistical investigation. The epoch-making re- searches of QuETELET and Galton on the anthropo- logical side have raised this study to the position of an independent science. Among biologists, Ludwig^ Wel- 6 Introduction. DON, Bateson, Duncker, Johannsen, Macleod, and others, have been active workers in this field. Fluctua- tion is either individual or partial : in the former case we are dealing with the statistical comparison of differ- ent individuals; in the latter with different but homolo- gous organs of the same individual; for example with the leaves of a tree. In both cases the capacity for variation is regarded by those who are competent to judge as a means of adaptation to the environment. Single organs vary partly in mass and weight and partly in number. The former case is referred to by Bateson as continuous variation ; the latter as discontinuous. But these terms are sometimes used by other authors with a different meaning. The laws of Mutability are quite different from those of individual variation; but, so far as our scanty infor- mation reaches they are just as independent of the mor- phological nature of the mutating organ. We can dis- tinguish between progressive and retrogressive Mutation. The former results in the origin of a new character ; the latter in the loss of one already existing. It is, obviously, to progressive Mutation, according to this theory, that the main branches of the animal and vegetable genea- logical tree owe their development; but the great major- ity of the cases of the departure of a single species from the type of the systematic group to which it belongs is due to retrogressive Mutation. It is to considerations of this kind that the first part of this volume will be devoted. In the first place I shall give a critical revision of the facts on which the theory of Natural Selection of Darwin and Wallace and others is based. In the second I shall deal with some examples of the experimental study of new forms. Introduction. 7 The experiments to this end were begun in the autmnn of 1886 and are now at least in one particular direction almost complete. A description of them will constitute most of the contents of the second part. The critical revision to which I have referred will form the substance of the first section. I shall confine my critique to the facts of selection and to the material, afforded by variability, on which selection operates. It will be shown that artificial selec- tion is, as already mentioned, a twofold process. On the one hand it consists in the isolation of constant strains from their neighbors and, inasmuch as the best are chosen, in their improvement. On the other hand it improves races and is the source of those superior fruits which we can only propagate by grafting and other vegetative methods. But this selection, so far as our experience goes, never leads to the origin of new and independent types. In this first section then it will be our object to render the difference between these two types of varia- bility as clear as possible. A correct apprehension of the nature of this difference will make clear the overwhelm- ing importance of Mutability, as opposed to individual variation, in the production of new species. In connec- tion with this critical treatment I have tried, by experi- ments on numerous examples of individual variation, to discover the limits to the amount of alteration that can be attained in this way. And we shall see that these are much narrower than a belief in the theory of Selection, as commonly entertained, would lead us to expect. For the main experiment I have chosen a plant in which I was enabled to follow in detail the phenomenon of Mutation through a number of years. This was 8 Introduction. Oenothera Lamarckiana which as long ago as 1886, formed the starting-point of this work. The second part will show that it has not disappointed me, and will give an account of the whole series of Mutations pro- duced by it. PART I. THE PRINCIPLES OF THE CURRENT THEORY OF SELECTION. I. SELECTION AND MUTATION. In his theory of Selection Darwin combined two principles relating to the origin of species ; and he laid stress sometimes on the one, and sometimes on the other, according to the nature of the available evidence or to the objections of his critics. One was the principle on which the controversy over the origin of species turned in pre-Darwinian days. It was the supposition of a progress by steps in nature, by means of which a new species arose suddenly from a former one. Such a phe- nomenon was called a Mutation. If the new form was distinguished from its parents by a single character the mutation was obviously a relatively simple process. And those who believed in the "sub-species" always regarded the matter in this simple way, even when they questioned the possibility of such mutations on the ground that they never saw them. This was the attitude of the French school in the middle of the XlXth century. They rec- ognized individual variation, and described it time after time; but they saw no connection between it and the origin of species. It always seems an extraordinary thing to me that the occurrence of mutations should have escaped the notice of the workers of that time. For they occur both in the cultivated state where they have been called single variations, and also in nature, where as I hope to show nOFERTT LIBRARY N. C. State College 12 Selection and Mutation. they correspond precisely to the anticipations of the Transmutationists of that time. The weak point of the whole position before Darwin's time lay in the application of the conception of mutation to the Linnean species, for these are not really elementary species but aggregate ones, and the question of their origin is obviously different from that of their con- stituent units. The second principle in Darwin's theory was the idea that individual variation could lead to the origin of new species by continued selection. This idea was at that time absolutely new, and found many adherents amongst whom Wallace, whose views are set forth in his book ''Darwinism," must be considered the chief. Moreover it is Wallace who has insisted that this form of the theory affords the only possible explanation of evolution. He absolutely rejects the theory of the origin of species by mutation. ''Single variations" according to him have no significance for the theory of descent. Experimental researches on individual variability and mutability hardly existed at all at that time. Investi- gfators had to be content with the information of breeders and general biological considerations. But the latter, although they often afford the .strongest argument for the theory of descent, seldom distinguish between the two processes in question. The experience of breeders demands in my opinion the most careful examination before it can be accepted as evidence in a scientific inquiry. Their experiments are neither designed nor carried out with this object in view. A critical revision of the whole range of facts on which the doctrine of selection rests is not only ad- missible, but is urgently called for. Darwin accumu- Selection mid Mutation. 13 lated a vast storehouse of facts and observations; hut our estimation of the importance and significance of the individual facts themselves has undergone a change. Moreover many new observations have been made which place the results obtained by breeders in a new light. Breeders with few exceptions do not work in the service of science ; and most of them take very little inter- est in the purely scientific aspect of their work. They do not make the general plan of their experiments as simple as possible in the hope of finding a rational explanation. On the contrary, as a rule, they prefer complex con- ditions especially where their efforts are directed to the production of new varieties. For the more numerous the factors the greater the expectation of getting something new and good. On the other hand scientific experiments on variability should, where possible, be free from the results of hybridization. But crosses are usually much more important to the breeder than pure races, and only in quite special cases has he the occasion to exclude crossing with the utmost care. Although mutations are often of much greater value to him than individual varia- tions he usuallv treats them both after the same fashion and often does not even distinguish between them. Moreover a systematic record of the culture, of the kind that is absolutely essential to work with a scientific object is not kept by breeders. It would cost far too much time and labor. The only records that are kept by most breeders are those which are necessary for the compilation of their catalogues. And if after a few years a new form proves to be something particularly good, its history is written, as I have been personally in- formed by one of the most distinguished breeders, partly from the information in the older catalogues and partly 14 Selection and Mutation. from memory in such a manner as best suits the pur- poses of advertisement. "It goes without saying," he said, ''that after three or four years one can no longer remember one's single fertilizations and selections." Many other well-known breeders have expressed them- selves to me in similar terms. ^ If we collect all that is known with absolute certainty about the ''How" of the origin of our innumerable garden plants, the result is extraordinarily meagre. Con- cerning the vast majority of them we do not know any more than that they exist ; in the case of others the firm which put them on the market is known, and the year of their introduction ; but the names of their raisers are usually kept secret especially where one is dealing with cases in which the crosses have not been performed pur- posely. And the question as to how the new forms arose, on the answer to which the value of this evidence as bear- ing on the theory of selection depends, can very seldom be answered. Public statements are dictated by exigen- cies of advertisement. It is often only found possible to maintain a well-defined improved race on the market by crediting it with further improvement. All such statements therefore require careful scrutiny before they can be utilized as scientific evidence. I am far from blaming breeders in this matter. It is to friendly intercourse with many of them that I owe in great measure my information on this subject. What I object to is the application by others of the results at- tained by breeders to questions for which they were neither intended nor devised. It was Darwin's insight ^ RiJMKER refers in strong terms to the difficulties which may arise from regarding the grossly exaggerated ilhistrations in seed catalogues as faithful records of the things depicted. See "Der zvirthschaftliche Mehrwerth guter Culiurvarietdten, 1898, p. 2 Selection and Mutation. 15 which enabled him to build his theory of descent on foun- dations supplied by breeders. At the same time he left many points untouched, or at any rate undecided, and for the final settlement of such questions I fear that the statements of breeders will seldom suffice. It is somewhat remarkable that purely scientific in- vestigation has not kept pace with practical experience. This wide field is still open to cultivation, and will, with- out doubt, some day bear a rich harvest. It is my object in this section to test the statements of practical breeders, so far as they admit of such criti- cism. I pay the sincerest tribute to their high practical value especially as in this case science is far behind them. But their application to the theory of descent is another matter. Real service to science can only be rendered by confining oneself to thoroughly authenticated cases. In conclusion : The analogy between the origin of new forms in nature and in a state of cultivation forms one of the chief supports of the theory of descent. But the fact of their origin does not help us to choose between the theory of selection and the theory of mutation ; noth- ing short of a knowledge of the nature and mode of their origin will help us to decide. But on this all-important point the experience of breeders teaches us very little. I shall try to relate what they tell us in the third chap- ter of this Part. 11. MUTABILITY AND INDIVIDUAL VARIA- TION. § I. THE TRANSMUTATION THEORY BEFORE DARWIN. In the introduction to his ''Origin of Species" Dar- win erives a short historical sketch in which he calls Fig. I. Papavcr bracteattim monopefalum. A. The detached Corolla. B. The whole flower.^ attention to the contributions made by his predecessors to the theory of evolution. Lamarck was the first whose * In the cultures of the firm Vilmorin-Andrieux of Paris there are found every year in the plots of Papaver bracteatum single plants whose petals are more or less completely fused. Fig. i is drawn from samples which H. L. de Vilmorin was kind enough to send The Transmutation Theory Before Darwin. 17 views on the origin of species attracted general attention. The chief of those who joined him in championing the common origin of all living form was Geoffroy Saint- HiLAiRE. Their point of view was a purely philosoph- ical one and rested on the principles of natural science current at that time, which sought to account for all natural phenomena without the aid of supernatural causes. Their followers however entered an entirely different field. They abandoned for the time the investigation of the phylogenetic relationship of all living forms and sought to discover the causes of the relationships of smaller groups. They adhered almost always to the Biblical concep- tion of creation, and sought to determine which units were created in the beginning. Some investigators re- garded the genera as creations, others the species of Linnaeus, and a third group the so-called ''subspecies" which would be much better termed elementary species. There can be distinguished among Darwin's prede- cessors and contemporaries four different lines of thought characterized by their different attitudes to Darwin's theory of descent. 1. The philosophical contemplation of nature by La- marck and Geoffroy Saint-Hilaire. 2. The rest of the Transmutationists who regarded the genera as created and the species and subspecies as derived from these. 3. The adherents of the Linnean species, who held that these were created. 4. The so-called school of Jordan who declared that rne. The plant is not on the market. It is not unreasonable to be- lieve that the appearance of the first ancestor of the whole systematic division of the Sympetalae occurred in geological time in the same way as sympetaly has arisen here as a variety. 18 Mutability and Individual Variation. the elementary forms which proved themselves immu- table when cultivated were the real independent creations. Let us first consider the views of the Transmutation- ists. Before Linnaeus the genera were regarded as the systematic units and the species were considered as sub- divisions of them. Many genera have popular names : these groups were known by the country folk, whilst the species were only in much rarer cases distinguished. TouRNEFORT gave the genera known to him their sys- tematic names ; but the species he distinguished only by symbols and not by special names. In his eyes the genera were the essential things, the species merely de- rivatives. The view that genera were created in the beginning and that species had developed from them in the lapse of time by transmutation had many adherents. Among them are to be reckoned Buffon, at least in his earlier works, then Bory de Saint- Vincent, Gmelin, Bur- DACH, PoiRET, Fries and many others.^ This view, at one time found an adherent in Linnaeus.^ He believed in a simultaneous creation of all forms in Paradise ; he suspected however that these forms corresponded to our genera whilst species had arisen from them in part di- rectly and in part by crossings.*^ This is important because it shows that the modern conception of species did not exist before the time of Linnaeus or at any rate that it was not the species which ^ GoDRON, De I'Espccc, pp. 8-10. ^ "Genus omne est nafurale, in primordio tale creatum." Syst. Nat. Veg. 14. Philos. Bot. No. 159, p. 104. C. LiNNE, Oratio de Telluris hahitabilis incremento. Upsala, 1743; Leyden, 1744. — Idem Amocnitatcs academicac. 1794. T. I., p. 71 {de Peloriis). The Transiimtation Theory Before Darwin. 19 were regarded as the real units of the system. This is also obvious from the meaning of the expression noinen specificiim which was in use at that time. Tournefort and his contemporaries wrote after the generic name a short diagnosis each time, in order to distinguish the single species from one another. So long as only a few species were known in each genus, one character sufficed. But as the number of species increased more characters became necessary, until finally many species could only be denoted by a description which occupied several lines. A circumlocution of this kind we now call a diagnosis; then it was called a nomen specificimi and had to be written out every time one wanted to refer to a particular species. LiNNAEU's substituted his binary nomenclature for these cumbersome noniina specifica^ and, in order to give his species the necessary importance he raised them to the rank of the units of the system. He advanced the proposition Species tot numeramns, qiiod diversae formae in principio sunt creatae^ and thus laid the foun- dation of the conception of species that is recognized to-day. And just as it had been supposed up to that time that species arose from genera by natural means, so, according to Linnaeus, smaller types had arisen from the species.^ But in order to insure as far as pos- sible the supernatural dignity of his species Linnaeus forbade his students to study the smaller types : Varie- tates lez'issinias non curat botanicus, ran the command."* LiNNAEUs's species were aggregate species and not ^ Philosophia Botanica, No. 257, p. 207. ' Ibid., No. 157, p. 103. ^ "Varietates sunt plantae eiusdem specie!, mutatae a caiissa qiia- CLinque occasionali," Ibid., No. 306, p. 243 ; No. 158, p. 104. * Ibid., No. 310. 20 MiLtahility and Individual Variation. true units. It seems that Linnaeus himself was fully aware of the fact, but it is certain that it was gradually lost sight of by his followers. In relatively few cases did he himself distinguish varieties within his species, and it is well known that when he did they were often raised to the rank of species by those who followed him. Well-known examples are afforded by Primida veris L. with the three varieties vid- garis (acaulis) (Fig. 2), cla- tior and officinalis ^ which are now universally re- garded as species solely on the authority of Jacquin without any further justi- fication. In like manner Lychnis dioica L. split up into L. diiirna and L. vcs- pcrtina, Platanthcra bifolia L. into P. bifolia and P. chlorantha and so forth. Numerous examples of a similar kind will occur to the reader. Conversely also Linnean species have been degraded to the rank of varieties : for example the Index Kewensis which recognizes the Primula species of Jac- ^ Primula acaulis is distinguished from the two other subspecies by the fact that its flowers arise singly by their stalks from the axils of the leaves and are not united to form an umbel. This species occurs in certain localities in the Netherlands, in the wild state, and from time to time bears umbels of which one is drawn in Fig. 2. Such cases are regarded as atavistic, as reversionary to some common ancestor of those Primulas which still possess umbels. But this atav- ism is not considered by the best systematists as sufficient ground for re-constituting P. acaulis as a variety and the main species P. veris as a species, in systematic works. From the point of view of the estima- tion of the systematic value of Atavism in general this case evidently is of much importance. Fig. 2. An umbel of Primula acaulis. The Transmutation Theory Before Darwin. 21 QUiN regards Datura Tatula L. as a variety of D. Stra- monium L.^ Linnaeus^ species, therefore, embraced his varieties and these varietates minores, which he would not allow his pupils to investigate. But it was not proved that all these smaller types had arisen from the species ; it merely followed from his definition of species. And so long as the Linnean species of the systematists provided them with sufficient work there was no reason for them to doubt his words or disregard his precept. But as the study and description of the "species" particularly of the European Flora gradually approached completion the attention of naturalists inevitably turned in the direction of the hitherto neglected Varietates minores. It soon became evident that these were much more numerous than Linnaeus ever supposed ; moreover, that they were distinguished by just as numerous and just as definite characters as Linnean species. Their discov- erers demanded for them the ''rank" of Linnean species and elevated them to it. Some authors went so far as to assert that by such discriminations they had created new species. The best known example is afforded by Draba verna which has been studied carefully by Jordan and by many other independent investigators after him. Among the latter I would mention De Bary whose results, which are in full agreement with Jordan's were published after his death, by F. Rosen in the Botanische Zeitung for 1S89. The European Flora includes about 200 (elemen- tary) "species" of Draba which together constitute the old species Verna and, so far, have remained constant ^ It is most remarkable that in the Index Kewensis which was published at Darwin's expense after his death no distinction is drawn between varieties and synonyms. 22 Mutability and Individual Variation. and distinct under cultivation. The extent of these dif- ferences is sufficiently indicated by a series of the most important forms, in Fig. 3. A heated controversy in which Jordan and Godron played the most prominent parts has raged over the ques- Fig. 3. Subspecies of Draba verna. i. D. violacea ; 2., 3. and 4. D. scabra; 5. D. subnitens ; 6. D. majuscula ; 7. D. obconica ; 8. D. glaucina \ 9. D. clongata ; 10. D. graminea. (After F. Rosen, Bot. Zeitung, 1889. Plate VIII.) tion as to whether these smaller perfectly circumscribed types should be called species or not. Jordan and the advocates of the smaller species based their views on the results of cultures ; and in this way they have en- The Transmutation Theory Before Darwin. 23 riched science with a collection of experimental facts of the greatest importance. A considerable portion of the fourth chapter of this work will be devoted to a critical consideration of these facts. Fig. 4. Subspecies of Viola tricolor. j.V.agrestis', 2.V. segetalis, general habit similar to that of I', agrestis ; 3. V. gracilescens; 4. V. pallescens ; 5. V. nemausensis. (After A. Jordan, Observ. s. plusicurs plantes rares ou critiques, II, 1846. Plates i and 2.) The description of any form as a species, or rather the supposed proof that any form zvas a species, carried 24 Mutability and Individual Variation. with it the assumption that the form under consideration had been created as such. The reasonableness of this position was recognized by both parties but especially by GoDRON and Jordan. But at the present day when the common origin of all species is hardly ever called in question it is very difficult to judge this controversy fairly. The origin of a new form from another was termed at that time a mutation.^ Godron and Jordan asserted that every one of the forms constituting their species was immutable. Jordan moreover tried to prove the truth of this statement by breeding experiments. He recog- nized individual variation and observed and recorded it accurately.- He was also acquainted with local races^ whose differences disappear whenever they are cultivated for a few years next one another in the same soil ; he knew moreover that as a rule it took only a few genera- tions to effect this change. Further, he was familiar with the results of accidental crosses by insects or wind, and mentioned the genera {Cirsiiim etc.) in which this was most apt to occur. But individual variability and mutability were abso- lutely different things in his eyes ; he frequently observed the first; but never saw an example^ of the latter. That was why he held species to be immutable.^"* ^ Already Lamarck used the terms "races muiahlcs on variables" ; see GiARD, Discours d'ouverturc de J. B. Lamarck, 1907, page iii (Note of 1908). ^A. Jordan, De I'origine dcs arbrcs fniiticrs, 1853, p. 9. ^ Loc. cit., p. 10 * Besides the common Viola tricolor, V. arvensis (Murray) is very familiar; it is reckoned by many authors as belonging to the same species. Compare for example Koch, Synopsis Florae ger- manicac et hclveticae. V. arvensis itself consists of a series of con- stant forms of which our Fig. 4 shows some of the more important. ^Jordan, De VOrigine des arbres fruifiers, 1853. Tn this work and in the other essays Jordan always uses the words "niufation" and "inimutabilite" where he is dealing with the supposed change of nftFERU LIBRART fl. C. State College The Transimitation Theory Before Darwin. 25 GoDRON also distinguishes quite clearly between spe- cific characters, and trivial fortuitous and purely ''indi- vidual" deviations which soon disappear when the condi- tions wdiich called them forth cease. The latter are united together by a series of transitions; the former are not.-' When Darwin's work on the origin of species ap- peared,^ the controversy over the ideas of species and mutability raged most fiercely in France. But it only turned on the question whether the larger or the smaller species were separately created, or whether they had both arisen from an original type. This original type, how- ever, was never thought of as being larger than a genus. '^ The transformation or transmutation theory of those days was therefore an entirely different thing from the modern theory of descent. Nevertheless Darw^in him- self says in 1858 at the suggestion of Lyell and Hooker he resolved to write a book on the ^'Transmutation" of species, a book which was published in the following year under the title of the Origin of Species.^ It is curious that the terms Mutation, Mutability, Im- mutability and so forth should have been so completely driven out of use by the theory of Selection. Darwin directed his whole energy wnth full knowledge against the dogma of the immutability of species. His ''Origin of Species" begins with the statement that until recently the great majority of investigators had believed "that species were immutable productions."'^ "I had become, in the year 1837 or 1838, convinced that species were mutable productions,"^ says he in his Autobiography; one species into another. See pp. 7, g, ii, 13, 34, etc. Also Godron, De I Espcce, e. g., II, p. 422. ^Godron, De l' Espcce, I, p. 175. ^ Nov. 24, 1859. ^ See Wallace, Darzvinism, pp. 3-6. * Life and Letters, T. p. 85. ^Origin of Species, 6th ed., 1898. Historical Sketch, p. xiii. ^ Life and Letters, I, p. 93. 26 Mutability and hidividual Variation, and in the passage cited in the Origin, he discusses the question whether in Paleontology the immutability of species was or was not assumed by the most prominent workers.-^ The prevailing opinion was that individual variability and mutability were two distinct phenomena. Variabil- ity was well known both in cultivated and in wild species, but most thoroughly in wild species which had been kept in cultivation through a number of years. It was found however to be limited, to depend upon the influence of the environment and to be useful as a means of adaptation. Mutability was not encountered in practical experience. No cases of a species arising from another had occurred in scientific cultures, nor were there any sufficiently authenti- cated instances of the origin of new forms in the nursery or the farm in spite of a thorough and critical scrutiny.^ The adherents of the Transmutation theory explained the systematic relationship of the single forms (species, varieties, and so forth) within the genera by the theory that they had a common origin. The opponents of this theory, in so far as they were upholders of the Linnean conception of species, held exactly the same views, except that they regarded the species as created and not the genera. Foremost amongst them was Godron. who con- sidered the races and varieties and even the species of Jordan as having arisen from the Linnean species by natural means, and made a very extensive collection of facts and observations to prove this view. The third school was sharply opposed to these two groups, the Transmutationists and the upholders of the ^ Origin, p. xviii. ^Jordan, De VOriginc des arhres fruiticrs, 1853, and Godron, De I'cspcce et des races. The Transmutation Theory Before Darwin. 27 Linnean conception of species. It relied exclusively on the Biblical story of creation and on experiment. Every form which proved itself to be immutable by experiment was, according to their theory, an independently created form. The experiment consisted in cultivating the par- ticular form in a garden for a few generations. They disapproved of the systematic grouping together of such pure forms into larger "species" on the ground that it was artificial and arbitrary. They recognized genera and the larger groups as necessary, but regarded them as manifestly artificial divisions. According as one belonged to the one or the other of these parties one was more or less prepared for Bar- wind's new teaching. The thin ranks of the Transmuta- tionists and the huge Linnean army admitted a priori the origin of races, varieties and Jordan^s species from other forms, and this in spite of the complete absence of ex- perimental proof. It was against these that Darwin turned his energies to show, what was indeed the chief object of his argument, that the supposition of a com- mon origin for genera and families was as much justified as the view held at that time, that the forms gathered to- gether in one species were descended from a common ancestor. The adherents of Jordan^s school who regarded the elementary species as created, were least prepared for Darwin^s teaching. There were however very few of them, and their system, by being so rich in species (Draba verna alone falls into more than 200), stood very much in the way of a wide acceptance of their views. At any rate they were not, or at most only very slightly, con- vinced by Darwin ; the bulk of them maintained their original position. The only one of them that I would 28 Mutability and Individual Variation. mention here is Michaele Gandoger whose Flora Eu- ropae is the most comprehensive work along this line of research. The controversy before the time of Darwin had therefore led to two essentially distinct results. These were : 1. The experimental proof of the existence of nu- merous, constant and mutually independent types within the limits of the Linnean species. 2. The general conviction that these constant types had arisen naturally from larger groups or species by mutation. ^ § 2. DARWIN'S SELECTION THEORY. The theory of Descent aims at the scientific explana- tion of systematic relationship. It is Darwin's immortal service to have obtained general recognition for this generalization. By doing this he revolutionized the whole of biological, systematic, embryological and pale- ontological science, tapping inexhaustible sources for new investigation and discovering everywhere mines where new facts wxre to be had for the picking up. The several propositions and hypotheses which Dar- win employed as supports for this theory should be re- garded now only as such, since their interest is mainly historical. They have served their purpose and are thereby fully justified. Whether they contain in part what is unproven or what is incorrect matters not. But they contain, over and above that, a large mass of im- portant facts which can be made use of to build further * The terms immutability and so forth have not entirely dropped out of use. E. g., B. J. Costantin, Accomodation dcs plantes aux cliniats froid et chaud, Bull. Scientif., public par Alfred Giard, T. 31, p. 490, 1897, and Bateson, Materials for the Study of Variation, 1894, P- 2. Darzmn's Selection Theory. 29 on the foundations laid by Darwin. This is especially true of the theory of selection, which now has served its time as an argument for the theory of Descent; happily this theory no longer stands in need of such support. We are now concerned to bring the origin of species into the field of experimental investigation. The position of the theory of Descent as a comparative science is com- pletely assured by the results Darwin obtained ; but as an experimental science it has made feeble progress.^ The cause of this lies in my opinion not so much in the difficulties of the investigation as in the lack of definiteness of this part of the theory. In the systematic sphere the discoveries could have been predicted ; this was far from being the case on the physiological side. Darwin was never quite clear about the physiolog- ical part of the theory of Selection. It seems to me that he always inclined first in one direction and then in another, never fully deciding between the two views. In his earlier works especially, he treated spontaneous variations (single variations) as the material afforded for natural selection whilst in his later works, in con- sequence of the objections of his critics he gave greater prominence to the part played by individual variation in the production of new species. But he never sharply dis- criminated between these two processes. Moreover, such a discrimination was not in the interests of his main ob- ject. It would have led him to many difficult points whose solution was not necessary to the theory of descent, and would have diverted attention too much from the main point at issue. As we have seen in the foregoing section, the * See also Bateson^ Materials for the Study of Variation, pp. 7 and II. 30 Mutability and Indiz'idiial Variation. proposition that races, varieties and subspecies of wild as well as of cultivated plants have arisen by certain modifications from "species" received general recogni- tion. Darwin had collected all the facts available for a history of these changes in the case of cultivated plants.^ They provide us with a history of garden plants and often also give the source and the date of introduction of varieties, but they do not tell us whence they came or how they arose.- ^'Varieties arc incipient species" and ''species have descended, like varieties, from other species" ; these are the two famous propositions which Darwin is contin- ually asserting and with whose proof he is chiefly con- cerned.^ In other words: the origin of varieties from species is granted ; why not species from species ? In order to prove this it is obviously not necessary to know the exact way in which varieties themselves originate. It is sufficient tliat the relation between species and genera is the same as that between varieties and species. Darwin asserts again and again that it must not be forgotten that under the term of variations mere indi- vidual differences are included.^ His variabilitv is there- fore always to be understood in a double sense. It con- sists on the one hand of individual difi^erences and on the other of single variations.'^ The former belong to those phenomena which we now term individual varia- ^ Of later works compare especiallj' Alph. de Candolle^ Suv Vorigine dcs plantes cultivccs. ^ See also Bateson, Materials, p. 17. ^ Origin of Species, 6th ed., pp. 2, 4, 86, etc. * Origin, ibid., pp. 64, 80, etc. ^ Life and Letters, III, p. 108. As examples of single variations are considered such cases as the color of the flowers of Datura Tatula (a blue form belonging to the white-flowered D. Strainoniiim) and the absence of spikes on the fruits of Datura mcrmis. See Fig. 5. Darwin's Selection Theory. 31 tion, and conform to Ouetelet's law. The latter are sporadic, spontaneous changes corresponding to our Mu- tations (Fig. 5). Darwin almost always speaks of these two types in his discussion on Selection but never separates them, and is always in doubt as to their relative importance in the origin of species. Fig. 5. I. Datura Tatula, with blue corolla and fo- liage tinged with red. 2. Fruit of D. Stramo- nium with thorns, unripe. 3. Fruit of D. (Straiiw- niuui) inennisj without thorns, ripe, dry and open. This being the case, it seems to me that it is almost unfair in a criticism of Darwin^s views to regard these two types as distinct. If I do so, I do it with the express object of showing that although Darwin was acquainted with the two phenomena he was not prepared to separate them completely on the basis of their significance for 32 Mutability and Individual Variation. his theory. Here, as everywhere, Darwin advanced with the utmost caution. Our problem then is this : In the formation of new species, does natural selection choose the extreme vari- ants of the ordinary individual variation, or does it choose occasional Mutations. In any large community there is always an abundant supply of extreme variants. More- over the struggle for existence does not preserve the single absolutely perfect ones only, but groups of the best, since it simply eliminates the least perfectly adapted. There is, so to speak, always plenty of material for se- lection in every species, and in every character. But in- dividual variability is, as far as our experience goes, by no means unlimited ; its limits are not indeed precise, but thev fall well within the rano-e of Ouetelet's Law. Single variations are chance phenomena into whose essential nature Ave have as yet no insight. We know that they occur and that they occur seldom ; but not too seldom. As to how they come about scarcely anything is known, but it is generally assumed that they appear suddenly,^ and they are consequently termed sports. They suddenly change a species into a new form ; or, from a variety, they make a new one absolutely different. Fre- quently they concern only a single character and then usually consist in the loss or latency of a character al- ready present, e. g., white flowers, absence of thorns {Datura incrmis, Fig. 5), hairs, runners (e.g., Fragaria alpina, Figs. 6 and 7, pp. 33 and 34),^ seeds, branching, ^ By far the majority of observations that have been adduced as instances come under the heading of hybridization. ^The Gaillon strawberries (Fig. 7) which are distinguished from the ordinary monthly strawberries (Fragaria alpina, Fig. o; solely by the absence of suckers and the correspondingly greater branching of the rosettes are often cultivated for the very reason of Darwin's Selection Theory. Z2> etc. ; these cases are instances of retrogressive mutability and have no signification for the elucidation of the main lines of descent. Apart from this quite definite group of modifications hy loss, single variations seem to be presented by all char- acters, to proceed in every direction and to be apparently without limit. To sum up, individual differences are al- ways present, occur in every direction and in every char- acter, but are limited and conform to definite laws. Single variations, on the other hand, are sporadic phenomena, appearing only from time to time, and suddenly changing Fig. 6. Fragaria alpina, IMonthly Strawberry. (Fraisier des quatre saisons.) the forms of life. They cannot be induced at will, but must be waited for.^ We have thus to decide between : 1. a selection of extreme variants. 2. a selection of mutants. The question for Darwin was. which of these two has played the greater part in the origin of species ?- this deficienc3^ See Vilmorin-Andrieux, Les plantes potageres, 1883, pp. 221-222. ^ Origin, loc. cit., p. 62. ^ The selection of extreme variants in nature forms the so-called local races and plays an important part not only in acclimatization but especially in many cases of adaptation to new environmental con- ditions. See III, § 4. 34 Mutability and Individual Variation. The breeder employs both, according as opportunity offers. Darwin asserts over and over again that their method consists in the accumulation of successive slight variations.^ But as to whether these small changes are variations or mutations he gives no decision. Natural selection, he says, like artificial, chooses these ''slight variations,''^ but to which category these severally be- long is left uncertain. Moreover it was Darwin's belief that Natural Selection was not the sole factor, for at the Fig. 7. Fragaria alpina, Monthly Strawberry with- out runners. (Fraisier des quatre saisons sans coulants, Fraisier de Gaillon). conclusion of the introduction to his Origin he says, '7 am convinced that natural selection has been the most important, but not the exclusive means of modification.'"^ In almost all works on Darwin's theory we find the .story of how he arrived at his theory of selection by reading Malthus's Essay on Population."* Already well ^ Origin, loc. cit., pp. 3, 63, 64, etc. ' Ibid. ^ See also Origin, p. 72. * See Life and Letters, I, pp. 83, 84. Darwin's Selection Theory. 35 acquainted with the struggle for existence and the un- ceasing destruction of countless individuals, he found in that book the long sought solution. He came to the con- clusion that selection played the same part among animals and plants as it does amongst mankind, and that in this manner species may have arisen. This conclusion, how- ever, is simply the idea of a genius and does not directly follow from Malthus's work. It has become one of the main supports of the doctrine of descent. But it was to the genius of the great thinker, not to the sound- ness of the raw material that the magnificence of the result was due. ' In the light of what we know now^ this story of the origin of the theory of selection often stands openly contradicted by Darwin^s own view. Natural selection, says he, works on ''chance variations''^ ''Unless such occur natural selection can do nothing.''^ From such utterances it is clear that Darwin attributed a very great and often preponderating, perhaps even an exclusive, significance to "single variations/' For individual varia- bility always provides natural selection with the required material in the form, sometimes of greater and some- times of less deviations from the type; it is, moreover, exhibited everywhere and in all directions. This fact was known quite well at that time, and Darwin himself was quite clear about it. But the laws formulated later bv OuETELET wcrc not known ; and the p'eneral insidit into the matter was much less deep than it is at present ; no one however questioned the universal occurrence of * It was in 1838 that Darwin read Malthus's book, and Quete- let's Anthropomctvie first appeared in 1870. ^ Life and Letters, II, p. 87, etc. ^ Origin, p. 64, etc. 36 Mutability and Individual Variation. variability. The chance variations were not therefore the extreme variants of the ordinary variabiHty; they were sporadic occurrences. Natural selection is on the lookout for these, says Darwin, and seizes on them ^'whenever and wherever opportunity offers.'''^ Darwin regarded these occasional deviations, these mutations, as appearing from time to time and in a gen- eral way conforming to definite laws as yet imperfectly understood. According to these laws it could not hap- pen that any considerable length of time should pass by without the appearance of at least a few considerable variations of this kind. To such variations would be due the progress which the majority of living forms ex- hibit in the course of the centuries. The longer the time the better is the prospect of the appearance of favor- able variations, 2 especially if these should only appear very seldom.*^ They provide us with ''intermittent re- sults."4 Moreover Darwin went so far as to believe in a certain periodicity. "Nascent species are more plastic," that is to say produce more sports and have therefore a better chance of splitting up into new species. Darwin cites Naudin and Herbert as the authors of this view, which they had derived from their comparative studies of the forms occurring within certain groups of plants.^ Schaafhausen^ mentions the unequal rate of the prog- ress in different branches of the genealogical tree, in some of them the changes taking place very quickly whilst in others absolute stagnation seemed to be the rule during long geological epochs. To produce a genuine new spe- cies, a variety must from time to time, perhaps at long * Loc. a7., pp. 65, 66. '^ /&f J., pp. 82, 86. ^ /^/cf., pp. 85, 92 * Ibid., p. 85. ^ Ibid., Hist. Sketch, p. xix. ^ Ibid., p. xx. Darzvms Selection Theory. 37 intervals, give off variations in the same direction. In this way it progresses *'step by step/'^ Let us look for a moment at Darwin's views on the influence of external conditions. On this matter again we find that his opinion is by no means fixed. Sometimes he would appear to think that it has played very little part in the origin of species, at other times he ascribes great significance to it. And inasmuch as he was quite familiar with the relation of individual variation to the environment, it follows that he was chiefly concerned here with single variations. In a letter to Hooker. 1856, he says, ''My conclusion is, that external conditions do extremely little, except in causing mere variability." "How much they do is the point of all others on which I feel myself very weak."- We are all familiar in the pages of Darwin's books with the important role ascribed to changed conditions of life. Especially in the case of the transport of a plant from one climate to another and the effects of the first years of cultivation on a wild species.^ Species therefore with a wider geographical distribution are more likely to produce new forms. In later years Darwin has again changed his views on this point ; after reading Hoffmann's famous re- searches he said : No doubt I originally attributed too little weight to the direct action of conditions. Perhaps hundreds of generations of exposure are necessary. It is a most perplexing subject. (1881.)^ The strongest influence on Darwin in his relation to this question was that produced by a criticism which was published in 1869 by Fleeming Jen kin in the ^ Ibid., p. 66. ^ Life and Letters, II, p. 87. ^ Origin, p. 64, etc. * Life and Letters, III, p. 345. 38 Mutability and Individual Variation. North British Review.^ This writer tried to prove, by calculations, that the likelihood of single variations main- taining themselves in the struggle for existence or of ulti- mately being victorious in it was very faint. Darwin allowed himself to be convinced by this and says forth- with: / ahvays thought individual differences more im- portant, but I was blind, and thought that single varia- tions might be preserved much oftener than I now see is possible. As the result of this criticism he made many alterations in the subsequent editions of the Origin. Finally I shall refer to the conclusion which Darwin derived from his theory of Pangenesis in its relation to these two forms of variability.^ There are two abso- lutely different groups of causes. First, the relative number of the units, their activity, their inactivity, their relative positions and the calling to life of those long in- active. Such changes occur without the units themselves being modified by them. Such changes zvill amply ac- count for much fluctuating variability, that is for that kind of variability which we now call individual, gradual or fluctuating variability. The second group of causes includes the direct effect of altered conditions on the organization of the indi- vidual, in which case Darwin supposes the units them- selves to be altered. If the new units have then suffi- ciently multiplied, to be a match for the units already ex- isting they will lead to the elaboration of new structures. These quotations convince me that Darwin believed the main branches of his genealogical tree to have arisen bv a modification of his oremniules and that he res^arded ^ Origin, p. 71. Life and Letters, III, p. 108. Animals and Plants under Domestication. 2d ed., 1875, II, P- 390. Wallace s Selection Theory. 39 fluctuating variability as a phenomenon of an entirely different kind.^ To sum up, we see that Darwin always distinguished between individual differences and single variations and that he ascribed to the latter at least a very considerable role in the origin of species. It was only by the pressure of criticism that he finally gave up this view and gave the place of honor to the ever present individual varia- tions. § 3- WALLACE'S SELECTION THEORY. In his book on ''Darwinism" Alfred Russel Wal- lace has collected in an excellent and convincing manner a valuable mass of evidence for the theory of descent.^ Few authors except Darwin have taken such a prom- inent part in fighting for this theory, as he. His book ''Darwinism" consists essentially of two parts. In the first sections Wallace deals with variability and selec- tion, in the second he describes the wonderful adapta- tions of animals and plants to their environment and seeks to explain them on the basis of Darwin's theory by bringing out as forcibly as possible the agreement between the demands of the theory and the facts them- selves. This latter half is undoubtedly the most interest- ing of the whole work. But I shall only discuss his theory of selection in this book. Wallace's selection theory differs from that of Darwin in one essential point. Wallace regards the ever present individual variation as the material from which natural selection forms new species. It is his main ^ See also my Intraccllularc Pangenesis, pp. 73-74, 210, etc. ^ A. R. Wallace, Darzvinism, an Exposition of the Theory of Natural Selection zvith Some of its Applications. London, 1889, 2d. ed. 40 Mutability and Individual Variation. object to show that animals and plants do perpetually vary in the manner and to the amount requisite.^ Single variations he regards as absokitely without significance; they have played no part (he says), or at most hardly any, in the origin of species.^ Our author holds himself to be at one with Darwin in essentials and only to have rendered his selection theory sharper and more precise. The hosts of doubts which, as we saw in the preceding section, were always so carefully brought forward and discussed by Darwin, disappear. The theory has become a compact, clear and surprisingly simple one. Wallace takes just as careful account of the systematic and biological facts as Darwin did in his cautious way, but Wallace^s theory is much more convenient and attractive than Darwin's. This very clearness in the mode of presentation makes it easy for the critic to discover the weak spot. In fact the author himself almost lays his finger on it. At the end of the first section he gives a summary of his collec- tion of facts and the method of his proof ; and one has only to follow carefully to discover the weak point in his argument.'^ It will be useful to recapitulate as briefly as may be this argument. Wallace's theory of natural selection rests on two series of facts. The first is the rapid multiplication and the resulting premature death of innumerable individuals. The second is variability and the survival of the fittest. Against this part of his argument I have no objection to raise. He then goes on to consider another im.portant ^Darwinism, 2d. ed., p. 13. '"My whole work tends forcibly to illustrate the overwhelming importance of natural selection." Wallace, loc. cit., pp. vii-viii. ^Darwinism, pp. 12, 13. Wallace's Selection Theory. 41 point. This point concerns the principle of the inherit- ance of variations and the artificial improvement of races by selection. In many cases cultivated forms have become so different from their wild ancestors by this means, that they can scarcely be recognized as their descendants. But the word races has evidently a double signification. It means not only the races improved by selection, but also the constant subspecies of unknown origin which already exist. ^ Without doubt many culti- vated forms diverge to a certain extent from the species to which they are considered to belong by systematists. But these forms are subspecies and their common origin from a single species is just as good a hypothesis as that of the common origin of the species of a genus. Culti- vated subspecies are in well-known cases older than cul- tivation itself; as Wallace himself for example shows in the case of the races of the dog.^ How they have arisen we do not know, not even in the case of those that have probably arisen in a state of domestication. On this slender foundation Wallace now proceeds to build further, and says, p. 12: ''It is therefore proved that if any particular kind of variation is preserved and bred from, the variation itself goes on increasing in amount to an enormous extent; and the hearing of this on the question of the origin of species is most im- portant/' But this thesis is by no means proved ; on the con- trary its truth is only assumed for the sake of the argu- ment both by Darwin and Wallace, and by the mass of their followers. Wallace evades this point in his book; he neither ^ As for instance the races of mankind. ^ Loc. cit., p. 88. 42 Mutability and Individual Variation. subjects it to a stringent criticism nor does he devote a separate section to it. Furthermore in the treatment of single instances this thesis is taken for granted with- out further proof. One sees this most clearly in the discussion of the apple :^ It is known, he says, that all our kinds of apples spring from the wild Pyrus Mains and that from this over a thousand different forms have been developed. This gives one the impression that cultiva- tion produced these numerous forms. But as a matter of fact the apple in the wild condition is a polymorphous species rich in subspecies and the w^ell differentiated types which are now cultivated already exist among the wild forms. The transformation of the wild crab apples into juicy and finely flavored fruits is all that has been brought about by cultivation. It is an absolutely unproved assumption that individ- ual variation extends its range by selection and increases "to an enormous extent^ This is the weak point in Wallace's selection theory. I admit that with this assumption it would be very easy and simple to account for the phenomena of adap- tation, and that this forms a very strong argument for it. And if it were only a matter of this explanation little purpose would be served by raising objections to it. But it is, as a matter of fact, fallacious. Selection certainly leads to enormous practical results, but that is a very different thing from enormous biological changes. The fact that a man can increase the yield per acre by one-half, has no significance from the point of view of the origin of species. In the third chapter I shall seek to prove this by the help of facts. it is not necessary to follow Wallace's argument * Loc. cit., p. 87. The Various Forms of Variability. 43 further. If his assumption is once granted everything else follows. On page 13 he again sums up his position. He is concerned there to show that variations of every kind can be increased and accumulated by selection not only in the cultivated but in the wild condition. I fully admit that Wallace has effected this proof in a masterly and convincing manner. But we also require proof that this increase and accumulation takes place ''to the amount requisite'' for the origin of new species and subspecies; and this proof Wallace neither brings forward nor seeks. Instead of it, his book is full of instances of the compound nature of cultivated and of wild species and of their so-called elementary or subspecies ;^ but how these have arisen we are not told. He has equally little success in proving that races which have arisen by selec- tion remain constant without further selection. Finally, we see that Wallace in his selection theory starts from individual or ordinary variability and allows no share in the process to single variations. He shows that the hypothesis thus simplified effectively coordinates systematic and biological facts, but he fails in proving that as a matter of fact specific characters can really arise by the selection of individual differences. § 4. THE VARIOUS FORMS OF VARIABILITY. Nothing is more variable than the meaning of the word variability. Many authors use this word in so comprehensive a sense that one cannot understand what they mean. (Fig. 8.) It is therefore important to distinguish as clearly as possible between the various phenomena included under See for example pp. 77-78, 85-86, etc. 44 Mutability and Individual Variation. this term. For they stand in absokitely different relation to our thesis. The following groups of phenomena usually fall within the meaning of the term variability: Fig. 8. Hedera Helix var, arhorea} 1. Systematic polymorphism and its supposed causes. 2. Polymorphism caused by crossing. ^ The best known example is afforded by Hedera Helix arhorca whicli is offered by many nurseiymen as var. arborca. It is not a variety, but consists simply of the erect flowering shoots cut off the ordinary ivy, stuck in the ground, and grown as trees. In April i88S I made some such cuttings, and have cultivated the best one till the present time. It forms a richly branched bush over a meter higli (Fig. 8). As is shown in the figure at a, b, c, creeping branches arise from time to time. In 1893 I sowed the berries of an older plant of this kind, in this case an ivy bush of about two meters, and obtained over a thousand seedlings. These still grow in our garden and have made, up till now, exclusively creeping stems and branches. The Arborea-iorm is evidently not inherited. Similar phenomena occur in many other genera, for example in the creeping species of figs in South Europe ; but they have not been sufficiently investigated. The Various Forms of Variability. 45 3. The differences, in individuals and organs, which follow Ouetelet's law. 4. The so-called spontaneous variations. The special problem which the mutation theory seeks to explain is the manifold diversity of specific forms; spontaneous variations are the facts on which this ex- planation is based. The truth of this explanation will then be tested by its application to hybrids; and, if possible, proved. Individual variability however will be shown to be of only secondary importance. It will be convenient to deal with these groups one by one. 1. Systematic Polyinorphism and its Supposed Causes. Linnean species are aggregate species. They include sometimes a small but often a large series of forms which are as sharply and completely distinguished from one another as are the best species. These lower-rank forms are usually called varieties or subspecies; varieties, if they are characterized by a single striking character, but subspecies if they are distinguishable by the sum of their characters, by their so-called habit. But on this point there is a great diversity of opinion. Some authors re- gard all these special forms as elementary species and consequently give them double names, thereby breaking up the Linnean species. It is well known that in this way Draha verna^ and Viola tricolor- and many other old species have been broken up (in the case of Draha verna into 200) smaller groups of perfectly distinct and usually local elementary species. By experiment and culture these forms prove constant, they do not change into one another, nor do they reproduce the typical or general form of the species. The majority of varieties * See Fig. 3 on page 22. ' See Fig. 4 on page 2;^. 46 Mutability and Individual Variation. are just as constant as these. Whether we give them binary and ternary names is not of much consequence. It has always been assumed both before and after Dar- win's time, that they have a common origin, biit in remarkably few cases is there historical evidence that this is so. When and how Datura Stramonium inermis, Rohinia Pseud-Acacia inermis, Lychnis diurna glaber- rima, and the wdiole series of glabrous thornless, white- flowered, laciniate forms, and so forth, have arisen we do not know. They exist and claim recognition equally with the best species. There are a few exceptions, for example Chclidonium laciniatwn Mill. (Fig. Z7 in V, § 25), Fragaria alpina Gaillon (Fig. 7, p. 30), etc., whose source is known. In practical horticulture matters are just as bad. End- less varieties are known but only in rare cases is there any historical information as to their origin.^ This section of the subject of variability therefore is a purely comparative one, its laws are morphological, and only rarely does it lend itself to historical or experi- mental study. 2. Polymorphism induced by hybridization is due to new combinations of the heritable characters of the forms crossed. Two groups of phenomena must be distinguished here : scientific experiment and horticultural and agricul- tural crosses. The scientific investigator chooses, if he can, the least 'Variable" species whilst the gardener prefers to cross types of which at least one is very 'Variable." For this variability can be inherited by the hybrid and increases the likelihood of getting new forms ; and this is of coarse * (Note of 1908.) A most interesting and complete list of these instances has since been given by Kokschinsky. See Flora, 1901, Bd. 59, pp. 240-363. The Various Forms of Variability 47 what is wanted. New elementary characters arise in hybridization experiments solely through this kind of variability, and not as the result of the crossing it- self; as for example Al- fred Bleu, the distinguished raiser of Caladiums, has as- sured me to be the case with his cultures. 3. Variability in the re- stricted sense or individual variability, is the name given to those dissimilarities of in- dividuals and organs, which can be described in terms of Ouetelet's laws.^ These laws, with which Darwin was not familiar, and which were only imper- fectly dealt with by Wal- lace^ have since that time been the subject of close investigation ; with the result that it has become increasingly evident that these varia- ^ See Figs. 9-13; also Fig. 22 (curve of 40.000 beets) in chapter 3, § II, where also the theoretical curve is shown. ^ Case containing- beans to demonstrate their variability in length. The glass case is divided by strips of glass into nine equal partitions. About 450 beans ( redspotted seeds of Phascolus vulgaris) were picked from a bought sample and the individuals measured. Their length varied between 8-16 millimeters, and in the following pro- portions : Partitions . . . . i 2 3 4 5 6 7 S 9 mm. 8 9 10 II 12 13 14 15 16 number i 2 23 108 167 106 2>}> 7 . ^ The beans were then placed in the subdivisions of the jar, in such a way that each compartment only contained beans of the same length (measured in whole millimeters) and in the order shown above. Without further treatment the beans show a grouping ac- Fig. 9. Glass Jar with Beans.^ 48 Mutability and Individual Variation. tions are of an entirely different nature from the rest of the phenomena included under the name of variability. They have this in common that they are always present and can be observed every year and in every group of 2 23 /ns 167 ;o6 33 ' Fig. 10. Curve of Beans/ individuals provided it is not too small. They are always grouped round a mean, and the numbers of the deviations cording to Quetelet's law. For a more exact demonstration a cor- rection would be necessary, since obviously the larger beans fill up their compartment more than a similar number of small ones. ^ Curve of the red-spotted beans. The curve is plotted from the observations reproduced in Fig. 9. It corresponds to the theoretical form (a-f^)" sufficiently exactly, as is evident by mere inspection. The length of the ordinates is proportional and almost equal to the corrected height of the groups of beans belonging to each compart- ment of the glass case. The number of beans found in each com- partment is found at the foot of the corresponding ordinate. A bean from each group is drawn as a sample to show the extent of fluc- tuating variability in length. The beans are seen to be very variable in form and coloring also. The Various Forms of Variability. 49 from this mean are inversely proportional to their mag- nitude. The variation may be exhibited in size or number and the results of observation can be treated by mathe- matical signs and formulae. Galton, Weldon, Bateson^ Ludwig, Duncker, and many other investigators have raised this line of inquiry to a special branch of science. But, unfortunately, a recognized term for the phenomena with which they deal does not exist. It has been called fluctuating, grad- ual, continuous, reversible, limited, statistical and indi- vidual variability. The latter seems to be the most widely distributed in zoological and anthropological literature, Fig. II. The Ogive-form of the curve of individual variation, made of the leaves of Pvumis Lauro-Cerasus.^ while tlie name fluctuating, which was often used by Darwin^ seems to be the best.^ On the botanical side individual is opposed to partial variability, the former ^ Individual variability can be very simply demonstrated by past- ing the leaves of a tree in a row side by side. They are arranged according to their size and are placed at equal distances along a horizontal base line in such a way that their midribs are parallel ; then their tips are joined by a line. In the above figure this line is placed at a little distance from the tips of the leaves for the sake of clearness. This line (the Ogive of Galton, who has made most use of it) at first mounts quickly, then in the middle only slightly and at the end rapidly ascends again, following Qi'Etelet's law. The points Q, M, Q divide it into 4 quarters (Q = Quartile). ^ See KoLLMANN in Correspondenz-Blatt d. d. Gesellsch. f. An- thropologic, Bd. 31. No. I, Jan. 1900. 50 Miitahility and Individual Variation. meaning the differences between individuals, and the latter the equally frequent differences between the organs of a single individual. The necessity of distinguishing between variability in space and time has been often insisted upon;^ I mean between (a) the diversity in a group of forms existing at the same time, and (b) the differences existing between K% tZJ 13 13Z Tt ns IS ISj 10 ItiJ 17 17.5 18 I3.S (9 Fig. 12. Amount of Sugar in Beets, at Naarden." On January 24, 1896. — On January 25, 1896. On January 28, 1896. parents and their children, and more distant descendants. Ploetz has proposed that contemporary varying indi- ^ W. Waagen, Die Formcn des Ammonites suhradiatns in Be- necke's Geognostisch-Palaontologische Beitriige, 1876, Bd. II, p. 1S6. ^ The three curves exhibit the sugar contents of beets from one large sample taken from three successive determinations on January 24th, 25th and 28th, 1896, by exactly the same method. The num- bers in each lot were 6848, 6781 and 6191, amounting almost to 20,000 polarizations. The sugar contents varied from about 12 to 19 per cent. These figures I owe to the generosity of Messrs. Kuhn & Co., the owners of the factories at Naarden. — In spite of the con- siderable number of values taken the curves do not exactly coincide. The third curve taken 3 days later has its apex shifted a little to the right. The differences between the 2 others are obviously to be attributed to unavoidable chance circumstances. In the compari- son of empirical curves with theoretical ones, a closer agreement than that between 2 curves from 2 samples of the same kind must obviously not be expected. For theoretical purposes therefore one should, where possible, compare two or more curves of the same phenomenon. The Various Forms of Variability. 51 viduals should be called Convariants, successive ones De- variants;^ and individuals departing widely from the mean are often called variants. Individual variability is exhibited by size, weight and number; Ludwig^s countings on flowers conform to Que- telet's laws as accurately as the anthropological meas- urements of that great writer himself. Variations in size and weight should be called quantitative, and Bate- son has proposed for variation in numbers the name discontinuous or meristic.^ Darwin asserted over and over again that this form of variability ^'perpetually occurs." It could therefore be described as perpetual or incessant, and tliis idea seems to me to be best expressed by the word continu- ous."' Individual variability, when tested by sowing, reverts to its original mean, the forms of its variants are con- nected together, are coherent and not discontinuous. It is centripetal inasmuch as the variations are grouped most densely round a mean. Finally — and this is very im- portant— it is linear; because the deviations occur in only two directions — less or more. This fact has given rise to the expressions plus-variations and minus-varia- tions. It is to the selection of the material afforded by in- dividual variability that the origin of many improved * Alfred Ploetz, Die Tilchtigkeit iinserer Rasse iind dcr Schuts der Schzvachen, I, 1895, P- 3i- ^Materials for the Study of Variation. ^ I have used the terms continuous and discontinuous in this sense in my essay Ueher halbe Galton-Curven als Zcichen discon- tinuirlichcr Variation. (Berichte d. d. Bot. Gcs., 1894, Bd. XII, Heft 7). Bateson uses the word in a sHghtly different sense inas- rnuch as he employs the term continuous solely for quantitative, and discontinuous for meristic variations {Materials for the Study of Variation, 1894). 52 Mutability and Individual Variation. races is due. But we must not forget, what we have ah'eady mentioned,^ that the word "race" is used here in a different sense from that in which it is used in anthro- pology. The principal difference between the so-called im- / ^ h . / / /^ f 1 I'f t 1 1 r^\ ! \^^ J \ '■•J ^ 2lMm22 23 Zii 26 27 2a 29 it 32 33 3<> Fig. 13. Exhibition of Variability by the Fan Type of Plotting.^ proved races on tlie one hand, and varieties, subspecies, elementary species, incipient species and so forth, on the other, will form the subject of our third chapter. ^ See page 41. ^ VariabiHty can be exhibited by other means than l)y Quetelet's curve (Fig. 10) or Galton's Ogive (Fig. 11). If it is a question of comparing successive generations with one another the "fan" type of presentation (Fig. 13) is to be recommended. The point from which the rays emanate gives the character of the mother plant. The length of the base of each triangle on the upper hori- zontal line gi\'es the length of the ordinates in an ordinary curve, as they are drawn above in the diagram. This breadth gives at a glance the frequency of individuals for any one scale character. The data for this figure consist of measurements of the length of the ripe fruits of Oenothera Lamarekiana taken in the year 1891 (99 fruits measured in whole millimeters). The lengths were distributed in the following way over the fruits: mm. 21 22 23 24 25 26 27 28 29 30 31 32 2)2) 34 I I 7 8 14 15 12 13 5 7 5 4 4 3 The crooked line follows Quetelet's law (a-J-Z?)". The Various Forms of Variability. 53 4. Spontaneous changes. We have long been familiar in practical horticulture with the phenomenon of the sud- den and unexpected appearance of varieties from time to time. Darwin calls these sudden transitions single variations. The finest examples are the so-called bud variations. The new form arises as a bud or twig on an individual of the old form and often remains a long time united with it. In a case like this there can be no doubt as to the mutual genetic relationship, and the fact that the transi- tion is discontinuous is at once evident. But even in this sphere there is great uncertainty because bud variations are often born by hybrids and the hybrid nature of an individual is sometimes even betrayed only by such varia- tions. Moreover bud variations are very common on varieties with incompletely fixed (mixed) characters, as in many forms with striped flowers (Antirrhinum, Del- phinium, Aquilegia, Dahlia, Fig. 14, etc.) 5. On the Magnitude of Mutations. We often hear of spontaneous changes described as sports or as sport- like variations. This term is not a happy one. Natura non facit saltus, said Linnaeus. But we are not told what we are to regard as a jump. There is much more point in describing the individual transitions as jerks and to speak of jerky variability.^ The jerks may only induce quite small changes, but each jerk represents a distinct unit. Galton has illustrated the difference between jerk- ing and ordinary variability in a very beautiful way. Imagine a polyhedron which can roll on a flat surface.^ Every time that it comes to rest on a fresh side it takes ^ "Variation par secousses" of some French writers. F. Galton, Hereditary Genius, 1869, p. 369. 54 Mutability and Individual Variation. Fig. 14. Striped Dahlias (Dahlia variabilis striata nana), grown from seed. The central flower of the plant in the middle had yellowish red stripes on a pale red ground ; the left flower on the same plant had one-half (shaded darker in the figure) entirely red, the other reddish yellow and pale red stripes. The two separate flowers figured below were of other color varieties ; in the right one red stripes were on a white ground, in the left dark violet stripes on a pale violet ground. The first plant had branches, in 1898, whose flowers were en- tirely red, afl'ording an example of so-called bud-varia- tion. The Various Forms of Variability. 55 up a new position of equilibrium. Little shocks make it totter ; it oscillates round its position of equilibrium and finally returns to it. A slightly stronger push how- ever can make it go so far that it comes to lie on a new side. The oscillations round a position of equilibrium are the fluctuations, the transitions from one position of equilibrium to another correspond to the mutations. The track left behind by the rolling polyhedron can be regarded as the line of descent of the species ; each sub- division of this track, corresponding to a side of the polyhedron, representing a particular elementary species ; each transitional movement to a new position a muta- tion. The more numerous one imagines the sides of such a polyhedron to be, the smaller, of course, are the muta- tions. But this illustration gives no insight into the causes which effect the changes in position. The observation of many single variations has intro- duced the view that mutations must always be consider- able changes and especially that they should be greater than variations. But this is by no means the case, and it appears that many mutations are smaller than the differences between extreme variants. This is imme- diately clear if one compares e. g. Draha vcrna or TypJia angitstifoUa and latifoUa. The single species of Draha verna (discriminated by Jordan, De Bary, Rosen and others), which have been shown by repeated sowing to be constant, differ less from each other than extreme variations in the same characters (form and size of the leaves, petals, pods, etc. ) usually do in other plants ; as can be seen best by comparing them with the partial variations of the leaves of our trees, that is, with the differences between the various leaves of one and the 56 Mutability and Individual Variation. same tree. And Davenport and Blankinship have recently shown in a valuable paper that in the case of Typha latifolia and augustifolia the curves describing their various characters overlap. A small leaf of latifolia can be smaller than the broadest leaf of augustifolia and so forth. Tlie curves overstep the limits between two species; they are transgressive and the species become ''inter grading groups:'^ The differences between the single species of Draha vcrna (Fig. 3) afford one of the best examples for making clear, in a general way, the nature and size of mutations. § 5. THE ELEMENTS OF THE SPECIES. Ever since Darwin's theory of descent obtained gen- eral recognition, the need of an experimental study of the origin of species has always been strongly felt. This demand was always kept in the forefront by the few opponents of this theory, who objected that, so long as it was not possible to produce new species, or at least ob- serve their origin directly, the foundation on which the theory rested was one of unproved h3^pothesis. In the discussion of this objection two entirely differ- ent things are usually confounded. The origin of species is not the same thing as the origin of specific characters. The former is a historical occurrence ; the latter is a physiological one. How, when and where species exist- ing at the present moment arose is a subject for historical investigation, and we can only discover anything definite about it in those rare cases in which records have been kept by contemporary eye-witnesses. The problem of ^ C. B. Daxtnport and J. W. Blankinship, A Precise Criterion of Species; Science, N. S., Vol. H, No. 177, p. 685, 1808. The Elements of the Species. 57 accurately tracing the formation of a given species is certainly a most attractive one ; but its solution falls within the province of comparative biology. The origin of specific characters is a matter for physiological investigation, and is of the very highest importance. AVe hardly know what specific characters are. We know, it is true, that elementary species and form.s closely allied to them, are distinguished from one another not by a single feature but by all their organs and peculiarities.^ The difi^erences between two closely allied forms often demand a long and extensive diagnosis. Nevertheless this diagnosis must be regarded as the ex- pression of a single character, a single unit, which arose as such and as such can be lost; the individual factors of which cannot be manifested separately. Theoretically such a group of characters must be regarded as a unit, as a single character.^ It forms a single side of Galton's polyhedron (p. 53). Darwin called such characters the elements of the species and consequently we may call each of the forms distinguished by such an element, an elementary species. How these elements of the species arise must sooner or later become the subject of experimental investigation. If we once succeed in solving this question we shall ob- tain not only a much surer foundation for the theory of descent but the prospect of the utilization of this discov- ery for the benefit of mankind. The only means by which the breeder can get new forms is by hybridization, and all that he can do by selection is to intensify the produce and yield of characters already present ; but ^ This fact forms a hitherto little noticed support for the theory of homotypic cell-divisions as advanced by Hertwig and others and by myself in my Iniracellulare Pangenesis (See e.g. p. 115). ^ Tniracellulare Pangenesis, p. i6. SS Mutability and Individual Variation. so far it is not within his power to call into existence new characters. We all know that it is said to be impossible to produce a blue Dahlia, a bright yellow Hyacinth and so forth. To give our large flowered Canna white flow- ers we must wait for the discovery of a new white flow- ered wild species and then cross it with it (Crozy) ; in the same way that our Gladioli have been made hardy and the flowers of our Begonias large by crossing them with newly discovered species which possess the character in question. As soon as we arrive at an experimental phys- iology of the origin of species, we expect to obtain con- trol over much that at present seems beyond our reach. But let us return to the facts. Whilst w^e may hope that the origin of new elementary species will one day become the subject of direct investigation, we must be perfectly clear as to the essential difference between these and the so-called Linnean species which are (usually) groups of elementary species. An elementary species can be identified in anv e^iven case bv the test of culti- vation ; how many such forms should be united to one Linnean species is a matter for so-called taxonomic in- stinct, just as is the settlement of the limits of genera and families. Let us return to the rolling polyhedron and look at the track it has left behind. Each piece of it, formed by one side, represents an elementary species, and we will imagine that all such species of a certain strip of the path have left living offspring. The question is where to place the boundaries of a "species" in such a group. Instead of a discussion I shall give the answer which one of the most famous of the older systematists. Hooker, has given in certain definite cases. First in regard to Oxalis corniculata. The forms of this collective species, The Elements of the Species. 59 which grow in New Zealand, have been raised by Cun- ningham into 7 well-defined species ; but since con- necting links between all these seven forms are found in different countries Hooker has united them into a single species.^ Another good example is furnished by Loiuaria pro- cera, a fern from New Zealand, Australia, South Africa and South America. If we were acquainted with the forms from one of these localities only, we should recog- nize, in them, a number of species. But when those from all these localities are compared they form a complete series, and they are consequently united as a single spe- cies. But this species comprises a much larger series of forms than all the remaining species of Lomaria put together. The limits of collective species arise therefore by the dropping out of links in the chain of elementary species. These gaps are apparent when one confines his attention to a single region; and real if they still persist when the Floras of the world have been examined. If the O.valis corniculata or the Loiuaria procera ceased to exist in any one country the present species would have to be split up into smaller ones. Or in other words : Linnean species arise by the dis- appearance of single elementary species from a hitherto unbroken series. This origin is therefore a purely his- torical occurrence and can never become the object of ex- perimental investigation.^ ^'Species" therefore have very little physiological sig- * J. D. Hooker, Introductory Essay to the Flora of N'e^i' Zealand, 1853, p. 18. Compare also Hooker's account of Aconitum Napellus. ^ The famous expression of Spencer, The survival of the fittest, is therefore incomplete and should run the survival of the fittest species. 60 Mutability and Individual Variation. nificance, whereas the study of specific characters will some clay form the most important branch of investiga- tion in the whole domain of biology. ':i^m P¥ / Fig. 15. ZeaMays tunicata {or cry ptosperma) . Three ears from a sowing of seeds from the same ear. The individual seeds are enclosed in the husks ; in A however, incompletely covered about the middle of the ear, and almost naked at the top. C is the intermediate form ; B has, especially l3elow, very large husks. Continuous Variation of Elementary Specific Char- acters.— The difference between fluctuation and mutation perhaps comes out most clearly in this connection. By The Elements of the Species. 61 mutation new characters arise all at once. Such char- acters are however just as variable, and vary in the same way as those specific characters with which we are already familiar.^ There are so many examples of this rule that it is difficult to make a choice.- Zea Mays tiuii- cata or cryptosperina has its seeds enveloped in husks, but the length of these husks varies in a high degree ; some- times they scarcely cover the seed, in other ears they are Fig. i6. Leaves of Soxifraga crassifolia in various degrees of pitcher formation, the succeeding stages being reoresented by the figures 1-5. This process can be imagined as consisting in the edges of the leaves folding upwards and fusing together. The degree of this fusion is seen to be very variable. 3 or 4 times as long, if not more. Very often they are much longer in the lower part of one and the same ear, than in the upper; and their length gradually decreases as the apex of the ear is approached (Fig. 15). Varie- gated leaves, double flowers, pitchers (Fig. 16), split leaves and so forth occur in great variety, and it would not be difficult to demonstrate the applicability of Oue- ^ See also Intracellulare Pangenesis, pp. 69-70. ^For examples from the animal kingdom see Bateson, Mate- rials, p. 68. 62 Mutability and Individual Variation. TELEXES laws to these cases. For each character there is a mean, around which the variants are grouped according to the laws of probabihty. In similar fashion, the split- ting of the leaves in Chclidoniiun laciniatwn, varies; even glabrous and unarmed varieties exhibit a certain de- gree of variability in the extent to which they manifest the character which they are supposed not to possess. (Young shoots of Bisciitclla laevigata glabra, fruits of Acs- cnlus Hippocastaniun incrmis, and so forth.) Five leaved clover {Trifoliiun pratcnse qiiinqiic folium) varies in the number of leaflets between 3 and 7 , obviously following Quetelet's laws.^ The characters peculiar to Papaver soninifcruni polycephalum (Figs. 27 and 28, Chap. IV, § 16) and Papaver bracteatiun monopetahim (Fig. 1, p. 12) are in the highest degree variable. The same is true of the syncotylous and tricotylous races. The widest range of variability and complete immutability are frequently associated.^ Such variation or fluctuation is therefore an occurrence of quite a different order from mutation. // § 6. THE MUTATION HYPOTHESIS. Although I do not intend to discuss the views of my contemporaries on selection and mutation at large in this work,^ I shall now call attention to the fact that ob- jections are continually being raised against the theory of selection from all sides. ^ The authors in question * Over het omkcercn van halve Galton-ciirven, Botanisch Jaar- boek of the Society Dodonaea, X. Jahrg., 1878, p. 46. 'Alimentation et Selection. Vol. Jubilaire de la Societc dc Bio- logie de Paris, Dec. 1899. ^ For a critical presentation of both sides of the question the reader is referred to O. Hektvvig^ Zeit- und Streitfragen der Biologie. * See also Bateson, Materials, p. 567. The Mutation Hypothesis. 63 express themselves more or less definitely in favor of the mutation hypothesis. E. D. Cope was the first to clearly formulate objec- tions against the doctrine of selection. Selection pre- serves the good and weeds out the bad, but whence does the good arise? Obviously ordinary variability is not sufficient, and causes of an entirely different kind must be sought for. Such causes he includes under the term bathmism. Carl Semper similarly rejects the selection theory and ascribes considerable importance to the influence of the environment, the so-called monde ambiant of the French school, in originating useful specific characters. Louis DoLLO was the first to express the view that revolution est discontinue from the standpoint of the theory of descent. He supports his statement by a series of facts, partly zoological, partly botanical, but especially derived from his own researches in paleontology. He puts forth the additional proposition that revolution est irreversible et limit ce.^ According to Wallace^s selection theory, progres- sive change by artificial and natural selection is supposed to be unlimited and even reversible. It must be reversed according to Wallace as soon as the conditions on which the selection depends are themselves reversed. According to the mutation theory, on the other hand, no cause can be assigned which would make a mutation reversible, apart from the loss or latency of characters. Each mutation is a definitely circumscribed unit. About a year later there appeared Bateson's famous work Materials for the Study of Variation Treated ^ Louis Dollo, Lcs Lois dc f evolution. Bull. Soc. Bclgc de Geo- logic, T. VII, p. 164, Annee 1893. 64 Mutability and Indiz'idiial Variation. Especially zvith Regard to Discontinuity in the Origin of Species. The special part of this book consists of an exhaustive catalogue of instances of variations in num- ber, or so-called meristic variations, in the animal king- dom. Variations and mutations in the number of verte- brae, of fingers, of joints of the tarsus, etc., are set forth, and are included under the term discontinuous variations.^ The general part is devoted to a criticjue of the modern theory of evolution. The theory of descent has, according to Bateson^ not merely to account for the kinship of organisms. This point is already granted. But it also has to explain the differences between indi- vidual forms and, with regard to this point, Bateson asserts with perfect right that the species alive at the present day are sharply and completely separated from one another, and that transitions between them either do not occur at all, or at most, very seldom. Existing species form a discontinuous series. The theory of descent, therefore, has to account not only for their relationship, but also for this discontinuity.^ The latter forms one of the weak points in the current theory of selection. For according to this theory, the series of ancestors of any given organism must be a continuous one, seeing that the only differences between parents and offspring are of the so-called individual or fluctuating kind. Whence then arise the gaps which now separate species from their nearest allies? The usual answer that is given is to point to the ex- istence of numerous intermediate forms These however are not transitional forms, but independent types, namely * See especially pp. 568 and 571 ; also pp. 15, 6t, etc. The argu- ment of Duncker (Biol. Ccnfralblatt. 1899, p. 372)^ is therefore not really directed against Bateson's use of the term discontinuous. ^ See pp. 5, 17, etc. The Mutation Hypothesis. 65 elementary species or subspecies. Bateson expressly points out that the law of elementary species holds good for the animal as well as for the vegetable kingdom ; but that these forms have as yet received much too little attention. Elementary species are sharply and completely separated from one another, they do not merge into one another, either in the wild state, or under cultivation (provided that crossing does not occur). This sharp delimitation of the elementary species is so general a phenomenon that it certainly points to a dis- continuous origin. The main object of Bateson's book is to arrange and collect the material in such a way as to give some insight into this discontinuity.^ A very serious objection to the theory of selection is brought forward by him in reference to the usefulness of specific characters." It has been repeatedly asserted by Darwin and others that the characters which separate allied species from one another are not of particular ad- vantage in the struggle for existence, but are as a rule useless and inconsiderable. Nevertheless these differ- ences are often, apparently, very complex and constant characters but "of doubtful value." The existence of such characters cannot be accounted for by Wallace^s theory of selection which explains useful characters in so beautiful and simple a manner. The mutation theory, on the other hand, gives a perfectly simple explanation of the existence of such characters ; for useless, but not dangerous, mutations must appear as often as useful ones, and have almost as much likelihood as these of per- sisting. ^ Species are discontinuous ; may not the variation by which spe- cies are produced, be discontinuous too? p. i8. See also pp. 69,568. ^ Page II. 66 Mutability and Individual Variation. Bateson's conclusion is expressed in the following words : The evidence of variation suggests in brief, that the discontinuity of species residts from the discontinuity of variation.^ W. B. Scott, in an exhaustive critique, has expressed his opposition to many of the views in Bateson's book.- He particularly objects to the statement that species form a discontinuous series. He adduces recent paleontolog- ical discoveries as proof that there are no such gaps in the genealogical trees of the horses or of many other mammals. Such series are only discontinuous when our knowledge concerning them is incomplete. In contin- uous series the progression took place by almost imper- ceptible gradations.^ These gradations seem, however, to be what Bateson calls steps. Let us return to the simile of Galton's rolling polyhedron. The question whether we choose to call this movement continuous or discontinuous depends on our point of view. Even a series of numbers can be unbroken and therefore con- tinuous. The word mutation has been used more in paleon- tology than in other sciences to express the differences between allied species. The actual process of mutation, the change of one species into another, can obviously not form a problem of paleontology. The paleontol- ogist can only study the series of consecutive forms. From such series however important information may be derived as to the size of individual steps, that is to say, the mutations. Waagen has said that the more com- plete the geological evidence is the less perceptible do '^ Loc. cit., p. 568. "W. B. Scott, On Variations and Mutations. Am. Journ. Sci., 8° series, vol. 48, Nov. 1894, PP- 355-374- ^ Page 360. The Mutation Hypothesis. 67 the specific gradations become.^ We can obviously never know how much more numerous, if at all, the mutations have been than the species whose remains we find ; count- less species may have arisen without leaving a trace behind, but whether this is the result of the struggle for existence, of natural selection, or of an advance in a predetermined direction, cannot now be ascertained. Phylogenetic changes make straight for the goal, seldom swerving to the side, hardly ever advancing in a zigzag line,^ but whether natural selection or variation in a definite direction was the determining cause is obviously a matter of personal opinion. The constancy of forms arising by mutation, as op- posed to fluctuating variability, is supported by the re- sults of paleontological research. Waagen as well as Scott and others have declared against Wallace^s se- lection theory on these grounds. They strongly maintain that mutations must be admitted to a more prominent place in any theory of evolution.^ Each ''mutation" (elementary species) serves as a new center of analogous variations. ' Scott deduces from paleontological data a further series of conclusions relative to the occurrence of muta- tions. I find many of these views supported by a critical study of the theory of variation as well as by my own experimental work. I shall have to return to them at the conclusion of tliis section, and in the first chapter of the following one. Last year Korschinsky definitely expressed himself as opposed to the present form of the selection theory. ^Waagen, Benecke's Geogr. PaVdontol. Beitrage, II, S. 170. * Scott, loc. cit., p. 370. ^ Loc. cit., pp. 372, 27Z- 68 Mutability and Individual Variation. He includes mutations or spontaneous variations under the term heterogenesis on the analogy of Kolliker^s heterogenetic reproduction and the "Heterogenism" of Hartmann.^ He bases his conclusions on the data of horticultural practice and gives a complete and very im- portant survey of the cases in which the history of the first appearance of varieties is more or less accurately known, or in which the occurrence of new forms other than by a series of transitional stages points to a sudden origin. Such heterogenetic changes (the mutations of the older investigators) can be progressive or retrogressive, that is the organs can become more complicated or more simple ; both kinds of changes must often happen, but the retrogressive ones can obviously occur more easily than progressive ones. Mutations can be induced, as Darwin also believes, by the cumulative operation of favorable conditions during development and by rich nutrition continued through many generations. Forms that have newly arisen can sometimes be so sharply dis- tinguished from the parent type that any systematist would take them for a separate species if he did not know their origin. Korschinsky concludes from a survey of the facts at his disposal that among garden plants all new forms, or more strictly all new characters, have originated by heterogenesis. New varieties are not obtained in horti- culture by the selection or cumulation of individual differ- ences. Selection is a conservative agency. It fixes new ^ S. Korschinsky, Heterogenesis und Evolution, Naturzvissen- schaftlichc Wochenschrift, 1899, Vol. XIV, No. 24. A larger work appeared in the Mcmoircs of the St. Petersburg Academy of Sciences, 1899, Vol. IX. See also Flora, Vol. 59, 1901, pp. 240-363. (Note of 1908). The Mutation Hypothesis. 69 characters that have ah-eady arisen but it cannot of itself produce new forms. This author then proceeds to a comparison of the fundamental principles of the selection and mutation the- ory. As a result, he finds that the theory that species have originated by the selection of individual differences is beset with series difficulties, whereas the belief that this has taken place by mutations (heterogenetic varia- tions) provides us with a satisfactory explanation or at any rate is in close accord with the facts. Two facts that strongly favor this view are (1) the absence of transi- tional forms and (2) the existence of at least apparently useless characters. Darwin's view is that the probability of a progressive development in animals and plants is great in proportion to the severity of the struggle for existence. Kor- SCHINSKY on the other hand holds that favorable con- ditions afford the best opportunity for the appearance of mutations. For the new forms in order to establish themselves require suitable opportunity for the develop- ment of their powers and fertility to the full. It will be seen that this contrasts strongly with Darwin's view, which is that the innumerable weaker variants simply cease to exist while the rarer stronger ones survive. I shall now close this historical sketch. I hope, at a later date to review more fully the views of modern authors; it will then be seen that the general opinion is that the theory of selection is unsatisfactory. For exam- ple DuNCKER says that individual variability is static rather than kinetic; and therefore does not provide material for natural selection.^ Lord Salisbury said in his presidential address at the meeting of the British ^ Biolog. Centralhlatt, 1899, p. 2,72- 70 Mutability and individual Variation. Association in Oxford in 1894:^ The theory of selection is by no means to be regarded as proven ; for a host of difficulties stand in the way of the acceptance of the explanation of evolution by the accumulation of ordi- nary (individual) variations. And still more recently Rosa,- from the standpoint of critical studies in phy- logeny has insisted on the distinction between mutations and fluctuations, considering only the first group as phylogenetic variations. ^Nature, Aug. 9, 1894. ^D. Rosa, La ridii.zionc progressiva dcUa variabilita c i suoi rapporti coll'estinaione et coU'originc dclle specie, Torino, C. Clau- sen, 1899, P- 93- in SELECTION ALONE DOES NOT LEAD TO THE ORIGIN OF NEW SPECIES. § 7. SELECTION IN AGRICULTURE AND HORTICULTURE. Botanical literature affords very few instances of scientific experiments on artificial selection. And so long as this continues to be the case we shall be thrown back on the experience of breeders. One of the best scientific experiments of this kind is Fritz Mueller's on Selection in Maize. ^ He dealt with the number of rows on the ears (Fig. 17) and started by choosing the ears with the greatest number of rows for sowing; the commonest were those with 10-12 rows; the others group themselves round this fig- ure in familiar fashion according to Quetelet's laws. Among many thousands of ears a single one was found with 18, but none with 20 rows. At the end of three years selection the mean had shifted to 16 rows, while single ears had as many as 26 rows. I have repeated this experiment over a longer series of vears, and have exhibited the result in the form of a genealogical tree, plotting the variability by the fan method as explained in Fig. 13, p. 52. The fans have been reduced by substituting the most essential lines for the numerous triangles of Fig. 13. The middle line of each ^Kosmos, i886, Vol. II, p. 22. 72 Selection Does Not Lead to Origin of Species. fan corresponds to the mean (the apex of the curve), the two dotted Hnes to the quartiles Q and Qi ; between which therefore He that half of all the individuals coming nearest to the average. The two outer lines of each fan denote the ears with the greatest and smallest observed number of rows; their divergence is of course largely m^ m'^:-- Fig. 17. Zca Mays. Ears with 8, 16 and 22 rows of seeds from my experiment in selection carried out from 1 886- 1 894. dependent on the size of the harvest ; this amounted on the average to about 200 ears a year. Of the two lines going from top to bottom the right- hand one represents the numbers of rows on the ears actually chosen for sowing. That is to say there were sown seeds from ears with 16 (1887), 20, 20, 24, 22, 22, and 22 rows ; ears with a greater number of rows Selection in Agriculture and Horticulture. 73 were usually too poor in seeds to be of any use for continuing the experiment. i ?o » n K 19 20 it ?♦ te u 1S9i 1893 1392 1391 1S89 ias8 1SS7 1886 8 10 12 1h 16 IS 20 2if 2t> Fig. i8. Pedigree of an Experiment in Selection with Maize. The numbers at the top and bottom of the figure show the number of rows in the ear. The experiment began in 1887 ; the hnes for 1886 give the characters of the race with which I started. The left-hand line joins the ends of the middle rays, which correspond to the apices of the curves for each 74 Selection Docs Not Lead to Origin of Species. year. It shows us therefore the mean progress of the yield. This hue comes closer every year to that of the ears used for seed. A proper study of its course would have necessitated the undertaking of effecting artificial self-fertilization which in the case of maize is known often to lead to a poor harvest. I should have preferred to have used another plant, had not the ears of maize lent themselves so admirably to a demonstration of this kind. It is not necessary to repeat that the experience of breeders provided the main support for Darwin's selec- tion theory. In its broad outlines the process of natural selection is like that of artificial selection. But as soon as we come to dissect the component factors of these processes we encounter serious difficulties, as I have al- ready remarked in the Introduction. The chief cause of this state of affairs is to be sous^ht in the circumstance that breeders rarely work with single characters but have as a rule aimed at improving their plants in every possible respect, They never attempt a separation, or even a separate observation, of special characters. The second cause is that breeders have no interest in distinguishing between their different methods of improving their plants. On the contrary it is usually the best plan to allow the various methods : of the choice of desirable mutations, of their gradual improvement by repeated selection, of natural or artificial crossing, of manuring and what not, to exert their combined effect. The breeder is only interested in the result. The means by which this has been attained is a secondary matter and is seldom thought worthy of an accurate record. Selection in Agriculture and Horticulture. 75 In choosing cases for the scientific study of the pro- cess of selection it is of the highest importance to ex- ckide all those in which crossing has taken place, or where it has not been excluded with absolute certainty. Many genera and species owe their present range of forms (which is what breeders call variability) almost entirely to the repeated crossings between the original wild forms that w^ere introduced, whether these were different Lin- nean species or numerous elementary species of such. There are two chief categories to be distinguished. First there are those genera in which a very wide range of form is desired, and for this purpose almost every conceivable cross between the different varieties has been carried out. The best, either from the utilitarian or from the decorative point of view, are then put on the market and are, in the eyes of the layman, an inscrutable medley. Fuchsias, dahlias, chrysanthemums, wheat and potatoes are the best known examples. The novelties of the breeders arise in these cases almost without ex- ception by the deliberate combination of characters al- ready existing in the old types. Secondly there are the genera which have developed in a definite direction since the beginning of their culti- vation, e. g., begonia, gladiolus, caladium, amaryllis, canna and many others. The improvement in these cases has almost always been the result of the discovery of new wild species. These have been crossed with the cultivated bastard race and in this way the desired char- acters of the former have been transferred to the latter. The large and beautiful blossoms, the caladiums with variegated leaves, the hardy gladioli etc. have been got in this way. The characters of the new varieties already existed in nature, distributed between the different spe- 76 Selection Docs Not Lead to Origin of Species. cies. The combinations are new in cultivation ; but the characters themselves do not owe their origin to it. Of course I do not deny the appearance of mutations in cultivation, but so far as 1 have been able to learn personally from the best known breeders these are rela- tively rare occurrences. It is impossible to insist too much that the much talked of progress in cultivation is a delusion if the part played by crossing is left out of account or if the results of this crossing are regarded as the effect of selection. And this happens only too often. Hybridization is so much more certain and easy a way than selection of get- ting something new that breeders would nearly always be working against their own interests if they did not expose their plants as freely as possible to natural cross- fertilization. The possibility of crossing should evi- dentlv onlv be excluded when we are concerned with the fixation of races or with methodical selection carried out according to strict principles. But it is only experiments of this kind that have any value for the theory of selec- tion. Unfortunately they are much more rarely carried out, or at least more rarely described than one could wish. Most of the brief notes made by breeders on varia- bility, which is apparently considerable, are open to the objection that the seeds in question have been collected from plants fertilized by the wind or by insects. And if, for example, we read through the mass of material col- lected by Darwin, with this in mind, we shall find that much that seemed to be variability or mutability receives a much simpler explanation by supposing that it was the result of crossing or of the collection of seeds from hybrid plants. It will always be found that the number Selection in Agriculture and Horticulture. 77 of cases of alleged variability in plants whether in the wild or cultivated state suffers a considerable shrinkage so soon as one views the individual statements from the point of view of possible chance crossings.^ I maintain, in a word, that much that has up to now been alleged as evidence of variability overstepping the limits of elementary species (that is mutability) should really be attributed to the result of unobserved chance crossings." It is worth while to draw attention to the further distinction between agricultural and horticultural selec- tion. For a clear perception of the relations existing between them w^ill facilitate our understanding of the difference between fluctuations and mutations. Every year there are put on to the market by pro- fessional gardeners a certain number of so-called novel- ties especially of plants propagated by seeds, for it is these that I have particularly in view.^ They are partly hybrids, partly really new varieties and subspecies, partly species brought from foreign lands. The varieties and subspecies arise suddenly and only a few individuals of them occur as a rule. They seldom appear in the nursery gardens but usually in those of customers, whose total area is of course much greater than that of the firms which supply the seeds, and where as a rule much more time and attention is devoted to the individual plants. * See also Hoffmann, Botanische Zcitung, 1881, p. 381. "The seeds of isolated flowering examples have shown no tendency to the formation of variations." ^ In a later section T propose to deal with this comparison ex- haustively and from the point of view of careful experiment. ^ The commercial aspect of breeding has been most thoroughly gone into by C. Fruwirth, Ziichtungsbcstrehiingcn in den Vcreinig- icn Staaten, in Fuhling's Landzvirthsch. Zeitung, 1887, Jahrg. s^^ p. 16. 78 Selection Does Not Lead to Origin of Species. The nurserymen then usually buy the novelties from the customers in question at a considerable price. As a rule 4 or 5 years elapse before such a novelty is put on the market. During this time, so we are told, it has been made constant by selection. It would be more correct to say that they are freed from the adulterating effects of free crossing. For the selection consists in weeding out the so-called rogues at the time of flowering (sup- posing that we are dealing with flowering plants) in order to save seed only from the pure individuals. But if we are dealing with vegetables the selection takes place long before they are in flower, so that there is no danger of crossing in this case. These rogues are nothing more than hybrids resulting from free crossing in the preceding summer. I have often had the opportunity of watching this weeding out. It takes place at the height of the flowering season. The pure individuals may therefore have been already partly fertilized by the rogues ; and for this reason some of the seed for the next year may have been contaminated. The sole object of the selection is to reduce the mixture with other forms to a minimum ; the pollination is left to insects from the first generation onwards, so that cross- fertilization always takes place. I have never been able to find that in ordinary cases selection had any other object than the purification of the new race from the effects of mixed ancestry. The use the gardener makes of his 4 or 5 years is to increase his stock of seed suflicientlv to make it worth while to put it on the market. This is in fact a far more important matter than the process of purification that we have been speaking of. As soon as the requisite quantum of seed is obtained it is put on the market. Absolute Selection in Agriculture and Horticulture. 79 purity is not guaranteed. I have often bought seeds of novelties and tested their purity by extensive sowings. They almost always contain impurities. But whenever I fertilized some specimens of the new forms with their own pollen after taking care that the visits of insects were excluded, they came absolutely true in the next generation. We all know that we are lucky if we get a purity of 97-99% ; the remaining 1-3% are, we are told, atavists; as a matter of fact they are practically always the survivors of the impurities that owed their origin to free crossing in the field. The whole profit on a novelty must be made during the first year of its appearance in the trade. ^ For as soon as it bears seed in other gardens its originator loses the monopoly of it. For this reason novelties are usually offered for sale towards the end of the year in special price lists to as many seedmen as possible ; they introduce them into their catalogues, and that is why one usually finds that most novelties are put on the market simul- taneously by numerous firms. Their price is at first con- siderable, but in a few years sinks to the normal, for by that time as much seed as wanted can be produced everywhere. \^ A horticultural novelty, when it has once arisen and has been freed from the results of crossing and put in sufficient quantity on the market, is everybody's plant. All that remains to be done to keep them constant is to avoid foreign pollen. The case of agricultural varieties, on the other hand, is quite different. I am referring now only to the gen- uine improved races and give the description of their pro- *I have often heard the vaUie of such a novelty set at iioo to £150. 80 Selection Does Not Lead to Origin of Species. duction according to the now current views. ^ They do not arise by chance, they are not the resuh of rare and sudden variations. The material out of which they are made is furnished by fluctuating variabihty. At the outset, the breeder seeks in his fields for those plants which seem the best for his purpose, and collects their seeds sep- arately. These plants differ very little in the eyes of a layman from the other specimens in the field. He sows seeds from these on a small scale, working every year on the same principles, in this w^ay gradually increasing the deviations from the original form in the desired direction. He has as a rule one or two qualities chiefly in view, but pays attention where possible to all other characters. He is not concerned with the improvement of one par- ticular quality. To achieve this many things are neces- sary, patience, an intimate acquaintance with the species of plants in question, and a firm and clear conception of the ideal to which he wishes his race to attain. And in spite of the possession of these qualifications the best known breeders are by no means successful with every experiment ; the greatest of them, that is those who have introduced the most widely distributed races, liave often only brought out one or at most a very few successful novelties. The value of such a race gradually increases. At first as seed for one's own purpose, but soon as seed for the market. But the seed is not put on the market in a single year but gradually during the period of improve- ment and multiplication. The improved characters de- ^ For the earlier constant products of selection, e. g., those of Patrick Shirreff, and for my own views concerning the description given in the text see the conckision of § 12, pp. 109 ff. and § 23, pp. 178 ff. (Note of 1908.) Selection in Agriculture and Horticulture. 81 teriorate as soon as the new race is cultivated on a large scale, on account of the consequent cessation of rigid selection. The harvest has therefore less value than the original sample of seed. In this vv^ay the breeder is assured the monopoly of his prize for many years until, may be, his race is superseded by another and a better one. The work done and the profits made by the horticul- turist are insignificant compared wnth those of the agri- culturist. The former introduces a few novelties into the garden every year. The latter increases the yield of whole countries. I have often heard farmers speak with pride of their results as compared with those of gar- deners. Finally I would mention a good example of the dif- ference in question. Beseler in Anderbeck by years of patient work improved his oats to such an extent that he was able to put them on the market under the name of Anderbecker Oats. This form was bearded, a feature which was found fault with from many quarters, and prejudiced its sale. It was a small matter to make Ander- becker Oats beardless, provided that beardless examples could be found. This happened to be the case ; and since that time Beseler's oats have been beardless.^ This difiference between the practice of agricultural and horticultural breeding has in my opinion been largely responsible for the present form of the scientific theory of selection. That which can onh^ be achieved by a few and at the cost of great sagacity and patience, produces a great impression ; that which chance can put into the hands of any one, makes none at all. And so it comes about that the former method has loomed much larger *v. RiJMKERj Getreidezuchtung, 1889, pp. 60, 75, and 94 S2 Selection Does Not Lead to Origin of Species. in our discussions on the origin of species, whilst the latter has been relatively neglected. But it must not be forgotten that the agricultural improved races do not possess the constancy of true species •} whereas the vari- eties and subspecies of the horticulturist can only be distinguished from true species historically and systemat- ically— not experimentally. In conclusion : we see that in estimating the value of the experience of breeders for scientific purposes we have to fix our attention on the simplest processes. Every- thing that can be considered the direct or indirect result of crossing must be excluded before we consider its bear- ing on the theory of mutation or selection. Furthermore one must sharply distinguish between the races that have been produced by continued selection, and the constant forms which owe their oriHn to a sudden fortuitous 'fc.' change. In horticulture varieties arise by mutations, and vari- eties are elementary species. In agriculture according to the current view and excepting in the instances of the unconscious isolation of elementary species, the highly improved races arise gradually through selection, but they never become species. § 8. SELECTIVE BREEDING FOLLOWED BY VEGETATIVE PROPAGATION. We shall now proceed to deal with the scientific sig- nificance of selection in those cases in which its products are multiplied vegetatively. * Or if they do prove to be constant, they usually turn out to be the result of the unconscious isolation of elementary species ; com- pare Nilsson's results, described in § 12, pp. 114 ff and Archiv fiir Rasscnbiologie, April i, 1906. (Note of 1908.) r Selective Breeding and Vegetative Propagation. 83 Properly speaking these cases have no significance for the theory of descent. But they are so much more striking than the results of selection in seedplants that they are often used as examples. If after extensive sowing, or after repeated selection of any species one gets a single example with large flowers or fruits or with any desirable character in an exaggerated degree there are two possibilities. First one may be dealing with a seedplant, that is with a species which can either be propagated only by seed, or in which it is usual to propagate it in this way in practice. Secondly, one may be dealing with a plant which is capable of vegetative propagation whether by division of the rhizom, by cuttings, by grafting, by tubers, or by any of the other ways in which this may be effected. In the first case the seeds conform to the law of re- gression. This was recognized by Vilmorin and after- wards scientifically studied by Galton. If we regard Galton's formula as generally true the mean of the offspring deviates from the mean of the type in such a way that it retains only a third of the deviation of the parent. So that to produce a given advance in the whole family we should have to sow seed from a plant which had advanced three times as far. To make the meaning of this regression clear I will select as an example a culture of Madia elegans. The mean number of ray-florets on a flower head is 21. and the other numbers are grouped round this in accord- ance with Quetelet's law. In the 1892 crop of my ex- periment the mean was 21 and the variation lay between 16-25 ; of these I chose 6 examples, each possessing 16-19 rays in the terminal head. From their seeds T obtained 84 Selection Does Not Lead to Origin of Species. a series in 1893 varying between 12 and 22 and with a mean of 19 rays. I now chose the seed from 13-rayed plants and got in 1894 a generation varying between 13 and 22 and with a mean of 18. The regression amounted in this experiment to about % ; that is to say the children deviated only one-third as much from the type of their species as their parents did.^ 1S0^. Curves of large and small amplitude. A, curve of the variability of the number of ray-florets of ClivysantJiemmn segetum growing wild. B, the same curve describing a crop ob- tained in 1894 by sowing the seeds of a 13-rayed plant. The numbers at the feet of the ordinates correspond to the number of ray-florets in the inflorescence. NUMBER IN THE AFTER OF RAYS OPEN SELECTION 6 0.3 0.0 7 0.3 0.0 8 6.8 0.0 9 4-3 0.3 10 3-1 0.9 II 7-1 2.3 12 9.9 9-3 13 34-2 65.3 14 14.2 14.8 15 8.0 ^■Z 16 Z-7 i-S 17 5-2 1.2 18 0.9 0.9 19 0.9 0.3 20 0.9 0.6 21 0.0 0.3 In Fig. 32 these two series of figures are exhibited graphically as curves. It will be seen at once that the dotted curve which describes the result of the culture in 1894 has a much higher apex and is much steepe^r than the other; that is, it has a much smaller amplitude. In other words, the deviations from the mean in plants growing in the field are greater in size and number than they are among the children of a single plant, even when this plant bore exactly the mean character of the type. It is clear that a generation corresponding to curve Variation and Adaptation. 153 A will have less difficulty in finding conditions to suit it than a less variable one, such as might be described by curve B. The offspring of seeds of varying parents are therefore at a considerable advantage. And now we come to the significance of crossing. The essence of fertilization is not the union of the two sexes but the mixture of the heritable characters of two individuals with a different past or at any rate of indi- viduals which have been subjected to different external conditions. The advantages accruing from the fusion of different variants afford, in my opinion, a fair ex- planation of the existence of sexual reproduction.^ Darwin's well-known aphorism : nature abhors per- petual self-fertilization, does not seem to me to express the matter quite exactly. It is not sufficient that isolated crossings should occur from time to time ; on the con- trary, it is necessary that a certain percentage of indi- viduals should always be crossed. For in this way varia- bility will be increased ;^ but the point is not that its range should be as wide as possible but that it should be main- tained at a limit which the environment demands.^ The degree of the deviation of the individual is al- ready determined in the seed. But seeds differ among themselves not only in relation to the characters of their parents, but according to the position on the plant itself and according to their weight. The significance of these factors and their bearing on variability has often been the subject of research ; numerous isolated papers on this subject exist but they need a comparative and critical ^ Intracelhdare Pangenesis, p. 29. ^ A. GiARD in Comptcs rendus de la Soc. de Biologie, 4 Nov. 1899, p. 2, and LiGNiER in Festschrift su Ehren Giard's, Nov. 1899. ^ See especially Ammon, Der Ah'dnderungsspielraum, loc. cit., P- 53. 154 Controversial Questions. treatment.^ Far too little attention has been paid to the relation between the range of variation of the individual characters and the degree of their adaptation to changing conditions of life; and the whole matter is still very much of a mystery. Here again it is probable that further study will tend to emphasize the fundamental distinction between variability and mutability. § 19. VARIABILITY IN MAN, AND SOCIAL QUESTIONS. A noteworthy feature of the last few decades has been the attempt to apply the results of evolutionary in- vestigation to the solution of the great problems of hu- manity and social life. Many have followed along the lines which the great English philosopher Herbert Spen- cer laid down ; and a considerable mass of literature has accumulated on this subject. There are at least two im- portant schools in this field of research. Otto Ammon is the founder of one of them: his method consisted in the application of the results of statistical investigations. The other, and much larger school, is that which aims at the application of biological, and particularly of zoological knowledge to the solution of social problems. Ammgn^s method seems to me to be justified by the fruit it has borne; but the writings of biologists in gen- eral and zoologists in particular seem to me to fall short of a desirable standard of lucidity and directness.- Many mistakes may in the future be avoided if a clear distinction be drawn between mutability and variability in the ordinary sense. * See Von Rumker, Der wirthschaftliche Mehrwerth, loc cif., pp. 140-141. ^ A general account of the methods and results of this school, and a bibliograph}- will be found in O. Hertwig's essay, Die Lchre vom Organismus iitid Hire Beziehimg zur Sozialwissenschaft, 1899. Variability in Man, and Social Questions. 155 The variability exhibited by man is of the fluctuating kind : whereas species arise by mutation. The two phe- nomena are fundamentally different.^ The assumption that human variability bears any relation to the variation which has or is supposed to have caused the origin of species is to my mind absolutely unjustified. Man is a permanent type, like the vast majority of species of animals and plants. The laws for permanent types apply to man ; though often with a qualification. But the laws which describe the changes by which indi- vidual permanent types arise cannot be so applied. As we have seen it is characteristic of these types to exhibit a certain amount of fluctuating variability. Man is no exception to this rule. Therefore all that we can apply to the treatment of social questions is our knowledge of ordinary variability. The facts of specific differentiation are interesting but not relevant. The mental qualities of the human race are closely bound up with their bodil}^ organization, and this has been shown to conform to the same laws as those by which we describe individual variability in plants and animals. Of late 3^ears, Kollmann has done more than any one else to insist on the distinction which should be made between persistent racial characters and fluctuating intra-racial characters in the case of man — a distinction which was also emphatically maintained by Virchow.- Favorable and unfavorable conditions of life, migra- ^L'Unite dans la Variation, loc. cit., p. 17. ^Kollmann, Die angebliche Entstchung netier Rassentypen in Correspondenzblatt der d. Gesellsch. fiir Anthropologic, Vol. 31, No. I. Jan. 1900. p. I. A bibliography of the subject will be found on P'lge 5- 156 Controversial Questions. tion to a different climate and so forth affect the fluc- tuating cliaracters of man to no small extent. But only for a time; as soon as the disturbing factor is removed, the effect which it produced disappears. The morpho- logical characters of the race on the other hand are not in the least affected by such influences. New varieties do not arise by this means. Since the beginning of the diluvial period man has not given rise to any new races or types. He is, in fact, immutable, albeit highly variable. In order to attain to some insight into the causes and significance of individual differences in man we must study the corresponding differences which are presented by an assemblage of forms belonging to a single species of animal or plant. Here is a wide and fertile field open for investigation ; but one in which the harvest of in- formation has been poor so far. Ammon, as we have already said, is the most con- siderable of the anthropological writers on this subject. Although he does not distinguish between the theories of selection and mutation, he sees clearly that our knowl- edge of the origin of species in nature has no bearing on social questions. And as it is on this point that most sociological writers are in error it will be worth our while to pa}^ some attention to his actual position.^ Ammon sets forth the modern theory of selection in five theses of which the first four deal with heredity, variability, the struggle for existence and elimination of the unfit (Natiirliche Auslese).^ The fifth thesis deals with the theory of descent. It runs : ''The forms and characters which, having arisen ^ Otto Ammon, Die Gcsellschaftsordnung und Hire nafilrlichen Grundlagcn, 2d edition, 1896, pp. 9-10. ^This happy phrase of Ammon is eminently preferable to Natiir- liche Zuchtwahl. Variability in Man, and Social Questions. 157 as the result of variability, are favorable to the survival of the individual increase in relative number by the nat- ural elimination of unfa^vorable ones. New varieties and species arise by the gradual accumulation, generation by generation, of the favorable deviations from the original type. And then he adds, 'The substance of the fifth thesis is often challenged on the ground that we are not in a position to state that deviations from a certain type can lead to the origin of a new species by the elimination of the unfit. Fortunately we need not wait for the settle- ment of this controversy. I have only enunciated the 5th thesis in order to give a complete survey of Darwin's theory; but it has no bearing zvliatsoever on our present socio-anthro polo gical inquiry/' This is not the place in which to go further into this question. The danger of the application of the theory of descent to social questions has already been pointed out by men who are cjualified to express an opinion. Quite lately Karl Pearson has severely criticized Ben- jamin Kidd's book on social evolution which is often recommended in England as the best up-to-date work on the subject. If the reader is not clear as to what is meant by the dangers, to which we have referred, which attend the application of the so-called scientific method to the treatment of these problems he will do well to read this critical essay carefully.-^ So long as it is impossible to investigate the social qualities of man directly it must suffice to do what we can by analogy. Material for this argument is afforded by the study of variability in the stricter sense of the ^ Karl Pearson, Socialism and Natural Selection, The Fort- nightly Review, 1894. 158 Controversial Questions. term ; but our knowledge of the mode of the origin of species will not help us in this investigation.^ The study of variability, in plants and animals, as well as in the physical characters of man may thus serve a higher pur- ix)se. It is singularly fortunate, in the present state of affairs, that these analogies should be limited to varia- bility as opposed to mutability. Variability is accessible to investigation from many points of view, which is far from being the case with mutability. Many principles in variability have been discovered and dealt with by OuETELET and Galton and their followers : the methods of this school can be partly applied directly to the in- vestigation of mental characters and partly effect a con- siderable simplification of treatment. There lies here a wide and fertile field of investiga- tion, especially for botanists.- One of the most impor- tant conditions in experiments on selection is the num- ber of individuals in each generation ; and plants readily lend themselves to cultivation by hundreds without any of the ill effects which usually attend overcrowding. Such experiments are Avell-nigh impossible in the case of ani- mals : and out of the question in the case of man. Here, as in many other spheres, the botanist must take the lead and the zoologist and anthropologist will follow after- wards. Of late years the statistical study of variability has become specialized as a distinct branch of science thanks to the labors of Bateson and Weldon among zoologists, LuDWiG among botanists and Karl Pearson and Dunc- ^ See also H. J. Haycraft, Darivinism and Race Progress, and further, on the possibility of replacing selection by improved nutri- tion : L'Unite dans la Variation, p. 21. ' L' Unite dans la Variation, loc. cit., pp. 14-15. Some Subjects for Future Investigation. 159 KER among mathematicians. Botanical work in this field, has also been done by Verschaffelt, Burkill, Haake, Davenport, Blankinship, Mac Leod and many others.^ Let us summarize the foregoing discussion. The mental and moral characters of men exhibit fluctuating variabihty. The laws therefore which describe this phe- nomenon can be profitably applied to such characters. And we shall have to be contented with this manner of treating the subject so long as a direct investigation by biometric methods, and by experiments in selection are out of the question. The foundations of sociolog}^ must be furnished by biology. And we may hope that the time is not far distant Avhen a fruitful cooperation be- tween these two sciences, apparently so much akin but actually so far apart, may be brought about. But no theory of the origin of species can have any bearing at all on this subject. § 20. SOME SUBJECTS FOR FUTURE INVESTIGATION. In the preceding discussion I have had occasion to draw attention not merely to the splendid achievements of my predecessors but also to the numerous gaps in our knowledge. The study of variability as opposed to mutability is a branch of human knowledge which has developed witli great rapidity in the last few years. The statistical method of dealing with this phenomenon is, as we have ^ A survey of the literature on this subject has been given by G. DuNCKER, Die Methode der Variationsstatistik ; Roux's Archh' filr Entzvickclnngsmechamk, Vol. VIII, 1899, p. 167; and by Oster- HOUT, Problems of Heredity in Contributions Bot. Semin. Univ. California 1898. 160 Controversial Questions. already said, firmly established : comparative and experi- mental methods are just coming on. I propose, therefore, to suggest a series of problems the solution of which would in my opinion throw much light on the essential difference between mutability and variability. 1. More examples of Quetelet's law are wanted: their number can never become too great. 2. The curves in question should be plotted from the same individuals or from the same batch of individuals in successive years. The constancy of their means and their amplitude (Galton's 0 and Q') should be deter- mined. Changes in these values, and changes in the symmetry of the curves, should if possible be traced to their causes. 3. Polymorphic curves should be looked for and ana- lysed. These may point to the existence of mixtures of perfectly distinct elementary species growing together or to the existence of antagonistic characters within the limits of a single species (for examples annual and bi- ennial forms in Daucits, Beta, etc.) They may also be due to diseases. Finally they may be the so-called ''double curves" in which the several apices are to be regarded as ordinates in a curve of a higher order, and not as indications of mutation. 4. Correlative variation is a phenomenon of the high- est importance.^ For example man presents many ni- stances of correlation between mental and physical char- acters. Correlations fall in two categories. In the one are those cases in which the two characters are dependent in the same way although not to the same degree on ex- *J. H. BuRKiLL, Variation in the Number of Stamens and Car- pels, Journ. Linn. Soc. Bot, Vol. 31. Some Subjects for Future Investigation. 161 ternal conditions. In the other are those cases in which variation in one character is the cause of variation in another/ as for example the various phenomena of growth which are correlated with differences in photo- synthetic activity. It is superfluous to refer the reader to Galton's method of studying correlation.^ 5. The relation between external conditions of life and variability ought to be investigated. Are there vari- ations which are independent of such, or are there not? If there are, what are their causes? Do the individual external factors exert a separate influence or not? Is there a definite relation between the extent of this in- fluence and the magnitude of the variation? Do all characters under the influence of high nutrition vary in a plus direction, and under a poor one in a minus direc- tion ?^ 6. The sensitive period in the development of char- acters should be determined. When the rudiments of organs are visible under the microscope it is usually too late to exert any restraining influence on their develop- ment. But there may be exceptions to this rule. During the time which a character takes to develop there is prob- ably one short period of extreme susceptibility; and this may be gradually attained and gradually lost. Here is matter for much interesting inquiry. 7. Galton^s regression is very important. Suppose we sow seeds of a self-fertilizing plant: and suppose that we know the amount by which it deviates from the ^DuNCKER, Roux's Avchiv, Vol. VIII, p. 163. ^ Galton, Natural Inheritance, and Proceedings Royal Society, Vols. 40 and 45 ; and Ed. Verschaffelt, Correlaticve Variatie by planten. Botan. Jaarboek, VIII, p. 92. ^Variability can also he influenced by grrafting and inoculation. See L. Daniel, Compt. Rend., 1894, T. CXVIII, p. 992. 162 Controversial Questions. mean of its ancestors in respect of certain characters. Then we determine the curve describing the result of our sowing. As a general rule, the mean of any char- acter in the filial generation departs less from the normal, than the character in question borne by the parent plant does. According to Galton, the relation between these two deviations is a constant one : the mean deviation of the children amounts to about a third of that of their parents. The question whether this is a universal prin- ciple naturally suggests itself; the experiments which I have made hitherto seem to point to the conclusion that it probably is. 8. Does this regression remain the same even when selection is continued for several generations? In other words, does the mean of a race never amount to more than a third of the value attained by the seed bearers chosen in every generation? Does the race in spite of its improvement persist in this relation to its progen- itors, that is to say, does it lag at every generation rela- tively further behind the selected individuals w^hich pro- duced it ? It seems to do ; at any rate the decision of this point dominates the theory of the origin of species by the natural selection of individual variations. 9. OuETELET^s law cuablcs us to calculate from a curve of variation the number of individuals that will exhibit a desired degree of deviation from the mean. It seems that this chance even in the case of small differ- ences is a very remote one demanding as it does millions of individuals. At any rate it is desirable to make such calculations for as many cases as possible.-^ 10. Artificial selection is a device for reaching a cer- ^ See DuNCKER, Biolog. Centralbl, 1898, p. 571. For each addi- tional 1000 individuals the range of variation only increases as from I to 1.049. Some Subjects for Future liwestigatioii. 163 tain magnitude of deviation from the average, with a minimum expenditure of trouble. Is this its only sig- nificance? Does the number of individuals with the undesired qualities diminish exactly at such a rate as we can calculate beforehand? That is to sav, is reo^res- sion independent of the ancestry of a given parent; in other words, does it make any difference whether the seed parent is the result of repeated selection, or is picked from a single sowing on a much larger scale? 11. In such experiments attention should be paid to one character and one only ; although interesting results may often be obtained by measuring a second or even a third character as a sort of collateral inquiry. The selections carried out bv breeders involve as manv char- acters as possible ; on account of correlations the improve- ment of the chief features can be carried further in this way, than would otherwise be possible. Such experi- ments should be made with a purely scientific end in view. 12. In starting an experiment attention must always be paid to the individual vigor of the seed-parents. If this does not happen to coincide with the desired devia- tion, it is advisable to take both the strongest individuals and those exhibiting the greatest deviation, as seed- bearers and to compare the posterity of the two. 13. There is a particular kind of selection experiment which should be carried out a great deal more than it is. I mean one which would start by choosing as seed- parents plants with the smallest petals, the smallest fruits, those with the least degree of hoariness or the least num- ber of spines, with the palest color in their petals, with the smallest number of stamens and carpels and so forth. According to the theory of natural selection such an ex- 164 Controversial Questions. periment should result in the origin of apetalous, fruit- less, glabrous, spineless, white-flowered, unisexual or sterile plants and so forth. Whereas of course on the mutation theory this would not happen; provided that crossing was rigidly excluded from the experiment. 14. What we must aim at is a complete control of variation. We must become so thoroughly acquainted with the underlying factors that we can predict the re- sults of our experiments. V. THE ORIGIN OF SPECIES BY MUTATION. § 21. SPECIES, SUBSPECIES AND VARIETIES. We saw in the second chapter that species cannot have originated by the natural selection of the extreme variants afforded by fluctuating variability. We have therefore now to show that the observations which have been made on this subject can be simply and completely explained on the hypothesis of sudden changes. When such transformations occur among cultivated plants — and they often do — they are called spontaneous or, as Darwin called them, single variations : moreover they are almost always inherited, if not in their entirety, at any rate to a very considerable extent. We may express therefore the essence of the Muta- tion theory in the words : "Species have arisen after the manner of so-called spontaneous variations.'' And in our critical survey of the facts we therefore have to con- sider how far the information at our disposal justifies this view. In order to be qualified to discuss this question we must first of all make quite sure what we understand by the term "species" and, more important still, we must form a clear idea as to which forms we are going to re- gard as the units of the natural system. For it is only in the case of the real units of the system that we can 166 The Origin of Species by Mutation. hope to obtain experimental proof of their common de- scent : the theory of Descent as apphed to groups of these units is, and will probably always remain, a comparative, science. At the time when the Linnean view that species had been separately created was generally accepted, it was naturally a very important matter to decide which forms should be regarded as species. I have already endeavored to give some account of the broad features of the con- troversy which raged round this question during the period just before Darwin's work appeared.^ Since the hypothesis of special creation of species was given up, the view that the Linnean species really were the units of the system was fostered by the persistence of binary nomenclature. But w^e are liable to forget that these species do not correspond to the units which exist in nature, but to groups of them. This is a fact which is clearly recognized and repeatedly asserted by the best systematists.^ Linnaeus himself, as we have seen, regarded his species as groups^ and not as simple things, and De Candolle often speaks of them as col- lective. The classification of plants into groups called species has exactly the same value and meaning as their classi- fication under the headings of genera, families and so forth. So long as our knowledge as to what are the real units of the system is as incomplete as it is at pres- ent, systematists and students of distribution, no less than evolutionists will have to be content to deal with ^ See chapter II, pp. 16-28. ^Alph. De Candolle, La Phytographie', and De VOrigine des Especes cultivees, 1883, p. 372. ^ A good example of this is afforded by the species Homo sapiens. species, Subspecies and Varieties. 167 the compound Linnean species and to regard the small local or elementary species as subsidiary to them.^ But it is clear that this conception of species must result in incomplete investigation and in fallacious con- clusions. For example it is well known that the geo- graphical distribution of species is analogous to that of genera; but it is evident that we should go far astray if we forgot that species like genera were collective enti- ties. The distribution of elementary species, in the geo- graphical region of the Linnean species which they com- pose, is very rarely made the subject of inquiry, yet it is just this point which is of the very greatest signifi- cance as bearing both on the origin and distribution of organisms. According to Jordan every species, as well as every genus, has a geographical center where the distinct component elementary species are most abund- antly represented, growing as they do close together on the same spot, whereas at the circumference of the region inhabited by the species its elements become few and far between.^ It is the actual theory of descent itself that would profit most by a proper appreciation of the conception of species. This theory which is recognized in mor- phology, embryology, in systematic work and in com- parative anatomy as the guiding principle of all specu- lation and inquiry has remained almost without influence on experimental biology. At first it raised the hope that science would succeed not only in discovering the ^ As is very properly done in the classification of parasitic fnnffi where some species are given a higher rank and embrace a certain number of species of a lower rank. See for example Klebahn in Pringsheim's Jahrb. fiir wiss. Botanik, Vol. 34, p. 395. ^ A. Jordan, De rexistence d'especes vegctalcs aifincs, 1873, pp. 4-8. 168 The Origin of Species by Mutation. common origin of all species but in bringing the origin of species within the range of direct observation and even in placing in our hands a certain amount of control over these natural processes. But we are to-day just as far from this goal as we were in Darwin's time. The opponents of the theory of Descent have from the very beginning argued that we ought at least to be able to observe the origin of species and, perhaps, even to effect it experimentally. This crit- icism must even now be recognized as fully justified, although it is of course no longer one on the answer to which the validity of the doctrine of Descent depends. It is just at this point that the prevalent confusion over species becomes most evident. What shall we make the object of observation and experiment? Our oppo- nents answer : ''The origin of the ordinary Linnean spe- cies of the systematist." But these are artificial groups whose limits can be altered by the personal taste of any systematist and are indeed as a matter of fact much too often so altered. The origin of such a species, like that of a genus, is a historical occurrence and it can neither be repeated experimentally, nor can the whole process be observed. A plant-form can only attain the rank of a systematic species by producing a series of new forms and by the subsequent elimination of those which formerly related it to its parent form. It is obviously as impossible to observe the origin of an artificially circumscribed group like this as it would be to observe that of a genus or familv- The object of an experimental treatment of these phenomena must assuredly be to make the origin of the units which really exist in nature the subject of experi- species, Subspecies and Varieties. 169 ment and observation. We must deal not with the origin of the groups made by the systematist but with those which are presented by nature. There is no question that these elementary species often do arise in the garden and in agricultural practice. But in the first place they are only noticed when they have become established and when therefore the chance of observing the mode of their origin is irrevocably lost. And in the second place we smooth the matter over by calling the new forms "Varieties." What are varieties? In wild plants they are usually very different from what they are in cultivated ones. Or rather the term variety has a number of definitions none of which is definite enough. In the eyes of those who perhaps unconsciously were anxious to maintain the supernatural value of species — and there are many of them even now — all forms, not the result of crossing, the history of whose origin is more or less accurately known, are called "varieties," Thus, all elementary spe- cies arising under cultivation fall into this category. Gardeners as a rule often draw no distinction between "varieties" and "kinds" on the one hand and between these and species and hybrids on the other. The description of all forms with whose origin we are familiar, as varieties, opens the door to endless misuse of the term. On this ground alone therefore it ought to be given up. Even some of the best known authors of pre-Darwinian days thought that they could prove the common origin of a group of species by describing them as varieties of a species of a higher order. In this way Naudin for example, according to Wallace, "proved" that the thirty species of melons, which had been recog- 170 The Origin of Species by Mutation. nized up to that time, were only varieties ^ And it will obviously continue to be impossible to demonstrate the origin of a "species" so long as this demonstration is regarded as ''degrading" the form in question to the rank of a variety. This would become a mere juggling with words. The conception of a variety held by those who are the best qualified to judge, rests on the view that a single character is not sufficient to confer specific rank on a given form. A beautiful example is afforded by the case which we have already mentioned of Datura Stramonium and Datura Tatula. Each was regarded as a species by Linnaeus himself, but they have been united to form a single species by more recent authors on the ground that Tatula is only distinguished from Stramoniwn by the possession of a blue pigment in its flowers, stem and petioles.^ This limitation of the idea of a variety is manifestly desirable scientifically, especially for the reason that the distinguishing feature is very often due to the loss or latency of a character: absence of Petals, of Hairs, of Thorns, of Color in the flower and so forth. Such cases afford the best examples of what we ought to call a vari- ety. But it should not be forgotten that the evidence for the relationship of such forms to their species ordinarily rests only on analogy ; and not, or very rarely, on actual proof. Such varieties are just as distinct and just as constant in cultivation as the best species. If it is still considered proper that they should be called varieties, then it fol- ^ Wallace^ Darwinism, p. 87 ^ In my opinion, Siramoninm is regarded quite wrongly as the species and Tatula as the variety. Every analogy points to the blue as the older and the white as the younger form (See Fig. 5 on p. 31). species, Subspecies and Varieties. 171 lows that varieties are nothing less than a particular form of species. Varieties are only small species, as Darwin has said.^ Jordan's elementary species are distinguished from one another not by one peculiarity but in nearly all their characters. This is an extremely important point. There is absolutely no justification for regarding them as vari- eties. If we wish to treat them as subdivisions of the old species they must be called subspecies. I prefer to call them elementary species. Darwin speaks repeatedly of specific elements when he is referring to their indi- vidual characters.^ There is little prospect that an agreement between all the workers in this field will ever be brought about. Theoretically in my opinion we should be perfectly justi- fied in applying the coveted distinction of ''species" to these elementary forms, whose limits are not set by our imagination. But practically it is for many reasons more convenient to refer to the artificial groups of these, that is, the collective species, simply as species. Where we are concerned with the investigation of the origin of a single species we mean of course an elementary one. The other species are groups whose origin is a matter of history and cannot for this reason be dealt with ex- perimentally. Thus we see that Linnean species are collective and artificial whilst Jordan^'s species are single and real. Each collective species consists of a larger or smaller group of subspecies or elementary species; in the deter- '^ Life and Letters, II. p. T05. Darwin's more famous aphorism that varieties are incipient species is less happy. We know nothing about the age of most varieties. ^ E. g., Variations in Animals and Plants, IT. p. 23. Each of these elements is represented in the germ, according to the theory of Pangenesis, by a unit, the Pangene. 172 The Origin of Species by Mutation. mi nation of the limits of these groups the systematist is guided almost entirely by the gaps which have arisen by the disappearance of more or less numerous sub- species. With regard to the nomenclature, it would perhaps be better if the binary system were replaced by a ternary; to retain the Linnean specific names as much as possible and to write after them the name of the elementary form.i The idea of a variety should be strictly confined to cultivated forms. ^ § 22. SPECIES IN NATURE. The species of the systematist are compound species; they consist of a greater or smaller number of subspecies which breed true when tested. The larger the geograph- ical area inhabited by a species, the larger is the number of component subspecies : they are concentrated in the center of the area and become scattered towards its per- iphery. In local floras therefore as a rule each species con- sists of only one or very few elementary species.^ The species of such local floras do not exactly agree in neigh- boring districts.^ From France alone Jordan brought ^ This course is adopted by Waagen in Benecke, Geognostisch- paldontologische Beitragc , 1876, Vol. 2, p. 187. — An example: any one can guess the meaning of Draba verna leptophylla whilst Erophila leptophylla has no meaning except to the initiated. ^ Subspecies are not to be regarded as subsidiary to, nor as de- rived from the species ; for each species consists of a group of sub- species. The only thing that can be said in favor of the conventional assumption of a forma genuina is that it is convenient. ^ Only one elementary species of Draba verna so far as I can find occurs round Amsterdam and the towns in its neighborhood: it agrees with Jordan's D. leptophylla. ■* For example Senecio Jacohaea is common in the neighborhood of Haarlem, but always without ray-florets, whereas in the adjoin- ing dunes near Leiden it is only found with these florets. species in Nature. 173 together over 50 species of Draba verna^ in his garden,^ and from other countries in Europe, especially from Eng- land, Italy and Austria, about 150 more, so that in 1873 he had more than 200 forms in cultivation.^ This richness in forms, or polymorphism as it is called, of the so-called *'good" species is quite a general phenomenon.^ Darwin repeatedly called attention to it and argued that as a result of it the most widely dis- tributed types had the best chance of giving rise to new species and so of gradually becoming genera. •"* In the case of rare forms he showed the prospect of doing so to be much smaller. Very few plants are as rich in subspecies as Draba verna. Perhaps Viola tricolor comes next^ with its well-known subspecies Viola arvensis which is itself a collective form.'^ In Germany or France the average number of subspecies per species may be placed at 2 or 3, for the whole of Europe the average is perhaps about 10. If all these forms were noted and described the Flora of Europe would be increased tenfold, which would be most inconvenient. But just as there are valuable treatises which only deal with the genera or at any rate only with these and their more important species, so it would be the business of the ordinary Floras to describe the species and their more important subspecies. The task of deal- ^ See Fig. 3 on page 22. ^ Dc I'origine dcs arhrcs fniitiers, 1853. ^ Dcs cspcces vcgetales, aiRnes, p. 13, 1873. *It is often spoken of as "Variability": but this cannot conceal the fact that the elementary species which compose the species are constant, and independent of one another. ^ Wallace^ Darzuinism, p. 80 and 98. ® A. Jordan, Observations snr plusieurs plantcs nouvelles, 1846- 1849, Vol. II, p. 7. "^ See Fig. 4 on page 23. 174 The Origin of Species by Mutation. ing with all the elementary forms that exist must be the duty of monographs of a greater degree of completeness.^ When it is a question of the origin of one elementary species from another this material is absolutely essential to the student of evolution. When however our object is the study of the relationship of the larger groups it certainly constitutes a mere dead weight, the fact of whose existence one is only too often tempted to sup- press. But I can see no rea- son why these two branches of inquiry should not exist side by side. Nothing but a belief in the supernatural value of the Linnean species can stand in the way. In the natural state it iS only very rarely that ele- mentary species are distin- guished by a single or by two or three characters- (Fig. ?)2>) ; they usually dif- fer in all their organs and characters. A complete diagnosis often requires a whole page. The tout ensemble of the plant is quite distinctive ; and the practised eye can recognize the various forms at a distance.^ This is especially so in the case of cul- ^ Compare, for example, the Flora Europae of M. Gandoger, which gives the elementary forms for all the more important species, but only refers to their characters in short dichotomous tables {27 vol.). 'On the heaths near Amsterdam there are to be found three forms of Potcntilla Tormentilla, one with narrow, one with broad and one with intermediate petals ; I saved seeds from each of these forms and found them to be constant. ^ Or they may be quite or nearly indistinguishable externally and differ from one another only in fundamental physiological characters as for example in the choice of hosts in the case of the Rusts — facts Fig. 22)- Potentilla Tormentilla, with narrow, broad and inter- mediate petals representing three constant subspecies found in nature. species ill Nature. 175 tures where groups of many individuals of the different types grow close together. The characters are sometimes of such a kind that they are easily recognizable even on dried material; but they very often disappear entirely or partly when the plants are pressed. The constancy and thus the distinctness of the local species can only be proved by cultivating the plants from seed.-^ Experiments of this kind have been carried out on a large scale by Koch and Fries and other well- known systematists but especially by Jordan and his pupils. In many cases these experiments have been re- peated and always with the same result. Thuret and Bonnet grew 14 of Jordan^s species of Draba verna, 4-6 species of Papaver dnhiinn, for about 7 years and convinced themselves of the constancy of these forms.^ This statement is supported by the high authority of De Bary, who satisfied himself as to the constancy and systematic distinctness of the numerous subspecies of Draba verna,^ as the result of his well-known researches, which were continued and published after his death by F. Rosen. This splendid work has received full recog- nition, but it has not had the effect which De Bary evi- dently hoped it would have on his contemporaries, of directing research more generally into these channels. A similar state of affairs obtains in zoology. Every ?oologist knows, as Bateson remarks,^ that in the case which we owe to the exhaustive and important researches of Ericks- SON. ^ Conclusions based on comparative study only should never he regarded as proofs in this field. See the Flora Europae of Gandoger. ^J. CosTANTiN, Accomodation dcs plantcs, Bull, scientif. publie par GiARD, Vol. XXXI, p. 507. ^ F. Rosen, Systematische und biologische Beohachtungen iiber Erophila verna, Bot. Zeitung, 1889, No. 35. * W. Bateson, On Progress in the Study of Variation, Science Progress, Vols. I and II, 1897-98. Vol. II, p. i. 176 The Origin of Species by Mutation. of many species the individuals differ according to the region which they inhabit, and that by means of these differences the species can be spht up into local races. The differences may be very slight and often only vis- ible to the initiated, and yet perfectly (Constant. But these facts are far from being appreciated as much as they deserve.-^ § 23. SPECIES IN CULTIVATION. Just as wild species at present consist of a larger or smaller number of constant and independent subspecies, so presumably will it have been with those species which man has brought into cultivation. Pliny was acquainted with the different kinds of a number of fruit trees, for example 43 sorts of pears, 29 of apples, 10 of plums, 8 of cherries and so forth. The Romans knew at least two sorts of beet, several kinds of which grow wild in the Mediterranean region. In about the year 1600 Olivier de Serres described in his Theatre d' agriculture the cultivated plants that were known at that time. He refers also to the main types of our modern vegetables. He mentions 61 vari- eties of pears, and 51 of apples, and also the commonly grown sorts of beet. Whence all these forms arose we do not know. It is possible that they arose in cultiva- tion; it is even possible that they arose as the result of cultivation. But it is equally possible that they existed be- fore it, growing wild either together or in different places, and that all or most of them were taken over into culti- vation as such. For there is absolutely no ground for the belief that the plants known to agriculture were only ^ See also Duncker, Roux's Archiv, Vol. VIII, 1899, P- 164. species in Cultivation. \77 once found by man in nature and only once brouglit into cultivation. So long as the chief concern of biologists was to establish the theory of descent there was some use in elaborating the probabilities in this sphere. But now, it seems to me that it w^ill suffice if we recognize the lack of historical information on this point. .A favorite theme for discussion is the question whether wheat owes its origin to a few or to many wild forms. For whether we are to assume that wheat has "varied" in a large or small degree during its cultivation depends on the answer to this question. It seems far more likely that wheat, just like Draba verna, was orig- inally composed of a vast assemblage of subspecies in the wild state. ^ And as fertilization in wdieat takes place mainly before the flowers are open, it is evident that many kinds can maintain themselves side by side in the same field, provided of course that they are really con- stant. The history of this subject contains a chapter which has a very strong bearing on this point. It concerns Colonel Le Couteur's cultivations in Jersey at the be- ginning of the nineteenth century.- He was visited by Professor La Gasca who pointed out to him that his field of wheat, far from being a uniform culture, con- sisted of at least 23 distinct sorts growing together. The natural supposition was that some of these sorts would have a larger share in the harvest than others. Le Cou- TEUR therefore harvested the seeds of typical individuals of these sorts separately and carried out comparative ^ Of all cultivated plants the cereals have changed least accord- ing to De Candolle in I'Origine des cspcces ciiltivces. ^VoN RiJMKER. Gclreidesiichtung, p. 67. 178 The Origin of Species by Mutation. sowings of pure bred kinds for a space of a few years to find out which of them were the most valuable. The offspring of these sorts proved to be pure and constant; and his original field must therefore have contained simply a mixture of these sorts. Le Couteur continued to grow the best of the kinds thus purified with such success that he put them on the market with no small advantage to himself; even now some of them are still very well known, as for example the Bcllevne dc Tala- vera. Wheat was therefore at that time a mixture of dif- ferent sorts; Le Couteur seems to have been the very first to isolate these units. ^ And even now the common types of wheat are still mixtures. The mixture main- tains itself without artificial selection, but the pure form does not.^ Later, Patrick Shirreff in Scotland worked on the same lines as Le Couteur w^ith various forms of cereals. He used to look in his own fields and in those of his friends for striking and apparently better examples : then he sowed their seeds separately and examined their off- spring. As a rule they turned out to be constant and often very productive. In this way he found the original of Mungo swells wheat in 1819, Hopetozvn oats in 1824, Hopetozvn wheat in 1832, and later Shirreff' s oafs.^ They were absolutely constant and as soon as a sufficient quantity of seed had been obtained by cultivation for two ^ At that time nobody thought of improvement : the idea did not arise till about 50 years later. ^ See p. 98. ' V. RiJMKER, loc. cit., p. 70. See also the account of Dr. Hesse's travels in Landw. Jahrb., VI, 1877, p. 850 et seq., and Shirreff's Improvement of Cereals, London, 1873. species in Cultivation. 179 or three generations they were put on the market without further selection. Space does not permit us to treat further of Le Cou- teur's and Patrick Shirreff's work. Suffice it to say that they show us in a general way that wheat, barley and oats^ were at that time mixtures of perfectly con- stant subspecies exactly as we have seen that the species of wild plants are. ' But we know as little about their origin in the one case as we do in the other. One of the most frequently discussed questions in practical horticulture is that of the origin of fruit trees, especially of the modern improved kinds of apples and pears. There is no doubt about the common origin of these forms. The question is only whether their common origin merely follows from the theory of descent or whether it is historically traceable. The latter is cer- tainly not the case with most of the chief types ; the past history is only known with certainty in the case of some of the recent sorts. It is to the Belgian breeder Van Mons that we owe the most valuable information on this subject that we possess. In the first half of the nineteenth century he put many of our well-known kinds on the market.- ^ Rye, which may be wind- fertilized, behaves differently. ^The literature on this subject seems to be little known and is difficult to get hold of : I have not succeeded in seeing the works of PoiTEAU and Chandeze. The following is a list of the most im- portant : Van Mons, Arbrcs fruifiers ou Pomonomic beige, 2 vols, 1835. Quotations from it will be found in Jordan's Arbres fruitiers, pp. 38 and 94.. PoiTEAU, Theorie de Van Mons ou notice historiquc sur les moyens qu'emploie Van Mons pour obtenir d'cxccUents fruits de semis. Ann. Soc. d'Agric, Paris, 1834, Vol. 15. G. Chandeze, La Theorie de Van Mons concernant la production de nouvelles varietes fruitier es. Belgique horticole, 1877, p. 354- Bot. Jahrb., V, p. 761. GoDRON, De VEspcce, II, p. loi. 180 The Origin of Species by Mutation. Van Mons expressly stated that he himself had not originated any new forms: "La nature seule cree.''^ He found all the sorts which he cultivated and put on the market, growing as such in the wild state^ and, as it happened, almost all of them in the Ardennes. The wild plants were thorny and their fruits small, tough and woody. As the result of being sown in a garden and under the influence of another climate^ they regularly lose their thorns and the tough consistency of their fruits, which become larger, fleshier and juicier. But the dif- ferences in form, color and taste and other valuable char- acters arose neither in, nor as a result of, cultivation ; they already existed in the wild forms. His new kinds are nothing more nor less than already well-known cultivated forms'* which he has improved in respect of size and juiciness, by selection for two or three generations'* without altering their varietal characters in the very least. ^ Van Mons was fully aware of the independence and constancy of these forms and it should be noted that he speaks of them as subspecies and not as varieties. The best way to raise a new type for the market is not to sow the seeds of the best sorts already in culti- vation but those of a fruit which, be it ever so puny, belongs to a hitherto unknown type. It seems that most of the new sorts that have been raised by other breeders have arisen in the same way. For example the splendid St. Germain pear owes its ori- gin to a single tree found by chance in the Foret de St. Germain near Paris; Besy de Charnnontel, Bergamotte Sylvanche, and Virgoideuse are also due to a lucky find. ^ Pomonomie, I, p. 445, *Loc. cit., II, p. 208. ^Loc. cit., p. 406, 444. ^Loc. cit., p. 462 and II, p. 208. ^ Loc. cit., p. 410. "" Loc. cit., T, p. 415. species in Cultivation. 181 Bailey^ has recently given a very striking example w^hich illustrates this point. Mr. Peter M. Gideon sowed a vast number of apple seeds and from these he got a single plant whose fruit he ultimately put on the market as the Wealthy Apple, because he made his fortune by it. This apple is now one of the most favorite and widely known in Minnesota. Mr. Gideon tells the story of how he got this mag- nificent fruit as follows. For nine years he sowed apple seeds so as to raise about a thousand young trees every year. But all this led to no result. Then he happened to buy a small basket of apples of a foreign kind in Maine : they provided him with about 50 seeds from one of which his Wealthy Apple arose. Sowing on a large scale had no result ; sowing on a small scale but from a new form fulfilled his highest expectations. Our argument is supported by the following evidence. If apples and pears are allowed to grow wild they are well known to revert to the type of the crab-apple and the wild pear in a few generations. But each sort retains the features characteristic of it ; they do not all revert to one and the same wild form. Whence does the host of wild sorts of apples and pears arise? We do not know. There are some who assert that they have arisen in cultivation and have then run wild. But this would hardly account for the large number of new sorts that have been obtained. It is the same with most cultivated plants as it is with cereals and fruit trees : almost every species consists of more or less numerous subspecies about whose origin we know nothing at all. Flax, the red clover and the poppy are very good ^ L. H. Bailey, Plant-breeding, New York, 1896. p. 108. 182 The Origin of Species by Mutation. examples of plants with such subspecies. The chief types of Chrysantheuimn indicum were imported as such from Japan into Europe ; the newer sorts have almost all been obtained by crossing them. A great variety of other examples can be easily collected. Fig. 34. Scduui crispiim after ]Munting, 1671. Many so-called \'arieties and even many monstrosi- ties have been known since the time when the species to '^Abraham Munting, Waarc Oeffeninge der Planten, 1671, p. 237. Hunting's Sedum crispuiii evidently is the same as Sediim cristatum Schrad. {Sedum reiiexum cristatum) ; the monstrosity must therefore be more than two centuries old. Since Hunting's time fasciation in this species has repeatedly been observed and recorded. Cf. Penzig, Teratologie, I, p. 467. The character is strongly inherited. I raised from seed a square-meter bed full of plants with more or less flattened branches, some of which I have photographed and reproduced in Fig. 35. Normal cylindrical or atavistic branches are shown both in the above figure from Hunting and in that from my own culture (Fig. 35 at). species in Cultivation. 183 which they belong were introduced; and have been de- scribed and drawn in early works on the subject. Abra- ham Hunting gave a long list of them in the year 1671.^ In it will be found examples of double flowers of Vinca, Colchicum, Hepatica, Cardaniine, Cheiranthiis Cheiri, Papaver, Viola, Caltha, Althaea, and others; of white- flowered forms of Ononis, Syringa, Centaiirea, Digitalis, Fritillaria, Hepatica besides white strawberries, white Fig. 35. Scdum rcUexum crisfatum. From nature, 1900, with expanded and ordinary (at) branches. raspberries and red gooseberries, and double Bellis and Matricaria. Also proliferating forms of Bellis, Calen- dula, Heliantlms and Scabiosa, fasciated Crown Impe- rials, Plantago major rosea. Primula veris, and P. Auri- cula with a double Corolla, fasciated Seduni ( Figs. 34 and 35), Celosia cristata, Amaranthus cristatus, etc. Moreover hundreds of varieties of the more impor- tant garden plants, e. g.. Hyacinths, Tulips and Ranun- culus were known at that time. ^ Waarc Ocffeninge der Planten, Groningcn. 1671. 184 The Origin of Species by Mutation. Many forms which are put on the market as new are, from our point of view, really quite old. I mention as an example the famous double Lilacs which Victor Lemoine of Nancy put on the market in the '80's. They consist of a number of new^ and in many respects excellent varieties which have now found a place in many gardens and parks. Thev were offered as new ; and I was anxious to find out how the ^'doubling" had been attained. I went, therefore, in 1892 to Nancy and asked M. Lemoine. After he had shown me his plantations of Lilac he told me the follow- ing story of their origin. "In 1870 I happened to see in a garden in Luxembourg, a double specimen of Syringa vulgaris azure a plena, a little-known form which is seldom seen in gardens. When some years later I came to think of growing Lilacs I simply bought this plant and crossed it with almost every variety on the market." This was the way in which he got his novelties. But as to the origin of doubling he was completely in the dark. Later I found that Hunting had mentioned the double form as early as 1671. We know just as little about the origin of the Cactus- dahlias which threaten to supersede all other kinds owing to their great variety, and to the splendor of their flow- ers. They are the result of a cross between one single plant and numerous older varieties. When I visited Mr. Van den Berg in Jutphaas who introduced this novelty, he gave the following account : "Many years back (1872) I asked a correspondent of mine in Mexico to send me a case of bulbs, roots or rhizomes of any kind of foreign plants he could possibly get hold of. The contents of the box reached Holland in very bad condition : almost everything was rotten; in fact everything but a single tuber which however produced a shoot. This plant was species mid Specific Characters. 185 the first C actus-T)3h\\2i. All efforts to find the same form, in the district where my correspondent lived, were in vain."^ The plant was there; but how it arose we do not know. It is just the same in many other cases, and with the most important types too. The breeder is delighted when he sees a new form ; but as to how it arises he is generally ignorant. It often happens that they arise singly in sowings on an enormous scale ; in which there is a greater likelihood that the seeds will be of different stocks, than in small ones. In this way D. B. Wier got his cutleaved maple in a sowing of about a milHon seedlings.^ And in the same way Donkelaar got the first double Dahlias in a culture of about 10,000 plants, and so forth. There is no object in citing more instances especially as most of the early ones are to be found in Darwin^s works. We may conclude therefore that even among culti- vated plants, species are mixtures ; consisting, as they do, of independent often numerous sorts of subspecies which have been found as such in the wild state. This fact is well known to many breeders and botanists : though the earlier botanists were more familiar with it than modern ones are. Hence the often repeated saying,^ 'Tf you want to raise a novelty you must first possess it!" § 24. SPECIES AND SPECIFIC CHARACTERS. The reader is now in a position to understand what I mean when I say that our business is not really with ^ See Van den Berg, in Gardeners' Chronicle, Nov. 8, 1879; and W. Miller, The Dahlia, in Bull. Ithaca, No. 128, p. 127. ^L. H. Bailey, Plant-breeding, 1896, p. 109. ^ Jordan, Arhres, fruitiers, p. 96. 186 The Origin of Species by Mutation. the origin of species but with the development of specific characters. The diversity of organic forms is due to the existence of a vast number of differentiating characters. And the question we have to answer is ''how have these characters arisen?" Subspecies become species by extinction of inter- mediate forms. New species can arise by crossing when the pecuHarities of two forms already existing are united to form a single new one ; and so on. But these are not cases of the origin of specific characters. Many species and even genera and still larger systematic groups have arisen by these characters disappearing or becoming la- tent. The origin of the monocotyledons from the dicoty- ledons is regarded by some as coming under this head (Delpino). But loss and latency are obviously special cases which do not directly touch the main question of progress in the animal and vegetable kingdom. The question is not how m.any characters peculiar to itself must an animal or plant possess to justify its ele- vation to specific rank, but : how have these characters arisen, or how can they arise ?^ In other w^ords : the mutation and the actual process of mutating must become the object of investigation. And if we once discover the nature of this process, not only will our insight into the actual relationship of living organisms become much deeper, but w^e may even hope that we may be able to gain some measure of control over the formation of species. If the breeder has obtained control over variability why should he not obtain it over mutation as well ? ^ "These factors are the units with which the science of heredity has to deal." Intracell. Pangenesis, p. 9. For their association in groups see pp. 21-22 and ZZ- Mutations in Cultivation. 187 It is clear that we can only advance by very small steps dealing at each step v^ith a single mutation. But even single mutations may be of enormous importance in horticulture or agriculture. Much that now seems unattainable may come within our power if only we can obtain some insight into the fundamental principles in- volved in mutation. There lies here a wide field of work the results of which will be as important to the biologist as to the practical man. § 25. MUTATIONS IN CULTIVATION. In a preceding section {% 23) I endeavored to show that man}^ of the elementary species which exist in a state of cultivation had arisen before they were intro- duced into it. But it does not follow that this is always the case as Jordan, Kerner and others believe. On the contrary in many cases there is historical evidence which at least makes it highly probable that mutations occur in the garden and the field no less than in the wild state. But it usually happens that the new form is not seen until it is alreadv established; how, when and where it arose cannot be discovered, or at most only with a small degree of certainty. According to the theory of selection the origin of a new form is a gradual process which we can observe whilst it is taking place. But the evidence at our dis- posal does not support this theory. It is true that forms wdiich have arisen suddenly exhibit a high degree of fluctuating variability and so give the selector the oppor- tunity of intensifying the new character. But that is a very different matter from the gradual origin of the new character. 188 The Origin of Species by Mutation. Good examples of mutations can be found in agri- cultural and horticultural literature. But before I give a selection of them I must point out how clearly the dis- tinction between races and subspecies is appreciated by practical authors. Prof. Kurt von Rumker in his often quoted Introduction to the Breeding of Cereals divides his treatment of methodical selection into two parts. One of them deals with selection w^ith a view to improve- ment, the other with selection with a view to the origin of new forms. ^ The object of the former, he says, is to fix characters already present, to stamp them so to speak, and to intensify desirable qualities. New forms, however, arise when the changes "do not consist merely in continuous improvement along one line but in the production of new qualities as lateral off- shoots." Such changes occur now and then in our fields and are known as spontaneous variations. ''Nothing is yet known with certainty about the origin of such spon- taneous variation and still less about the causes of their origin." All that we know is that they are inherited. After these quotations from Von Rumker the com- mon phrase "the production of new forms" will sound, to say the least, exaggerated : we should be nearer the mark if we spoke of the search for new forms (and of their subsequent improvement, in the usual sense of the term). The awnless form of Beseler^s Anderbecker Oats is a very famous example of a form which was found ready to hand in the fields. I propose to give now a series of further examples. In almost all the cases the new sorts have come absolutely true to seed from the very beginning when the possibility * P. XIV, and 56 and 83. Mutations in Cultivation. 189 of crossing has been rigidly excluded. Sometimes the new character appears very slightly developed in the first instance as in the case of ''double" flowers. In such cases the characters have to be improved by selection. In some the variation appears once and for all, in others it continually reappears. It is well known that every breeder should look anxiously for possible novelties ; but when he has found one, it depends on him and on him alone whether it attains its full beauty. The origin of the new form is emphatically due to chance and not to the skill of the breeder, as it is in the improvement of races. Chelidonium laciniatum Miller, a subspecies of Cheli- doniuni ma jus, is one of the most beautiful examples because more is known about its origin than about that of almost any other plant, thanks to the painstaking in- quiries of E. RozE.i j^g gives the following history of it. About the year 1590, Sprenger, an apothecary in Heidelberg, found in the garden where he grew plants for his business (amongst which was Chel. majns), a new form of Chelidonium which dififered from C. majus in the possession of deeply cut leaves and petals. He called it Chelidonia major foliis et floribns incisis and sent some examples to Jean Bauhin, Gaspard Bauhin. Clusius, Plater and other well-known botanists of his time. All of them declared that the plant was unknown to them and new. It had never been found wild before, nor has it been found since ; although from time to time it has escaped from gardens. It comes absolutely true from seed, has maintained itself till the present day and is very generally grown in Botanical gardens. Miller, ^ E. RozE, Le "Chelidonium laciniatum" Miller, Journal de Bo- tanique, 1895, Nos. 16-18. 190 The Origin of Species by Mutation. RozE and many others have tested its constancy by cul- tures extending over many years and have observed no reversion to C. majus. I have repeated the experiments with the same result. We may conclude therefore that C. laciniatum arose about the year 1590. Unfortunately Sprenger does not say v^hence the seeds came which gave rise to it ; whether Fig. 36. Chclidonium laciniatum. A flower of it to the left. Below a flower of C. majus. from seed saved by himself from C. majus or from some other source. The former is the more probable since otherwise he would have known from whence he had obtained it. Transitions between the two species in question do not occur to-day any more than they did in Sprenger^s Mutations in Cultivation. 191 time. We may presume therefore that the younger form arose suddenly from the older one. W. T. Thiselton Dyer has described a series of spontaneous variations of Cyclamen latifoliiim, a very interesting species from the fact that it is one of the very few garden plants with which crossing had not yet succeeded.^ The supposition of a hybrid origin of its sub- species is therefore excluded. A form with horizontally projecting petals and an- other with hairy struc- tures in its flowers, re- minding one of similar structures in the flower of the Iris, have been de- scribed. The first form has arisen manv times ; it was at first thrown away as unsuitable for cultiva- tion, but has since been put on the market. The incised petals also have arisen several times, for example in 1827, when they were described in the Botanical Register, but were subsequently lost. Since 1850 they have appeared ni several nurser}- gar- dens. The hairy structures suddenly appeared in 1890 in the nursery of Messrs. Hugh Low & Co., although in a veiy rudimentary form. They were greatly improved by repeated selection, and after a few years put on the Fig. 38. Chelidonium ma jus. ^W. T. Thiselton Dyer. The Cultured Evolution of Cyclamen Latifolium. Proceed. Roy. Soc, Vol. LXI, No. 371, p. i35- 192 The Origin of Species by Mutation. market. They also appeared in France as early as 1885; but there they were not cultivated further. They exist both in the red and in the white variety. Strawberries without runners belong to the species Fragavia alpina and are known under the name of Gail- i^ON-strawbcrncs.^ Forms are known both with red and with white fruits.^ The history of their origin is re- corded by P. P. A. De Vilmorin in the Bon Jardinier.^ He found a single individual bearing this character in a crop of the ordinary Fragaria alpina. The seeds of this individual gave rise solely to plants without runners : the new sort was absolutely constant from the beginning. The cauliflower and Kohl-Rabi were raised from iso- lated monstrosities of Brassica olcracca.^ The Chou de Milan dcs Vert us likewise arose spontaneously from an-' other sort of cabbage and soon became one of the most popular vegetables in the Paris market.^ Merctirialis annua laciniata was discovered in 1719 by Marchant as a new form ; since that time it has come true from seed.^ That is the last of these examples I shall refer to. Some species have appeared twice, or even more often, in localities widely distant from one another and under circumstances which almost completely exclude the ])ossibility of a common origin. I may quote the example of the copper beech, to which Prof. J. Jaggi has devoted an exhaustive historical monograph."^ Three localities ^ See Fig. y on page 34. ■" Vilmorin Andrieux et Cie., Les plantes potageres, p. 222. ^L. De Vilmorin,, L' amelioration des plantes par le semis, 2d ed., p. 48. *A. P. De Candolle, Transact, hortic. Soc, 5, p. i, quoted in Hofmeister, Allgemeine Morpliologie, p. 565. " Vilmorin, L'amelioration, he. eit., p. ig. ^GoDRox, De I'Espeee. 1, p. 160. • J. Jaggi, Die Blutbuche su Biich am Irchcl, Zurich, 1893. Mutatiuns in Cultivation 193 for it are known. The Stammberg near Buch am I rebel in tbe Zurich Canton; a wood near Sondersbausen in Thuringia;^ a wood above Castellano near Roveredo in the southern Tyrol The first locabty was known as early as the 17tb century; the second in tbe second half of the 18th century; the third only at the beginning of the 19th. In the same way Fragaria monophylla (Fig. 38) was found by Fries in the neighborhood of Skaru- gata in Lapland ; then it arose in a garden near Versailles about 1761 and is now to be found in many botanical Fig. 38. Fragaria vesca monophylla. a, two leaves; /', a young plant on a runner with single, double and triple leaves — a case of atavism. gardens.^ Fagus sylvatica aspleniifolia was found in a wood in Lippe-Detmold and in the neighborhood of Paris.^ Alniis glutinosa laciniafa (Fig. 39) and Bctula alba laciniata are found wild in Sweden and Lapland.^ * l\Ir. ^., DoRiNG in Sondershaiisen informed me, that this tree is still living; its foliage as well as that of its offspring is, however, only of a pale red. (Note of 1908). "Braun, Abh. k. Acad., Berlin, 1859, p. 113; Hofmeister, Allg- Morphologie, p. 557 and 571; Bot. Zeitung, 1878, p. 283; Alph. de Candolle, Geographie botanique, II, p. 1081. ' E. Faivre, L'espece, p. 44; the former statement is from Braun, Verjilngung. * Braun^ loc. cit., p. 332. 194 The Origin of Species by Mutation. In the nursery gardens the same novehy often appears simultaneously in different places; as for example Age- ratiim mexicaniim naniim luteum which arose about 1892 in both Paris and Erfurt.^ There is a series of varieties on the market, of the most diverse botanical species, of which it can be said that it would be practically impossible for them to grow wild. They have often been brought forward as evidence Fig. 39. a, Alnus gliitinosa lacl- niata with fruits ; b, leaf of Alnus slutinosa. Fig. 40. Rammculus acris pcia- lomana, a form which has be- come completely sterile by profuse petal formation. From a plant found in a meadow. for the view that varieties arise suddenly in cultivation by so-called spontaneous variation or mutation. I recall those fruits which cannot dehisce as Papaver somni- feruni inapertmn and Linum usitatissinium (L. crepitans is the only subspecies which open its fruits so as to scatter the seeds). Then there are the large and heavy seeds of cereals and some Legwninosae but especially of maize ^ I was told this by Mr. Otto Putz., a nurseryman in Erfurt. Mutations in CultivatioH. 195 whose seeds seem to have no means of becoming distributed. Lastly there are the sterile varieties; Currants (Corin- thian grapes), Bananas, many sorts of apples and pears, astrakhan grapes, some strawberries, the green rose, the green Pelargonium sonale and green Dahlias (of which I have cultivated two different sorts, one with elongated and the other with ordinary flat flower- heads). Ranunciihis acris and Caltlia palustris which have become sterile by petalomany (Fig. 40) and many other examples of this kind of doubling;^ then there is the sterile Maize (Fig. 41) many examples of which have ap- peared in my own cultures but which, so far as I know, does not seem to have been noticed elsewhere.^ The great majority of forms which have arisen suddenly, be they varieties or subspecies, come absolutely true from seed; that is to say every single seed gathered reproduces the new form when sown, provided that the seed pa- rent was fertilized w^ith its own pollen, or with pollen from another example of the same form. Constancy is one of the properties of elementary species. ^ K. GoEBEL. Pringsheim's Jahrhiichcr fiir wissensch. Bot., Vol. XVII, p. 207. ^ Over stericle Maisplanten, Botan. Jaarboek 1889, Table V, p. 141. Steriele Mais als erfelyk 1890, p. 109. Fig. 41. Zca Mays sterilis. Three un- branched "pan- icles." a, without bracts ; b and c. with slight bract formation at the tip. Dodonaea, Vol. I, ras, ibid.. Vol. y, 196 The Origin of Species by Mutation. Apparent exceptions to this rule are so numerous that we might be inchned to doubt its universal validity. But in most cases it will be found that those who record such exceptions have paid no regard to the possibility of cross- fertilization by insects or by the wind. Crossing is cer- tainly the simplest and most obvious explanation of them. The whole subject of so-called atavism in plants demands a careful re-investigation, for most of what passes as atavism in the nursery and private gardens is nothing more nor less than the result of accidental crossing. At least so my researches into these phenomena lead me to believe. I shall however return to this subject and deal with it more thoroughly in a later section ; and shall confine myself now to citing some of the more important in- stances of constancy. The complete constancy of many varieties is well known. As for example in the case of Matricaria Cham- omilla discoidea and the corresponding varieties of Bi- dens tripartita and Senecio Jacohaea. Also of Datura tatula ineruiis/ of Ranuncidus arvensis iuermis,^ of the peloric varieties of Antirrhinum majus,^ of Nigella sa- tiva apetala,^ of Ilex Aqtd folium with yellow berries,'* of weeping oaks and weeping birches,^ of red-leaved Bcrheris,'^ of the peloric form of Corydalis solida,^ of Hordeum trifurcatum, Rubus fruticosus laciniafus be- sides countless garden plants and vegetables (sugar peas, thornless spinach and so forth). ^ Botan. Zeitung, 1873, p. 687. * Masters, Vegetable Teratology, p. 227. ^Hoffmann, Botan. Zeitung, 1881, p. 410; a number of other examples are recorded here. * Darwin, Variation in Animals and Plants, II, pp. 24, 26. ^GoDRON, Mem. Acad. Stanislas, t868, p. 3. Mutations in Cultivation. 197 I have already said that the so-called cases of atavism, brought forward as evidence against this constancy, are really cases of crossing. The copper beech illustrates this well. Its distinguishing character is reported as be- ing inherited to a highly variable extent, according to the locality in which it lives. Sometimes all the seeds come true; sometimes only 20%. But as the trees in question grow amongst ordinary beeches, and as arti- ficial fertilization is of course out of the question, they must usually be fertilized by pollen from the surround- ing trees. If we want to draw any conclusions from the posterity of a copper beech w^e must confine our atten- tion to properly isolated trees. In conclusion we may refer to the familiar fact that in cultivation mutations follow on one another so that the plant gradually becomes separated from the original form by an increasing number of characters; which is exactly what, in all probability, occurs in nature. The great number of long names of garden plants is evidence of this ; as for example Scabiosa atropiirpurea nana pur- purea from which a Forma carnca and a Forma rosea have subsequently arisen ; Calliopsis tinctoria pumila pur- purea, Tagetes potula nana with dark leaves, and another form of this dwarf with bright yellow flowers and so forth. The succession of names often indicates the stages of development of the form in their historical sequence. Finally, then, we may say that a gradual origin of elementary species has not yet been observed ; but that there are hosts of instances in which new ''species" have arisen suddenly or in which at least such an origin is in the ver}^ highest degree probable. Scarcely ever has the new form been isolated immediately it appeared : it is usually left like its parents to pollination by insects. So 198 The Origin of Species by Mutation. far as this circumstance allows us to judge, these new species are as a rule just as constant as the older so-called "good" species. § 26. THE HYPOTHESIS OF INDISCRIMINATE MUTABILITY. The chief merit of Darwin's theory of selection was that it explained the adaptation which is seen on all hands in organic nature on purely natural principles and without the aid of any teleological conception. It is be- cause it does this so completely that the theory of descent has gained such universal acceptance. The universal belief in the kinship of living forms, in its turn now makes the experimental study of the manner in which one species arises from another, possible. Nay, it chal- lenges us to such an inquiry. How the species which exist at the present time arose in the past is evidently a historical question which can only be directly answered in a very few cases. But the determination of the mode of origin of species must soon become the subject of inquiry just like any other physiological process. According to the Darwinian principle, species-form- ing variability^ — mutability — does not take place in defi- nite directions. According to that theory, deviations take place in almost every direction without preference for any particular one, and especially without preference for that direction along which differentiation happens to be proceeding. Every hypothesis which differs from Dar- win's in this respect must be rejected as teleological and unscientific. The struggle for existence chooses from among the mutations at its disposal those which are the best adapted ^ Intracellular e Pangenesis, pp. yZy 210, etc. The Hypothesis of Indiscriminate Mutability. 199 at the moment; in this way alone can their survival be explained. According to Wallace's and his followers' modi- fication of the theory of selection that process concerns the individuals of one and the same species only. Accord- ing to the theory of mutation, however, the units with which selection deals are the species themselves. Some survive and extend the limits of their distribution ; others are wiped out ; the former may again be the source of new species, the latter vanish and leave no posterity. The essential idea of this theory may be expressed by saying that by natural selection species are not created but elimi- nated. Wallace's theory of selection and the theory of mutation — specializations in two different directions of Darwin's theory — both have to account for evolution without calling in the aid of a theory of variation in a definite direction. Wallace's theory obviously does this inasmuch as according to it the material on which selec- tion operates is individual variability ; moreover the study of this variability rewards the student with a rich harvest of facts which might afford a strong support for this theory were there not other reasons for rejecting it. The theory of mutation is in this respect less for- tunate. For mutations themselves can only be directly observed in a very few cases ; and in fewer still have they been properly studied. Mutations are naturally much rarer than individual variations, which every animal and plant exhibits ; they do not lend themselves to investiga- tion in the same way as the latter. Nevertheless they can be made the subject of research, and for many reasons they ought to be investigated just as minutely as varia- tions have been. 200 The Origin of Species by Mutation. One of the greatest faults of those who hold the cur- rent theories of selection is that they have focussed their attention much too exclusively on the phenomena of se- lection and individual variation and much too little on mutations. There can be no doubt th^it this is one of the chief causes of the depth of our ignorance of the facts of mutation. This circumstance explains how it is that we can do no more in the matter of testing the hypothesis of indis- criminate mutal^ility by the facts at our disposal, than find out how far the special hypotheses put forward by various authors are in harmony with fundamental and undisputed Darwinian principles. Nor is this task an easy one. The question is ob- viously : what share in the origin of the larger or collec- tive species is to be ascribed to mutability and what to the natural elimination of elementary species. Many authors have suggested that altered conditions of life exert a direct influence on animals and plants in such a way that new characters are developed which render their possessors better fitted to their new environment. The environment has, according to them, the power of directly evoking in the organism an adaptive response. But this assumption seems to be no more than a begging of the question we are trying to answer. Dar- w^in's idea was that mutability took place in all directions and that the most favorable mutations were preserved. And this view of the matter will, it seems to me, remain the simplest and most probable answer to the question imtil such time as we have collected sufficient experi- mental evidence to decide whether this mutability exists or not. We must now discuss in some detail the views of W. The Hypothesis of Indiscriminate Mutability. 201 B. Scott, one of the most prominent champions of the theory of mutation, who, however, has declared against the hypothesis of indiscriminate mutability on paleonto- logical grounds. For it seems to me that this hypothesis agrees perfectly well with the facts of paleontology and especially with those wonderful genealogical series which have lately been discovered. Unfavorable species may well have arisen in far greater numbers than we should ever imagine, without having left the shghtest trace in geological strata. The continuous series certainly point to selection in a constant direction during long periods of time, but by no means do they in my opinion demand a theory of mutability in a definite direction ; for their explanation. A closer examination of Scott's arguments will show how far my view is justified. Scott asserts that those paleontological series which are well knowni, are con- tinuous and without gaps ; whereas those in which the gaps are many are just those which are imperfectly known. This incompleteness is due either to the absence of individual strata from certain periods or to the fact that it has not yet been possible to examine the strata in question, properly. But where the series of strata is continuous and without gaps, and their examination thor- ough, the genealogical tree has also proved to be contin- uous and without gaps. This is evident in the case of the well-known pedigree of the horse, in those of many other mammals, of Ammonites and so forth. Such series always possess this remarkable feature : they proceed, so to speak, in a straight line. Evolution makes straight for its goal without deviation, swerving 202 The Origin of Species by Mutation. neither to the right nor to the left so as to form a zig-zag line.-^ The question we have to answer now is how can such a definite and apparently predetermined series of changes be explained by natural principles and especially by the principles of descent enunciated by Darwin ; in other words : how must we picture mutability and natural se- lection to ourselves in order to gain a satisfactory expla- nation of these series. Two ways of explaining them are possible. 1. iMutability may take place in almost all directions; and it is natural selection which operates in one direction during long geological periods. 2. Mutability takes place only in one direction and itself determines the direction of change. The former obviously represents the view of Dar- win ; the latter that of Scott. In the first place it must be remembered that when we are dealing with paleontological facts it is hardly possible to decide between mutability and selection, and, as Scott has remarked, no ''explanation" can ever be much more than a guess. There is every reason for supposing that in the gen- ealogy of every organism numlDers of species may have arisen but have never multiplied sufficiently to insure their preservation in the rocks, and have disappeared without leaving behind them either posterity or record. Paleontology can obviously not help us to decide as to the admissibility of such a theory. Let us therefore com- ^ Weldon regards this objection to the theory of selection the most serious of all. See his Presidential Address : On the Three Principal Objections which Arc Urged Against the Theory of Nat- ural Selection, 8th Sept., 1898. Brit." Ass. Adv. Science, Bristol, 1899, p. 887. The Hypothesis of Iiidiscriininate Mutability. 203 pare the number of species in these geological series with the wealth of our modern collective species in elementary types. Can there be any question that this richness ex- isted at all times in spite of the fact that there is no geo- logical record of it?. We will again refer to the composite species Draba verna, which has been so fully elucidated by Jordan. Thuret, De Bary, Rosen and others. It is generally assumed that all the elementary species of Draha vcrna spring from a single original form ; yet they differ from one another in every conceivable direction. They must liave arisen as mutations from this form ; which must therefore have produced them in all directions and not in one particular one. They afford sufficient material for natural selection ; whatever view of it we hold. Supposing that the ancestors of the horse exhibited a similar indiscriminate mutability, what chance would there be of their preservation in the fossil state? This question is a difficult one to answer and calls for further treatment. The present number of elementary types be- longing to a species is no measure of the number of mutations it may have produced since its origin. By far the greater number of mutations presumably perish, nipped in the bud by natural selection. Other forms may continue for one or two years, but after a time, they too disappear. It is only a very few which ultimately come to take part in the great struggle for existence. Much that appears, must forthwith disappear. Even between the male and female individuals of one and the same species there is often a strong competition whicli may result in a permanent alteration of their numerical proportions. As a rule male plants are more delicate, and we find quite regularly that in unfavoral)le positions 204 The Origin of Species by Mutation. the females have increased in number in proportion to the males. This was observed by Hoffmann on Spi- nacia, Rvunex and LycJinis, and by others on many other species. In Matthiola incana the strongest seeds give double-flowered individuals but the proportions of such depend on the conditions in which they are cultivated ; when seeds are gathered in the open they do not exceed 50%, when sown in puts they attain 60 and sometimes 70%. It seems to me therefore to be a warrantable assump- tion that in geological times many newly arisen forms were promptly anniliilated and have left no trace. If the hypothesis of mutability in one direction ren- ders the theory of a selection operating in a constant direction superfluous, then we must regard mutations as in a very high degree limited. Only those species, whose remains have been found in paleontological strata could have arisen by it ; and strictly speaking only those wliich lav in the direct line of descent. All lateral branches which have died without posterity point to a selection operating continually in the direction of the main line of descent. It seems to me that the more we consider Scott's view in detail the more do the differ- ences between him and Darwin tend to disappear. The question how far the theory that selection may have operated in one direction during long periods of time is justified lies outside the scope of this book : but it will be admitted first, that it has never been proved to be false, and secondly, that this theory has at least as much justification as that of mutability in one direction. In short: The mutation theory demands that organ- isms should exhibit mutability in almost all directions. The facts of paleontology and classification are in accord The Hypothesis of Periodic Mutability. 205 with this theory. And the fact that ordinary or collective species consist of groups of elementary species whose characters may differ in every conceivable zvay empha- sises the existence of indiscriminate mutability. § 27. THE HYPOTHESIS OF PERIODIC MUTABILITY. Tlie constancy of species is a demonstrated fact ; their transmutabihty is still a matter of theory. This is the old objection against the theory of descent. La- marck, Darwin and Wallace met this difficulty by assuming that the immutability was only apparent and was due to the fact that the changes are so slow that in the short time during which we are able to observe them they can not be detected. This however is merely an assumption, as I have al- ready shown. No one doubts that many species have undergone vast changes during the course of centuries ; but no one knows whether they have taken place grad- ually or by leaps and bounds. The contrary supposition that species can remain ab- solutely unchanged during long periods of time but under certain circumstances begin to produce new forms seems to me at least equally justified. The ancestors of species that exist to-day have on this theory passed through im- mutal)le and mutable periods : the division of the large species into elementary species would be the result of the last or of some of the last periods of mutability.^ We repeatedly find the idea of a periodic transmuta- tion of species expressed in Darw^in's works. "Changed ^ KoLLMANN remarks on this subject : 'Tn no species of animal or plant is this process — the formation of new races — a perpetual one but is confined to certain periods. If this were not the case we should have only changing forms and always new species instead 206 The Origin of Species by Miitaiioji. conditions of life" are the chief causes of this transmu- tation, and Darwin cannot have imagined the environ- ment to have been perpetually changing. Moreover Dar- win often refers to the fact that a plant exhibits little variability during the first few years after it has been brought into cultivation but after 3 or 5 years begins to give rise to new forms. Even if the explanation of this phenomenon should turn out to be different from that given by Darwin, the fact that he insists so strongly upon it shows at any rate that the idea of periods of greater and of less mutability was present in his mind.-^ As the cause of these periods, Darwin believed that ex- ternal influences must act for many generations before they can induce any change of this kind. But if mutability is a periodical phenomenon we get round the difficulty of having to suppose that mutations should appear equall}^ at all times; and we are also in a position to account for the apparent periodicity in evo- lution. The existence of long intervals of time during which characters remain unaltered is, at any rate in the case of a great many species, a matter of tolerable cer- tainty. The frequent, although not universal, existence of the same elementary species in localities which have been separated for centuries points decisively in this direction. Moritz Wagner's famous theory of migration is based on the same fundamental idea.- We have no rea- son to expect mutability so long as its external causes are absent. So long, that is to say, as the climatic, phys- of the constant forms which actually exist." Correspondenzhlatt d. d. Ges. f. Anthropologie, Vol. 31, No. i, p. 3, Jan. 1900. ^ "I do believe that natuval selection will generally act very slozvh', only at long intervals of time.'" (Darwin, Origin, 6th ed., p. 850 "Wagner^ Das Migrationsgesets der Organismen. Mutation Within Mutation Periods. 207 ical and biological environment remains the same we must suppose that the species will not change. But if the plant extends its range, or if those with whom it com- petes for the means of subsistence, change in any way, the opportunity for the appearance of mutations is at once given. Either of these occurrences might result in a shorter or longer time in a rapid and considerable in- crease in the number of individuals and this might be the cause of the appearance of mutations on the scene. For a rapid multiplication of this kind presupposes the germi- nation of such seeds as under ordinary circumstances either w^ould not have germinated at all, or would have come to nought. This might be the case for example with seeds of weak lateral branches, of the tips of in- florescences or of flowers from accessory buds and so forth. But these are after all only suggestions; and I feel strongly that we ought to make this matter a subject for experimental inquiry; to look for species which hap- pen to be going through a period of mutation and still more to discover what are the factors which will enable us to induce such a period in a species at will. We have a doctrine of descent resting on a morphological founda- tion. The time has come to erect one on an experimental basis. y/ § 28. THE PHENOMENON OF MUTATION WITHIN THE LIMITS OF THE MUTATION PERIODS. Observations on periods of mutation have not yet been made. On the other hand many attempts based on a priori considerations have been made to discover what the phenomenon of mutation may be expected to be like. 208 The Origin of Species by Mutation. Two theses, which help to remove many difficulties standing in the way of the mutation theory, have been put forward : 1. The assumption that the new form or species does not arise merely once from the parent species but, while the period lasts, a great many times and with some degree of regularity. 2. The possibility of the appearance of useless or even harmful specific characters — whose existence is not compatible with the ordinary theory of selec- tion. The object of these considerations was to show that newly arisen forms could increase sufficiently to enter the struggle for existence with at any rate a fair prospect of success, without the help of natural selection. But the fact that the actual behavior of new forms when they arose was insufficiently known and that arguments there- fore could not start from a posteriori premises had the result that this subject received little attention. Gulick and Delboeuf are the two chief writers who have de- voted themselves to this aspect of the question. Gulick's generalization was: A^i initial tendency due to accidental variation can increase and develop in succeeding generations, zvithout reference to the advan- tage of the species. He is referring not to an extreme variant of individual variation but to a mutation; and moreover to one on which natural selection, at first at any rate, has no effect.^ J. Delboeuf is concerned to show how the final usur- pation, by the transmuted forms, of the space and means of subsistence which supported the original type is a * See Journ. Linn. Soc. Zool., Vol. XI, p. 496 and Vol. XX (iSSS") p. 215. Mutation Within Mutation Periods. 209 necessary consequence of the continuation of the cause, which gave rise to the first deviation, however shght it may have been.^ A sharp distinction between the selection and the mutation theory was not drawn at the time when Del- BOEUF was writing, so that his attention was directed indiscriminately to both of them. I shall consider the application of Delboeuf^s thesis to the latter only. And I shall further limit my analysis to the consideration of those cases in which the new form is immediately con- stant, and this, as we saw in § 25, is almost always the case. Delboeuf starts with the supposition that a mutation does not arise only once but is given off e^•ery genera- tion in a definite although perhaps a small number of individuals for just so long as the cause of the mutation continues. He further supposes that the new form can multiply in peace, and that its increase is neither aided nor hindered by the struggle for existence. Under these conditions the new form must always increase in numl)cr of individuals in relation to the parent form with a speed which will vary directly with the percentage of mutating individuals produced in each generation. From a knowl- edge of this percentage one could calculate the number of generations it would take for the new form to equal the old one in number, and also how man}- years must elapse before the new form entirely replaces its pro- genitor. In the numerical tables of Delboeuf's paper some of the more important cases are worked out in detail. The general principle, however, is quite clear: A ncn' ^J. Delboeuf, Ein auf die Umzvandlungsthcoric anzvendharcs mathcmatischcs Gesets, Kosmos, 1877-1878, Jahrg. i, Bd. II. pp. 105-127, especially p. 112. 210 The Origin of Species by Mutation. form ixHtJwut any advantage whatsoez'er in the struggle for existence zvill maintain itself provided (1) that it is sufficiently vigorous and fertile and (2) that it does not arise merely once but repeatedly during a long period of time.^ Delboeuf^s generalization has received little atten- tion. Nevertheless it seems to me, in principle and in the light of the facts of mutation, to be sound. It explains in a very simple way the existence of the vast number of specific characters which are quite useless or at any rate as to the use of which we have no idea at all — as for example the differences between the oft-cited species of Draba verna. According to the commonly accepted theory of selec- tion onl}^ characters advantageous to their possessors should arise; according to the theory of mutation on the other hand useless and even disadvantageous ones may also appear. And according to Delboeuf^s view, the latter may also persist through long intervals of time side by side with the useful variations. The premises from which he starts are at any rate warranted by actual experience. ^ With regard to the probability of this last condition I refer the reader to the instances in § 25, pp. 193-196, and to the repeated ap- pearance of sterile maize (in my experiments) both of which support this view. VI. CONCLUSIOX. 1. The student of morphological and historical evo- lution is concerned with the origin of the Linnean or col- lective species, genera, families and larger groups. The student of experimental evolution is concerned with the origin of elementary species, or rather with the origin of specific characters. 2. ''The real difficulty of Darwin's theory is the transition from artificial to natural selection" (Paul Janet). This difficulty can only be surmounted by ad- mitting that the improvement of races and the origin of new forms are really entirely different, and only ap- ])arently similar, processes. In Darwin's time no distinction was drawn between these tw^o processes. 3. "No two individuals in a generation are abso- lutely alike." This well-known saying refers to fluctu- ating variability and has nothing to do with the theory of descent. 4. ''Species have arisen by natural selection resulting from the struggle for existence." This statement also needs some explanation. The struggle for existence, that is to say the competition for the means of subsistence, may refer to two entirely different things. On the one liand the struggle takes place between the individuals of one and the same elementary species, on the other be- • tween the various species themselves. The former is a 212 Cofichision. struggle between fluctuations, the latter between muta- tions. In the former case those that survive are the in- dividuals which find conditions favorable to them — that is to say, as a rule, the strongest individuals. It is by this process that local races arise, and by it that acclimatiza- tion is rendered possible. If the new conditions of life are relaxed, the adapted race reverts to the form from which it sprang. The natural selection of newly arisen elementary spe- cies in the struggle for existence is an entirely different matter. They arise suddenly and without any obvious cause; they increase and multiply because the new char- acters are inherited. When this increase leads to a struggle for existence the weaker succumb and are elim- inated. According as the young or the parent form is l)etter fitted to the environment wmII the one or the other of them survive. Species no more arise as the result of this struggle for existence, than they do as the result of the struggle between the variants of one and the same type — though for different reasons in the two cases. In order that species may engage in competition with one another it is evidently an essential condition that they should already be in existence ; the struggle only decides which of them shall survive and which shall disappear. It is evident moreover that this ''elimination of spe- cies" must have weeded out many more than it has pre- served. In a word, from the standpoint of the theory of muta- tion it is clear that the role played by natural selection in the origin of species is a destructive, and not a con- structive one. 5. Herbert Spencer's well-known expression : ''The survival of the fittest" may mean one of two things : Conclusion. 213 either ( 1 ) the survival of the most favorable individuals within the hmits of the constant species or in the forma- tion of local races or (2) ''the survival of the fittest spe- cies" as the basis for the theory of descent. The two expressions are quite independent of one another and refer to two entirely different spheres of inquiry. 6. According to the theory of mutation species have not arisen gradually as the result of selection operating for hundreds, or thousands, of years but discontinuously by sudden, however small, changes. '' In contradistinction to fluctuating variations which are merely of a plus or minus character the changes which we call mutations are given off in almost every manner of new direction. They only appear from time to time, their periodicity being probably due to perfectly definite but hitherto undis- covered causes. The theory of the inheritance of acquired characters comes under the heading of fluctuations. Acquired char- acters have nothing to do with the origin of species. Nor can the theory of descent be applied to the solution of social problems. PART II. THE ORIGIN OF ELEMENTARY SPECIES IN THE GENUS OENOTHERA. I. THE PEDIGREE FAMILIES. § I. OENOTHERA LAMARCKIANA, A MUTATING PLANT. (plate I.) The chief obstacle in the way of getting material suit- able for investigating the origin of species is our complete ignorance of the conditions under which this process takes place. In order to obtain this material I started in 1886, to search the country round Amsterdam for spe- cies, exhibiting such monstrosities or other peculiarities as I thought would suit my purpose. As a result of my quest I brought over one hundred species into cultivation, but only one of these turned out to be what I really wanted. From this I conclude that most of the species in this locality are passing through a period of non-mutation, and that plants which happen to be actually passing through a mutable phase are encountered at any rate, relatively rarely. The plant in question is Oenothera Lamarckiana, which together with its nearest allies O. biennis and O. nitiricafa have been introduced into Europe from Amer- ica. The species Lamarckiana differs from the others by its taller growth, by its much larger and more beauti- ful flowers, by the fact that self-fertilization rarely oc- curs, bv its different leaves, and so forth. ^ O. La- ^ For the synonyms, and a discussion of the relationship as well as for a more detailed account of its origin see Sur Vintroduction dc 218 The Pedigree Fain Hies. niarckiana was introduced from America into our gar- dens, from which it has subsequently escaped. At any rate this was the case in the locahty in which I found it. This was close to Hilversum and afforded peculiarly favorable circumstances for the most minute investiga- tion. I visited the place during the summers of the years 1886-1888 almost every week, and, since that date at least once or twice nearly every year. The plant grev.' in a Fig. 42. Oenothera Lamarckiana. A flower nearly natural size. One of the petals has been removed to show the eight stamens with the pistil and its stigma. disused potato-field to wdiich it had spread from a neigh- boring park. It began to spread in about the year 1875. and during the 10 years 1875-1885 it extended over about half the field. In the succeeding years it multiplied still rOenothera Lamarckiana dans les Pays-Bas, in Nedcrlandsch Kruid- kundig Archief, T. VI, 4, 1895 ; also the later sections of this Part. Oenothera Lamarckiana, a Mutating Plant. 219 more rapidly ; until the field was finally planted witli forest trees. At the present day traces of the plant still exist. A rapid multiplication of this kind during the course of a relatively short period of time has often been con- sidered as one of the conditions for the appearance of a mutable period. This consideration led to a closer inves- tigation on the spot, which confirmed the conclusion. The plant exhibited a high degree of fluctuating varia- bility in all its organs and characters. It presented also numerous variations of another kind, of which I shall only mention fasciation^ and "pitcher"-like malforma- tions.- Most of the plants w^ere biennials, but many were annuals ; and a few lived three years, as in the case of the beet. That I really had hit upon a plant in a mutable period became evident from the discovery, which I made a year later, of two perfectly definite forms which were imme- diately recognizable as two new elementary species. One of them was a short-styled form: 0. hrevistylis, which at first seemed to be exclusively male, but later proved to have the power, at least in the case of several individ- uals, of developing small capsules with a few fertile seeds. The other was a smooth-leaved form with much prettier f,oliage than O. Lamarckiana and remarkable for the fact that some of its petals are smaller than those of the parent type, and lack the emarginate form which gives the petals of Lamarckiana their cordate character. T call this form O laevifolia:^ ^ Over de erfelykheid der fasciatien. Kruidkundig Jaarhock Do- donaea, 1894, P- 72. Cf. pp. 92-95. " Over de erfelykheid der syniisen. Ibid., p. 129. Cf. p. 165. ' Both forms are described and, in part, illustrated by Prof. Julius Pohl: Ueber Variationsweite der Oenothera Lamarckiana, in 220 The Pedigree Faniilies. , Both 0. brez'istylis and 0. laevifolia come perfectly true from seed as will be shown later on. They differ from O. Lamarckiana in numerous characters, and are therefore to be considered as true elementary species. When I first discovered them (1887) they were rep- resented by very few individuals. Moreover each form occupied a particular spot on the field. O. brevistylis occurred quite close to the base from which the Oeno- thera had spread ; O. laevifolia on the other hand, in a small group of 10 to 12 plants, some of which were flowering whilst others consisted only of radical leaves, in a part of the field which had not up to that time been occupied by O. Lamarckiana. The impression produced was that all these plants had come from the seeds of a single mutant. Since that time, both the new forms have more or less spread over the field. I could find neither of these forms in the herbaria of Leiden, Paris or Kew ; nor have the3^ so far as I have been able to discover, been described from other local- ities. Whether or no they did arise in my locality can of course no longer be determined. But I think that until proof to the contrary is forthcoming this must be re- garded as extremely probable. So much at any rate is certain that the discovery of these two species increased my hope of witnessing the origin of other species from the same stock — a hope which was soon to be fulfilled. In the autumn of 1886 I brought two samples from Hilversum to Amsterdam for cultivation in the experi- mental garden. One lot consisted of nine particularly fine rosettes with almost fleshy roots; the other, of the seed from a quinquelocular fruit from a plant growing Oesterr. Botan. Zeitschr., 1895, Nos. 5 and 6. (O. laevifolia is re- ferred to there as O. oxypetala) . The Lamarckiana-Faniily. 221 in the middle of the field. Lastly, in the autumn of 1887 I collected the seeds of O. lacvifolia. I obtained in this way three groups which, in conformity with the prin- ciple of nomenclature adopted by growers of beets, I call families ; and these I continue to grow, separately, to the present day. From these three families and their numerous lateral branches I have derived my wdiole culture, which has em- braced several thousands of individuals almost every year. Latterly several hundreds of plants have been arti- ficially fertilized for seed purposes every year. Furthermore I have imported O. hrcvistylis direct from Hilversum, because it did not arise in mv cultures. I have also occasionally made collections of seed in the field to afi^ord material for control experiments. In each of these three families new species have arisen in my garden ; and they have been essentially the same in the three groups. I shall deal with them separately : first with that derived from the rosettes, the progeny of which I shall call the LamarckiafiaA^xmXy. Of this family the main trunk (§2) and a lateral branch (§5) wnll be dealt with separately, for the sake of simplicity of treatment ; but the results arrived at with the latter agree, in their broad features, with those obtained from the former. From the seeds of O. lacvifolia the Lacvifolia-imn'iW (§6) arose; from the seeds of the above-mentioned fruit a group wdiich I shall call the La/a- family. § 2. THE LAMARCKIANA-FAMILY. The original parents of this family were, as we have already seen, moved to the botanical garden in Amster- dam in the autumn of 1886. They flowered in 1887, 222 The Pedigree Families. bearing large blossoms both on the main stem and on the numerous lateral branches, and set a quantity of seed. They were grown on an isolated bed and considered as the first generation. From their seeds I raised, in 1888 and 1889, a second generation which flowered on the same isolated bed. I chose six of the strongest to gather seeds from. The third generation was mature in 1891 ; it was not isolated, but flowered in that vear before the other cultures of Oenothera began to bloom; some days before this hap- pened all open flowers and all the buds were removed. From the seeds of the first and second generation there appeared, besides the normal plants, three hitherto unknown forms : O. nanella and O lata in some numbers, and a single example of O. rnbrinervis. My hope had been fulfilled. But the difficulties of the experiment had meanwhile become so great that I was ol)liged to give it up for a time. The laezHfolia-ia.m{\y was meanwhile continued and experiments in methods of cultivation, manuring and artificial fertilization and so forth were carried out on a large scale. The result was that in 1895 I was able to take the Lamarckiana- family up again with results which far exceeded m\- highest expectations, as a glance at the genealogical tree on page 224 will show. Since that time I have manured my plants heavily, isolated any mutating individuals as soon as they could be recognized as such and have then chosen the strongest rosettes, as early as possible, as seed-parents for the next generation. I have, further, treated my plants as much as possible as annuals; and have always chosen those which were to produce seed for the next generation in the main line from among these. So that from 1895 to 1899 I always had one generation The Lamarckiana-Family. 223 each year. Fertilization was always artificial; the flow- ers set plenty of seed when impregnated with their own pollen. The visits of insects were precluded by the use of prepared paper bags.^ The production of new species has not in the least suffered from all these precautions. I shall now summarize the whole history of this fam- ily in the form of a genealogical tree (p. 224), including in it only the main line of descent and the individuals which mutated directly from it. The table shows the eight generations in succession; the first 1886-1887, consists of the nine plants collected in the field at Hilversum; this and the two following generations each occupy two years. I did not sow the seeds which I harvested in 1891 till 1895 ; from that time on, each generation occupies only a year. In the column over which O. Lam. is written are given the approximate numbers of individuals which were examined either as seedlings or as grown plants, in each year. These num- bers do not refer to the total number of seeds sown or even to the number of seedlings that came up, but to plants which were examined separately. The table also shows the number of plants which mutated in each generation, so far as they could be rec- ognized with certainty. It is probable that these numbers are in many cases too small because I had not nearly space enough to grow all the seedlings separately until they had so far grown that their true character was a matter of absolute certainty. They had as a rule to be examined as seedlings and it is probable that m this way many cases of mutation were overlooked. I have only recorded the more important mutations ^''On the Use of Transparent Paper Bags for Artificial Fertili- sation," in Hybrid Conference Report ; Journal Royal Horticultural Society, Vol. XXIV, April 1900, p. 266. 224 The Pedigree Families. OENOTHERA LAMARCKIANA. A THE LAMARCKIANA FAMILY. I TABLE SHOWING THE ORIGIN OF NEW SPECIES FROM THE TYPE. (The figures refer to the numbers of individuals.) GENERATIONS SPECIES gigas albida ob- longfa rubri- nervis Lam. na- nella lata scin- tillans VIII VII VI V IV III II 8th gen. 1899 (annual) 7th gen. 1898 (annual) 6th gen. 1897 (annual) 5th gen. 1896 (annual) 4th gen. 1895 (annual) 3rd gen. 1890-91 (biennial) 2nd gen. 1888-89 (biennial) 1st gen. 1886-87 Hilversum and Amsterdam (biennial) 1 0 1700 21 1 9 0 3000 11 11 29 3 1800 9 5 1 25 135 20 8000 49 142 6 1 15 176 8 14000 60 73 1 1 10000 3 3 15000 5 5 The Lamarckiana-F amily . 225 in the table : others have arisen, but they have either not flowered or, being partially sterile, have set no seed ; or are of minor importance for other reasons. As examples of such we may just mention O. sublinearis and 0. suh- ovata and two or three allied types, which could not be distinguished with certainty because they bore no seed. From others as, e. g., O. leptocarpa, 0. elliptic a, and O. seviilata I have made sowings with successful results although the experiments were carried out on a small scale (cf. §§ 16-20). In the case of one form, O. spathu- lata, I have so far only obtained rosettes, and the same is true of other forms to which I do not propose to give special names. The above-mentioned 0. laevifolia and O. hrevistylis, which were found in the original locality never appeared in my cultures. The numbers on the table show that my experiment dealt in seven generations with about 50,000 plants and that of these over 800 mutated; i. e., about 1.5%, a figure which must for many reasons be regarded rather as too small than too large. In the case of every mutated in- dividual it is certain that since 1886 its ancestors w^ere normal 0. Lamar ckiana. Whether this was the case with the earlier ancestors is obviously now beyond the range of proof, but it mav be assumed to have been so with a great degree of prob- ability because of the extreme rarity of forms that showed any deviation in the field at Hilversum. 226 The Pedigree Families. § 3. THE MUTATIONS IN THE LAMARCKIANA-FAMILY. I shall now describe the mode of origin and the more important characters of the seven new species mentioned in the table. I. O. gigas.^ A vigorous plant with broad leaves, large flowers and short fruits and, so to speak, better in habit than O. Lamarckiana in every respect. It has the ap- pearance of being just as well fitted for the struggle for existence as any species of the genus to which it belongs. Even the radical leaves of quite young plants betray the identity of the new type. They are broad with a broad base which passes into the petiole abruptly instead of gradually as in the case of Lamarckiana. The leaves that appear later possess this character in a less degree, but it is always recognisable. The form of the leaf is moreover very much more variable than in any other form of the subgenus Onagra; examples with very nar- row and others with very broad leaves occur in quite small groups of individuals. Its stem is thicker than, though about the same height ^ With regard to the nomenclature it must be mentioned that my plants are burdened with a formidable series of synonyms from the very moment that they appear. Some authors regard O. La- marckiana as a variety of O. biennis. Others separate the subgenus Onagra as a distinct genus. O. gigas has therefore ihe following equally legitimate synonyms ; • O. gigas. Oenothera Lamarckiana gigas. Oenothera biennis gigas. Oenothera biennis Lamarckiana gigas. Onagra gigas. Onagra Lamarckiana gigas. Onagra biennis gigas. Onagra biennis Lamarckiana gigas. The same is true of the other new forms. It may also be noted that Oenothera is written by many authors Onothera, whilst Lamarckiana mav be written lamarckiana or Lamarkiana. The Mutations in the Lajnarckiana-Fauiily. 227 as that of 0. Lamarckiana. It is much more densely clothed with foliage than the parent form, a state of affairs brought about by the fact that the nodes are closer together and that the leaves hang down. The inflorescences are extraordinarily luxuriant, with short internodes, broad bracts and very large flowers which form a much fuller and more beautiful group than those of 0. Lamarckiana. The fruits are short and thick and more or less conical ; the seeds are very large. In spite of the high degree of variability which this plant exhibits it can be distinguished with ease from its relatives at every stage of its development.-^ This species arose only once in the Lamar ckiana-isiru- ily as the table on page 224 shows. In the other families it has also only appeared twice. Its appearance was on this wise. In 1895 I had a crop of about 14,000 plants constituting the 4th genera- tion of the Lamarckiana-ia.m{\y. All the mutated indi- viduals had been transplanted from this crop, and the majority of the Lamarckianas had been weeded out, to give more space to those which were to provide seed for the next generation. At the beginning of August I had about 1000 of these plants in flower, but many were still in the rosette stage. I chose 32 of the strongest and finest of these rosettes and planted them in a separate bed the proper distance apart. These plants grew up the next year and flowered in July and August. One of them caught my eye with its thick stem, rather compressed inflorescence and notice- ably larger flowers. On the 10th of August I picked off all the flowers, both the open ones and those which were ^ For a more detailed description of this and the other new spe- cies see the next chapter. 228 The Pedigree f am Hies. through blooming, enclosed the inflorescence in a paper bag; and, later, fertilized the flowers myself with their own pollen. The plant set a quantity of seed ; the fruits were short and thick, the seeds large. This plant was the parent of the new species O. gigas. Its ancestors were at least for three generations ordinary O. Lamarckiaiia. The new form arose without any inter- mediate stages or previous warning; it is so striking when in flower that it could not have been overlooked if it had existed before. And it must be remembered that the number of seed-bearing plants in each of the three generations were respectively only 9, 6 and 10, and that they were under continual and close observation. The self-fertilized seeds of the original plant of O. gigas were sown separately in 1897. They gave rise to somewhat over 450 plants. All of them proved to be like their parent and constituted without any question a type distinct, from the very outset, from O. Lamarckiana. Only one plant did not conform to this type; it had all the characters of O. gigas, but possessed the dwarf habit of 0. nanella. It will henceforth be referred to as O. gigas nanella. Not a single one of the 450 plants re- verted to O. Lamarckiana. Lack of space prevented me from keeping more than a quarter of this crop till the end of the summer. Many stayed in the rosette stage, others produced stems and flowered; all were pure O. gigas. I saved seed from some of these plants whose flowers had been covered with paper bags and self-fertilized. This experiment proves that this species was perfectlv constant from its very first appearance. And it remained so for the three subsequent generations.^ ^ And afterwards, until now (Note of 1908). The Mutations in the Lamarckiana-F amity. 229 We may postpone further details of this case to a later section and proceed to the following generalization as being fully warranted by the evidence. A new elementary species can appear without any oh- z'ious cause in a single individual and he absolutely con- stant from the very outset. As I have already stated this form has appeared twice again in my experiments. 11. O. olbida. A pale green, rather brittle, and very delicate form with narrow leaves; never attaining any- thing like the height of 0. Lamarckiana and bearing pale flowers and weak fruits which contain little seed. It appears every year in most of the cultures in larger or smaller numbers ; as a young rosette it is immediately recognizable. They are so weak that in the first gen- erations I imagined them to be diseased and did not record them ; that is the reason for the absence of any mention of their occurrence in the years 1888 and 1890 in the table on page 224. In spite of their frequent appearance it was not till 1896 that I could get one of them to flower. I mention this in order to allay any suspicion that crossing may have been the cause of their repeated appearance, before the 6th generation. All that remained of the 1895 crop in 1896 was a single plant, which was consequently biennial. A few isolated flowers appeared on it but they bore scarcely any pollen and set no seed after I had fertilized them with their own pollen and covered them with a paper bag. In 1897 however I succeeded in getting five biennial plants to flower; and in obtaining seed from them after artificial fertilization. From these seeds I raised a 230 The Pedigree Families. second generation in 1898 and a third in 1899, both of annual individuals. Both generations were absolutely constant and exhibited no signs of reversion ; but con- sisted only of a few individuals on account of the paucity of the harvest which each generation gave (86 in 1898 and 36 in 1899). III. O. rubrinervis. A beautiful species w^hicli often has red veins on the leaves and broad red stripes on the calyx and fruit. Markings of this kind do, it is true, occur sometimes on 0. Lamar ckiana, but they are never so pronounced that tlieir possessors could possibly be mistaken for O ruhrinervis. The flowers are somewhat larger and a rather darker yellow. The stem, especially in annual cultures, is generall}^ shorter than that of La- marckiana and suffused with red. The species cannot as a rule be recognized in the very young rosette stage '} in fact not till the plant has 10-20 leaves, or later if the plants are grown too close together. Ruhrinervis was therefore not detected until after the latas and nanellas had been identified and removed. It is then easily rec- ognizable by its narrow and long leaves with red veins, and by its vigorous habit. A very peculiar feature of this species is the brittle- ness which characterizes the annual forms to a much greater extent than the biennial ones. The stem and leaves can be broken by a moderately hard blow. If such a blow is administered to the flowering plant from above, the stem splits into several pieces with perfectly smooth surfaces of fracture. The cause of this is the extra- ordinarily slight development of the bast-fibres which however are not entirely lacking as microscopical investi- ' The distinguishing characters have since been found in verv young seedlings (Note of 1908). The Mutations in the Lamarckiana-F amily . 231 gation sliows. If a fruit is pulled off without great care the stem is usually broken in the process, an event which has more than once caused me considerable annoyance at harvest time. In all its other characters O. rubrinervis is a very healthy plant — quite as strong as O. Lamarckiana at any rate. It is the only one of my new species which is not inferior to the parent type in richness in pollen and seed. Apart from its brittleness it seems to be fully qualified for the struggle for existence. I have not however yet organized any experiments to determine this point. 0. rubrinervis appeared in the main line of descent, as the table on page 224 shows, in the third, fourth, fifth and sixth generations. There were 32 examples of it altogether. In the other families it was also ob- served from time to time; and appeared as early as 1889 in the laez'ifolia-immly. In 1897 only three appeared in the main line of the Lamarckiana-i2im\\y, whilst 10 ap- peared in the branch lines of descent. Their ancestors for at least five generations back were all 0. Lamarckiana, or at least not of the rubrinervis type. O. rubrinervis appears each time without visible prep- aration : and what strikes one most is the absolute con- stancy of the characters although these were quite distinct from those of its ancestors. When once I recognized a plant in its rosette stage as being a rttbrinervis I could predict that it would have a fragile and brittle stem and red calyx and fruit. This constancy in character is a feature of all my new ele- mentary species and is even more striking in the case of O. lata than in rubrinervis. I first started experiments on the constancy of the new rubrinervis in 1896 and 1897. In 1895 I covered 22^2 The Pedigree Families. the eight individuals which appeared in that year and fertihzed them with their own pollen, (see tahle on page 224). From the seeds thus obtained I raised a consider- able number of plants in 1896; the seeds were sown in pans and all the seedlings picked out onto a bed. W^ith the exception of a few that were to serve as seed- parents, they were all pulled up while they were flowering ( from August to October) i. e., not before the time when the rubrinerms- character was fully de- veloped in the stalk, calyx, flower and young fruit. There were some young stems and ro- settes left over after this process, but these also proved to be ex- amples of riihrinervis. The number of plants that flowered amounted to about 1000. Among these was a single La- Fig. 43. Oenothera riihrinervis. Top of the plant, showing flowers, buds, and unripe fruits. iiiarckiana which had presumably grown from a seed left in the bed from last year's sowing. Otherwise all the plants were rubri- nervis; except that some of them also exhibited the The Mutations in the Lamarckiana-Fajnily. 233 features of Oenothera leptocarpa, a new species at that time. The existence of the single Laiuarckiana was i)lainly a difficulty, so that I not only continued but repeated the experi- ment. For the continua- tion of the experiment I used the seeds saved in 1896. I raised from them, in 1897, 1114 plants; every single one of which was a riibri- nervis. For the repetition of the experiment I used the self-fertilized seeds of the four plants which had arisen in 1896 from the Laniarc- kiana-i^mily and could therefore boast four generations of pure Lamar ckiana ancestry. From these seeds I raised altogether 1862 plants, which were without exception riib- ' rinerz'is. From these facts I conclude : 1. That rnbrinervis is an absolutely ci^nstant elemen- tary species. Fig. 44. Oenothera ohlonga. Upper part of a plant at the commence- ment of flowering. 234 The Pedigree Families. 2. That every example of nihrinervis that arises in a family of another kind is capable of producing per- fectly constant progeny. IV. 0. ohlonga. The seedlings of this species can first be recognized as such at the appearance of about the sixth leaf, that is a little after 0. lata and O. nanella and consideral)ly earlier than 0. rubrinervis and 0. scintil- laiis. The leaves are narrow and with long stalks ; the transition from the leaf to its stalk is not gradual but abrupt, and the broad and pale veins have a reddish tinge underneath. 0. ohlonga can only be recognized uniformly early when the plants among which it appears are grown sufficiently far apart, but if the undoubted examples of ohlonga are removed from time to time more examples of it will be found as a result of the additional space put at the disposal of the plants. The typical form of the leaf, to which we have re- ferred, was maintained in the rosettes that were planted out. Some of the plants bore stems in the first year, others turned out to be biennials. In both cases the plants reach a moderate height only, rarely attaining a meter in height and being very much smaller than plants of Lamarckiana grown under identical conditions. The annual forms branch but little, and the branches them- selves remain short. The terminal spikes are thickly covered with flowers and buds ; the flowers themselves are smaller than in 0. Lamarckiana and develop small fruits which contain verv little seed. The biennial forms branch more, and bear plenty of pollen ; they form short 1)ut stout fruits which contain abundance of seed. Towards the end of their flowering period the oh- longa plants can be recognized from quite far off by the The Mutations in the Laniarctziana-Fcunily. 235 way in which their small unripe fruits are crcjwded to- gether. In tlie fourth generation of my Lamarckiana-i2i\w\\\ (grown in 1895 and consisting of 14,000 plants) there were 176 O. oblonga; in the fifth generation (grown in 1896 and consisting of 8000) there were 135. That is, in the one case 1.3, in the other 1.7 %. In the sixth gen- eration this proportion was maintained (29 in 1800 = 1.6%). In the last two the number has been mucli smaller because the counting had to be discontinued too early. In 1896 I got seeds from biennial plants of the fourth generation, and from annual ones of the fifth by arti- ficial self-fertilization in paper-bags. There were seven of the biennial seed-parents : each of them produced be- tween two and three hundred seeds wdiich were sown in separate lots. Altogether 1683 plants were raised from them. They were all oblonga with the exception of one which had the characters of albida. There were no ex- amples of Lamarckiana among them. Ten of the annual plants of the fifth generation set seed which was, however, scanty and germinated badl}-. Only 64 plants were raised ; of these one was O. nibri- nervis, the rest oblonga. There were no Laniarckianas. I have tested the constancy of other examples oi ob- longa wdiich have arisen in other families with the same result. Oenothera oblonga is, therefore, perfectly constant directly it arises, but it has the power of, itself, giving rise to new forms. V. 0. nanclla (O. Lamarckiana nanclla). Dwarf Oenothera. We are not now concerned with the ques- tion whether a particular form is to be described as a spe- 236 The Pedigree Families. \ cies or as a variety. Our business is to test its constanc}- by experiment. But tbe result of this will not help us to decide between species and variety. ^^j^gfy* ^v>»^ Fig. 45. Oenothera nanella. A thinly and a densely leaved individual on one of the first days of the flowering pe- riod. The plants were about 15 and 20 cm. high, whereas annual plants of O. Lamar ckiana begin to flower when they are about one meter high. If a form proves to be constant but is distinguished from another form onlv bv a single character, it is re- \ i The Mutatiofis in the Lamarckiana-Family. 237 garded by most authorities as a variety; and this is es- pecially the case with dwarf forms, which are known for a whole series of species, and attain only half or less of the stature of the species to which they belong. On this ground O. nanella may be regarded as a variety; but it must not be forgotten that, from the experimental point of view, it is just as good an element- aiy species as those which we have described already.^ The dwarfs are, perhaps with the exception of 0. lata, the easiest to recognize in my cultures. They ap- peared annually, in every culture, except the smaller ones. Among the 50,000 individuals which composed the whole Lamar ckiana-idirmXy , 1 58 were nanella ; that is about 0.3%, a proportion which was remarkably constant in successive years. The dwarfs can be easily and certainly recognized during the whole course of their development. If grown far apart and well lighted they are recognizable as soon as the second leaf appears; the first leaf is, at that time, broad with a broad almost heart-shaped base closely set on its short petiole. In 1897 I identified and recorded them by this character. Plants about which there was the slightest doubt were allowed to develop further. The broad stumpy leaves are succeeded by one or two lozenge-shaped ones with long stalks ; and the plant looks as if it were reverting to O. Lamarckiaiia. This however is not the case, for there soon follow a number of very broad leaves, with very short petioles, closely crowded together; with the result that the highly char- acteristic dwarf rosette is formed. The way I dealt with these plants in 1896 was to plant them out after ^Jordan, {De Torigine dcs orhrcs fruitiers, 1853) has pointed out that varieties are only a special form of elementary species. 238 The Pedigree Families. the second leaf had appeared in well manured soil and at a good distance apart. They were about 6 weeks old when I finally identified them. The rosettes nearly always bore a stem in the first year; I only obtained biennial plants by sowing the seed late or by crowding. The biennial form is richly branched ; the annual has very few lateral branches (Fig. 45). The internodes are numerous and very short, the broad short- stalked leaves are therefore much crowded.^ The petioles are brittle. The first flowers often open when the plant has scarcely reached a height of 10 centimeters; after their first appearance flowers are usually born at regular intervals but sometimes sporadically. The flowers are almost as big as those of O. Lamarckiana ; so that the plant in flower is very showy. The fruits are not per- ceptibly smaller than those of the parent species. In order to protect the first flowers from the visits of insects I enclosed the whole plant in a bag of parch- ment, the margin of which is attached to a metal ring which is firmly pressed into the ground. It is not until the inflorescences have attained a considerable length that the flowers can be enclosed in parchment bags in the ordinary way. The first dwarfs I fertilized in this way were some which flowered in 1893. Their ancestors which had not been protected from the visits of insects and only in- completely isolated had notwithstanding this, already ex- hibited a high degree of constancy. The seeds collected in 1893 gave rise to 440 seedlings which were all nanella. In 1895 I self-fertilized a series of dwarfs which arose in the fourth generation of my Lamar ckiana-ia.m'i\y ^ The characteristics of the dwarfs are in part due to a disease; see § i8 (Note of 1908). The Mutations in the Lamarckiana-family. 239 and had therefore at least three generations of ancestors with the normal high stature. In the same year I also sowed some seed saved from the second generation (1888- 1889) and I self-fertilized some of the dwarfs that ap- peared in the crop thus raised. There were altogether 20 of them ; they set a quantity of seed from which 2463 seedlings were grown. They were without exception 0. nanclla. Thus we see that every one of the twenty dwarfs which arose spontaneously from O. Lauiarckiana had a perfectly constant progeny. As I have already stated, the plants were not registered as dwarfs until they were strong rosettes and had attained the age of about six weeks. I repeated the experiment on a larger scale in the following year, when I had found out how to identify the plants without transplanting them. I used the seeds of nanella plants which had arisen in the fifth generation of the Lamarckiana-isimWy, i. e., plants whose ancestors had therefore been normal for four generations. From the seeds of 36 plants I raised over 18,000 seedlings. These were without exception nanella ; but 3 of them ex- hibited in addition to the dwarf habit the characters of O. ohlonga and constituted an elementary species of the second grade, O. nanella ohlonga. Moreover, whenever nanella appeared in other fam- ilies it proved immediately constant, not only in the first but in succeeding generations as well. Combinations with other characters occurred in these cultures ; but very rarely. I have often had examples of 0. lota nanella and O. nanella elliptica, and now and again variegated or pitcher forming individuals of 0. nanella, and so on. VI. O. lata. This species is solely female: it never 240 The Pedigree Families. forms the slightest trace of pollen.^ With the pollen of i^amarckiana or any of its deviations it is perfectly fertile and gives a proportion of /a/a-plants which varies about 15-20%. Julius Pohl has investigated the cause of its steril- ity.^ The pollen sacks of the open flower are dry to the touch ; they seem to be empty, but as a matter of fact contain a little pollen which is seen under the microscope to consist of empty grains which are not merely poor in protoplasm but shrivelled and stunted. The development of the anthers is at first normal, up to the formation of tetrads. About this time the surrounding cells of the tapetum elongate ; and subsequently grow into the cav- ity of the sack. These cells disappear at a later stage, when the pollen grains may be found lying in the mucus which they leave behind. I spent a great deal of time in transferring these scanty masses of pollen to the stigma in the hope of obtaining a few seeds if possible, but all in vain. If the visits of insects were prevented the plants set no seed. The exclusively female character of this mutant is very important, for it shows in a most direct way that the remarkably regular production of O. lata year after year in the Lamarckiana-i3.m\\y cannot be due to acci- dental crossing — an explanation of the frequency and regularit}^ of its appearance which is also disproved by ' Of late I have discovered a hybrid strain of lata which produces pollen in a greater or smaller part of its flowers. These plants, when self-fertilized, produce seed, which gives 15-20 % examples of lata] and 80-85 ^c of Lamarckiana. Also the same figures as by pollina- tion with the parent-species. This proves the O. lata to be an in- constant species. Seeds of this strain have been distributed to some of my correspondents, who also found the type to be inconstant. — (Note of 1908.) "Julius Pohl, Ucher Varintionszvcite dcr Oenothera Lamarc- kiana, Oesterr. bot. Zeitschrift, 1895, Nos. 5-6. The Mutations in the Larnarckiana-Family. 241 the artificial fertilization of the whole ancestry in my pedigree-cultures. Fig. 46. Oenothera lata. Top of a stem, with buds in the axils of the broad bracts ; a, b, c, separate buds in various stages of development; A, B, C, buds of Oenothera La- marckiana in corresponding stages. The buds of lata are seen to be palpably fatter than those of O. Laniarckiana. 0. lata can easily be distinguished at every stage of its existence from all its allies ; in fact as soon as the 242 The Pedigree Families. second or third leaf unfolds: these leaves are broad with broad bases and long petioles. But most characteristic is the broad and round shape of the tip of the leaf, a fea- ture which is more or less distinctly pronounced during all the rest of the life of the plant. The plants are always low although the rosettes are large and strong, stronger sometimes than those of Lamarckiana itself. The stem is limp so that the top hangs over to the side even in the healthiest plants. It is thickly covered with dense foliage. The leaves are rounded at the apex and much crumpled. The top of the growing stem, both in its young stages and when it is covered with flowers, is in the form of a com- pressed rosette. Everything in this plant is stout and broad, so that they came to be known among us in the garden as ''fat- heads." This character w^as particularly noticeable in the case of the flower-buds just before they opened, and has been well brought out by Pohl^s figures. The petals do not unfold themselves completely but remain more or less wrinkled. The stigmas are peculiar. As in the case of O. Lamarckiana their number varies from a nor- mal of four up to 8 and more, forming a so-called half- curve. This matter has been made the subject of a thor- ough study by Verschaffelt in the case of 0. La- marckiana. The fusion of neighboring stigmas, which occurs in the parent form, occurs in O. lata also. The unequal development of the stigmas which occurs now and again in Lamarckiana is exaggerated to an extra- ordinary extent in the daughter species; and the most curious malformations arise as the result of the fusion referred to.-^ They do not however interfere with fertili- zation. ^ For figures of these see Pohl's paper, loc. cit., Taf. X, Fig. 27. The Mutations in the Larnarckiana-Fainih. 243 The fruits are short and thick and contain relatively few but, as a rule, large seeds. O. lata appeared pretty regularly in my cultures, but in proportions which varied greatly. And as they could be easily and certainly recognized, even under unfavor- able circumstances such as crowding, these deviations are indices of a real variability in proportions and not of the difficulty of identification which may have affected the proportions in the case of the other forms. The pro- portion of lata plants was sometimes as low as 0.1% ; in the fifth generation it was as high as 1.8% ; i. e., about the same as that of O. oblong a. VII. O. scintillans. Except for 0. gigas which has so far only arisen three times, and 0. spathulata, 0. subo- vata, O. leptocarpa and others which I shall refer to later on, O. scintillans is by far the rarest form in my cultures. It arose only eight times in the Lamarckiana-idimWy , and in other families still more seldom. It does not, like the other species, breed true when self-fertilized, but behaves in a very peculiar way. Seeds from it give rise to three forms in considerable numbers : 0. scintillans, O. oblonga (Fig. 44 on page 233) and 0. Lamarckiana. This is a different phenomenon from that with which we are already familiar in the other elementary species, namely the ver}^ occasional production of a mutation in about 1 in 1000 plants. There often arise in this way elementary species of the second order, i. e., species which combine the characters of two species. These also arise in the case of 0. scintillans; e. g., 0. scintillans nanclla and 0. scintillans elliptica. But only very occasionally, i. e., one such among thousands of normal O. scintillans. A verv remarkable feature of this instabilitv of O. 244 The Pedigree Families. scintillans is that the proportions, in which the different forms occur, are by no means low at first, and that they cannot be increased by selec- tion. Perfectly definite prin- ciples underlie these propor- tions ; for the behavior of an O. scintillans from one stock is the same as that derived from another. Some mutants of O. scin- tillans had a capacity of produ- cing 35-40 % scintillans; and others a capacity of producing 70 % or more. Moreover, this capacity seems to be inherited. I first noticed O. scintillans in 1888, in a culture from seeds of the /a/a-family (§7). The plant was biennial : I did not sow its seeds till 1894. In 1895 I fertilized some of them (which, it will be seen, flow- ered in their second year) with their own pollen. I treated 14 plants in this way, but they set little seed. In 1896 1 raised only 400 plants from them as fol- lows : 68 % O. Lamarckiana Fig 47. Oenothera scintil- lans. Top of an annual plant. 15 % O. scintillans 15 % O. oblong a 2 % O. lata and one plant of O. nanella. The Mutations in the Lamarckiana-Family. 245 In 1898 there arose in another 0. /a/a-family a single example of 0. scintillans which flowered in its first year and was self-fertilized in a bag. 148 young plants were raised from its seeds, as follows : 55 % O. Lamarckiana Z7 % 0. scintillans 7 % O. oblong a 1 % O. lata. I have tested the hereditary capacity of three exam- ples of 0. scintillans which arose directly from the 0. Lamar ckiana-i2im\\y. The first was a plant which arose in 1895 but did not flower till 1896; it had a number of lateral stems on all of which the flowers were fertilized. The various sowings of their seeds gave the following species and proportions : 52-59 % O. Lamarckiana 34-36 % O. scintillans 3-10 % 0. oblonga 1 % O. lata. I manas:ed to obtain self-fertilized seeds from nu- merous O. scintillans which appeared in this culture ; the proportions of the various kinds produced by them were subject to considerable fluctuation, but they were essen- tially the same as those given by their parents. Of the six plants which, as shown in the table on page 224, arose (1896) in the fifth generation of 0. La- marckiana I only succeeded in bringing two through the winter and in getting them to flower (1897). They were self-fertilized in bags. The seeds of one plant gave like the others : 51 % 0. Lamarckiana 39 % O. scintillans 8 % O. oblonga 246 The Pedigree Families. 1 % 0. lata 1 % 0. nancUa. But the second plant gave a much "better" result for the 200 seedlings were distributed as follows : 8 % 0. Lamarckiana 69 % O. scintiUans 21 % O. oZ^/on^ra 2 % 0. lata, O. nan ell a, etc. I fertilized about 25 of those O. scintiUans with their own pollen. I sowed the seeds of each one separately and got a proportion of scintiUans in the next generation varying between 66 and 93 % about a mean of 84 %. One plant seemed to give O. scintiUans only ; but the crop from this plant was small. In 1899 and 1900 I continued my experiments in the same way, in order to find out whether this species could be brought, by selection, to the same level of constancy as that which characterizes the other elementary species. Perhaps some day there will appear in the Lamarc- kiana-i2irm\y or in some other a scintiUans which will breed true straight away. An estimation of the constancy of scintiUans mutants has been made four times in all. Three times it gave offspring like itself in the proportion of 34-39 % ; in the next generation it did the same. In the fourth instance the proportion was 69 % and the mean, in the next gen- eration, had mounted to 84 %. These numbers show in my opinion that 0. scintiUans is an inconstant species which moreover tends to give rise on its first appearance to other forms and especially to O. ohlonga.^ * The deviation in the first series of figures (15 % instead of 34- 39%) which occurred in a second generation, from a scintiUans The Laws of Mutation. 247 It now remains to give a short description of the spe- cific character of this form. Scintillans cannot be recognized until quite late :^ as a rule they could not be identified before the rosettes had quite a considerable number of leaves about 10 cm. long. The leaves are small, narrow and with long petioles ; with shiny surface (whence the name), dark green, and with hardly any trace of crumpling. The veins are white and often broad. The ends of the leaves turn down, so that the leaf makes an arch over the ground. The stems never attain a great size ; they are thin and l)ear short leaves; they produce flowers early, forming- long spikes. The annual forms are usually only feebly branched ; the biennial ones more profusely. The flow- ers are small ; a little smaller or about the same size as those of 0. biennis. As in O. Lamar ckiana, the stigmas project beyond the anthers. The fruits are small, the quantity of seed in annual plants is also small ; and many of the plants begin to flower too late to set any seed at all. The dark green color and shiny surface occurs on tlic stem leaves as well, and gives the plant a peculiar a])- pearance quite different from that of 0. Lamarckiana. § 4. THE LAWS OF MUTATION. I propose now to recapitulate the conclusions which I have drawn from my experiments. The various ele- mentary species we have dealt with behave in essentially the same way; so also does a secondary branch of the whose proportion in the first generation is unfortunately unknown, seems also to point to some susceptibility of this proportion to factors we do not yet understand. * Since writing this, I haA^e succeeded in recognizing them as young seedlings, with only 2-4 leaves (Note of 1908). 248 The Pedigree Families. family we have just considered, as well as two other primary families which will be described in section 5 ; not to mention a number of subsidiary families and cul- tures. The main conclusion is that the facts of muta- bility can be described by laws just as definite as the laws of variability. The following generalizations apply in the first in- stance to the new forms which have arisen from Oeno- thera Lamarckiana; but it should be stated that they are completely in accordance w^ith a whole host of observa- tions, for the most part of a horticultural nature, on other genera and families. I. N'ezu elementary species arise suddenly, without transitional forms. A great point in my experiments has been that the ancestors of the newly arisen forms have always been accurately known, and often for many generations back; and that they were either isolated as a group (1887-1891) or that they flowered in isolating- bags and were artificially fertilized (1894-1899). There is no mention of any such precaution in horticultural records. This precaution enables us to be certain that each new form arose from the seed of a normal specimen of Oenothera Lamarckiana. The new form always arises with all the characters proper to it. Once the identity of a seedling is recognized the characters which it will gradually assume can be predicted, and in every case the prediction has been fulfilled. Many opportunities for testing the degree of cer- tainty in identifying seedlings offered themselves during the course of the experiments, and especially when the chosen seedlings were planted out and flowered in my garden. When there are hundreds of individuals to record, The Laws of Mutation. 249 it Is natural that one should occasionally be in doubt over some of them ; particularly over such as happen to grow between others, and have not, on that account, sufficient space for their full development. I have usually given these plants an additional lease of life, in many cases the whole summer. They then very soon proved to be a pure type ; or perhaps, to be compound forms such as 0. lata nanclla, O. scintillans elliptica and so forth, or, lastly, new forms altogether. But they never turned out to be intermediate forms. Transitions be- tween the various elementary species did not occur. As a matter of fact I have thought on one or two occasions that I had discovered examples of such inter- mediates. For example \ once noticed a plant which was like O. lata in many respects but bore plenty of pollen. I fertilized the plant artificially and raised 270 plants from its seed. These were all like their parent except for 1 % of them which were true lata, that is to say no larger a percentage of O. lata than the Laniarckiana- family Itself can give rise to. I have called this form O. semilata (§ 17). Other cases of a similar nature have been observed. The seed of a newly arisen form will, If sown, always give rise to plants with exactly the same characters as their parents : and this purity of the new form is main- tained in subsequent generations. II. New elementary species are, as a rule, absolutely constant from the moment that they arise. The seeds set by an example of a newly arisen species after arti- ficial self-fertilization give rise solely to plants like itself; without, as a rule, any trace of reversion to its parental form. This is equally true of O. gigas which has only arisen 2dO The Pedigree Faniilies. three times, and of forms which have appeared as fre- quently as have 0. albida, 0. ohlonga, O. rubrinervis and O. nan el I a. The point cannot be decided in tlie case of O. laevi- folia or O. brevistylis, both of which were found on the spot where O. Lamarckiana was originahy discovered, but have not arisen in my cuhures. Both these, when self-fertihzed, come perfectly true from seed. O. hrevi- stylis breeds true in spite of its small fruits which some- times set no more than a solitary seed. Indeed I thought at first that these fruits were absolutely sterile. Oenothera scintillans and 0. lata are exceptions to this rule. The seeds borne by self-fertilized plants of the first named form produce a generation only about one-third of which is 0. scintillans. This is true of the seeds of three distinct individuals which have arisen quite independently of one another. From the seed of a fourth individual however 69 % of O. scintillans were raised ; and these again in the next generation gave from 60 to 90 %.i This constancy of the new species is an extremely important characteristic. It has enabled O. laez'i folia and 0. brez'istylis to maintain themselves in the spot where they arose — mere scattered examples among the host of Laniarckianas which surround them ; and, what is more, pure in respect of all their characters (apart of course from accidental crossing) That the struggle for existence is a pretty keen one in the field in question may be gathered from the fact that a vigorous Lamarckiana can bear 1 00 fruits and that each fruit contains between 200 and 300 seeds. The whole ^ For further information on this point see section 19 of this part. The Laws of Mutatioii. 251 field contains no more than some thousands of plants, that is to say not much more than could be supplied by the seeds of a single individual. The seeds which do not produce adult plants either do not germinate or the seedlings which come up die young. Yet in spite of the severity of the competition, O. laevifolia and O. hrcvi- stylis have maintained themselves for more than twelve years. -^ III. Most of the new forms that have appeared are elementary species, and not varieties in the strict sense of the term. Elementary species are distinguished from their nearest allies by nearly if not quite all their char- acters. The differences are often so slight as to escape notice by an eye not trained to observe them ; and they are particularly apt, as systematists so often complain, to become lost in dried specimens. This latter is how- ever fortunately not the case with the new forms whose origin I have witnessed ; for they are distinguishable from one another and from O. Lamarckiana as herbarium specimens far more easily than, for example, specimens of this last species are from O. biennis. This close familiarity with each form can only be attained by a careful and minute study and description of all the organs of the plant at every stage of its devel- opment. Once a plant is thoroughly known in this way it can be recognized at almost any stage. Varieties are distinguished from the mother species usually by one single character or at most by two or three, whilst they resemble them in all others. Apart from this point, the difference between species and vari- eties is to a large extent arbitrary, since when tested ex- perimentally the one is just as constant as the other. *And afterwards until now (Note of 1908). 2}i2 The Pedigree Families. It is rather curious that all the new forms which have arisen in my experiments should have been species in this sense and not varieties. I have always hoped to get a white flowered form or some other such distinct variety but so far in vain. O. nanella is perhaps the only form which can be called a variety in the horticultural sense of the term. It is a charcteristic of varieties that they crop up in a great number of unrelated species, genera and families. For example the varieties rosea, alba, laevis, inermis, la- ciniata, prolifera, bracfeata, and penditla. It is the same with monstrosities: e. g., var : plena, fasciafa, torsa, ad- nata, fissa and so forth. The same is true of dwarfs or the var. nana. But with the exception of O. nanella I cannot find in other families and genera any series of forms analogous to mine. It is for these reasons that I do not consider them varieties. A very popular definition of varieties is that they are forms which are known to have arisen from other forms. This position is obvious!}^ untenable. The proof of their origin may exist in the case of some few horticultural varieties but with the vast majority of them and with all wild varieties this proof does not exist at all. Their origin is a thing of the past and when, as is usually the case, it was not witnessed by human eyes the so-called "proof" of it is based on deduction or analogy. And in all cases, where we are not dealing with direct observation, the origin of varieties is in no sense what- ever more certain that that of collective species or genera. I have dwxlt on this point because I feel quite certain that many of my readers will regard my new forms as The Laws of Mutation. 253 varieties for the very reason that I have been able to ob- serve their origin.^ IV. New elementary species appear in large numbers at the same time or at any rate during the same period. Scott's palseontological results have led him to conclude that species-forming variability, or, as he also calls it, mutability must appear simultaneously in large groups of individuals and that the causes of these changes have probably been working through long periods of time.^ The palaeontologist investigates the problem of the origin of species only in broad outline. It is the experi- mental physiologist who deals with the separate individ- uals themselves and with their posterity, of whom not a millionth part would ever be preserved in the fossil state. We have no right therefore to expect more than a general agreement between the conclusions attained by these two lines of investigation. And when we do find such agreement, as we do in the present instance, I think it is extremely desirable that it should be put on record. Amongst the species which have arisen in my experi- mental garden Oenothera gigas has only been observed once. The others appeared every, or nearly every, year in varying, and often, indeed, in considerable numbers. More than 800 individuals of the seven new species we have described arose independently from one another from the Laniarckiana-isimiiy. And as about 50,000 plants were cultivated during this period of time the ^Fortunately, as a matter of fact, this has not been the case (Note of 1908). ^W. B. Scott, On Variations and Mutations. Amer. Journ. Sci.. 3d Ser., Vol. 48, No. 287, Nov. 1894. See p. 2>72>- '"Forces both ex- ternal and internal similarly afifect large numbers of individuals." 254 The Pedigree Families. number of new forms amounted to between 1 and 2 % of the total cultivated. In other words : The new elementary species arose from the parent form in a ratio of 1-2 %. Sometimes more than, but oftener less than, this value. And this ratio was maintained throughout the whole course of my experiment, so far, at least, as the difference in the methods of investigation which have been employed at different times permit me to estimate it. This figure, 1-2 %, is more probably too small than too large. For it was only in the years 1895 and 1896 that I went to the labor of determining it accurately. In previous years the average was considerably lowered by other circumstances, the most important of which was the omission of such forms as O. ohlonga, O. nihrinervis and O. scintiUans which at that time I could not recognize in their early stages. The table on page 224 shows, for the two years 1895 and 1896, 22,000 individuals of La- marckiana and 711 of the new forms. That is, more than 3 %. O. laevifolia and 0. brevistylis formed far smaller a percentage than 3 % of the number of Oenotheras growing on the original field at Hilversum ; yet they, obviously, arose in quantities sufficient for them to main- tain themselves. We may conclude therefore that a yearly appearance in the proportion of from 1 to 3 % would be sufficient for the establishment of a new species.^ V. The nezv characters have nothing to do with in- dividual variability. Oenothera Lamarckiana exhibits a degree of fluctuating variability in all its characters which is certainly not less than that exhibited by other plants. The new species fall right outside the range of ^Compare the calculations of Delboeuf, as given above (I §28). The Laws of Mutation. 255 this variability; as is evident from the fact that they are not connected with the parent type by intermediate or transitional forms. New races can of course be evolved by repeated selec- tion in one or another direction in Lamarckiana just as much as in any other plant. Indeed I have, myself, produced a long-fruited and a short-fruited form in this way. But such races remain dependent on selection and differ from their type only in one feature : they do not bear the slightest resemblance to elementary species. Elementary species themselves exhibit fluctuating variability, and often indeed to a greater extent than the parent form. Nearly all their organs and characters vary, but never in such a way as to approach the original form. VI. The imitations, to zvhich the origin of new ele- mentary species is due, appear to he indefinite, that is to say, the changes may affect all organs and seem to take place in almost every conceivable direction. The plants become stronger (gigas) or weaker (albida). with broader or with smaller leaves. The flowers become larger (gigas) and darker yellow (rttbrinerz'is), or smaller (oblonga and scijitillans) and paler (albida). The fruits become longer (rubrinervis) or shorter (gi- gas, albida, lata). The epidermis becomes more uneven (albida) or smoother (laevifolia) : the crumples on the leaves either increase (lata) or diminish (scintillans). The production of pollen is either increased (rubrinervis) or diminished (scintillans) ; the seeds become larger (gigas) or sm.aller (scintillans) , more plentiful (rubri- nervis) or more scanty (lata). The plant becomes fe- male (lata) or almost entirely male (brevistylis) ; many forms which are not described here were almost entire!)' sterile, some almost destitute of flowers. 256 The Pedigree Fain Hies. O. gigas, O. scintillans, 0. ohlonga tend to become biennial more than O. Lamarckiana ; and O. lata tends to become less so; whilst O. nanella cultivated in the usual way scarcely ever runs into a second year. This list could easily be extended, but for the present it may suffice. To regard the new forms from another point of view, some of them are fitter, some unfitter than the parent form, and others neither the one nor the other. Until experiments have been made with the new forms sown in the field it is obvious that no definite conclusion on this point can be arrived at : nor do the observations which have so far been made on the plants growing in the field at Hilversum throw any light on the subject. Nevertheless it is evident that the female 0. lata is at a great disadvantage; and that 0. albida with its nar- row leaves is, at any rate in its early stages, far too deli- cate. O. rubrinerz'is looks quite robust but is very brittle and liable to be broken. Annual plants of O. ohlonga bear hardly any seeds, whilst O. nanella is very small and its petioles are often brittle. All these forms appear to me to be less fit as compared with O. Lamarckiana. On the other hand 0. laevifolia seems to be at least a match for its parent; and O. gigas in many respects superior to it : all its organs are larger and stronger and apparently better adapted to perform their functions; the whole plant is stouter. Sowings of this species in the open should give favorable results. The forms which have not yet been described (0. spathulata, suhovata, etc.) are hampered in the struggle for existence by their almost complete sterility. O. sub- linearis with its slender grass-like leaves is much too weak in its early stages — and so forth. The Laws of Mutation. 257 Many authors already hold that species-forming vari- ability must be indiscriminate. We are strongly opposed to the conception of a definite ''tendency to vary" which would bring about useful changes, or at least favor their appearance. The great service which Darwin did was that he demonstrated the possibility of accounting for the evolution of the whole animal and vegetable kingdom without invoking the aid of supernatural agencies. Ac- cording to him species-forming variability exists without any reference to the fitness of the forms to which it gives rise. It simply provides material for natural selection to operate on. And whether this selection takes place between individuals, as Darwin and Wallace thought, or whether it decides between the existence of whole spe- cies, as I think ; it is the possibility of existence under given external conditions which determines whether a new form shall survive or not. We can go a step further and say that many more useless and unfavorable variations must arise than favor- able ones. This becomes sufficiently evident when we consider the complexity of the conditions which an organ- ism has to satisfy before it can supplant its fellows. The mutability of Oenothera Laniarckiana satisfies all these theoretical conditions perfectly. Nearly all or- gans and all characters mutate, and in almost every con- ceivable direction and combination. Many combinations must obviously be fatal to the life of the germ within the ripening seed and cannot on that account be observed. Others hinder the development of the seedlings and wdiole series of experiments with apparently mutated plants came to nothing in spite of every care on account of the premature death of the young plants. Many combina- tions reduce the fertilitv so much that we cannot jro fur- 258 The Pedigree Fainilies. ther than observe the mutated individual itself. A num- ber of other combinations are, I suppose, lost in my ex- periments because they cannot be detected until the plants are fairly old, by which time the great majority have been weeded out to make more room.^ Such considerations seem to me to explain how it was that I was able to cul- tivate only so small a number of new species through more than one generation. And it is of course open to question how many of those that I did cultivate could survive the struggle for existence. I conclude therefore : Mutability is indiscriminate. Some mutations bear no offspring and disappear forth- with. Between the others and the species already estab- lished natural selection must decide, unless artificial se- lection steps in. VII. MufabUity appears periodically. I am led to this conclusion by my experiments ; but I express it at present only tentatively. The fact that of all the species that I have examined so far, only one has proved to be in a state of mutation appears to me sufficient evidence for this conclusion. But further investigations are neces- sary for the establishment of the generalization : and such I have only recently started. I am not of course now in a position to give experimental proof of the existence of mutable and immutable periods : but I have enunciated the hypothesis of their existence as the sim- plest explanation of the remarkable fact that I have so far observed mutations only in a single species ; though plentifully enough in it. The above generalizations refer in the first instance to the case which we have observed, namely the muta- * For example O. brevistylis and O. leptocart i" 1899. "the longer can the plant keep the water for its own use." How far O. laevifolia is inferior to other Evening Primroses from its lacking these depressions is not an easy question to answer; but this much is certain that I have always found it weaker and smaller than the parent species. In the experimental garden however where the \ The deviations from the type of leaf, characteristic of the species, which often occur at the bottom of branches are classified by Delpino as subvariations. See Delpino, Teoria generate delta Fillotassi, 1883. They are often of an atavistic natuir ^VoN RiJMKER^ Zuckerruhensiichtung, 1894, P- 6. Oenothera Laevi folia. 313 plants never lack water this character makes no differ- ence. A very characteristic feature of O. laevifolia is af- forded by the flowers on the weaker shoots. They have narrow petals which exhibit every transition from the broad, obcordate contour of those of the strongest flowers to oval or elliptical forms as shown at c and d in Fig. 59. This character is very constant. It was through it, that I first discovered the new form and it was only after this had been in cultivation for some time that I became acquainted with the smooth leaves. Weak plants bear such flowers on the main stem ; stronger ones either on the whole extent or only at the base of the lateral branches. In the height of summer these flowers are rare ; but towards autumn and often as early as the beginning of September they appear in greater numbers. By cultivat- ing only healthy plants without lateral branches it would be possible for a whole year to go by, without seeing one of these flowers : this has sometimes happened in my experiments. There is something extraordinarily attractive about these flowers. They are smaller and neater than the rather gross and stout flowers of the common Evening Primrose; their color is often paler; their form, in a sense, freer, inasmuch as the petals scarcely touch one another. I have often stuck them into my journal or photographed them. I have found them from 1887 up to the present day, always the same, exhibiting the same varieties of form but without progressing in any one particular direction, just like all the other new species which have proved constant in all their characters from their origin. 314 Origin of Each Species Considered Separately. The forms of the petals of a single flower often differ from one another (Fig. 59 b). The petals of plants grown in the field on dry sand were narrower than those of plants grown in the garden on manured soil. The petals of the former were almost twice as long as broad, in those of the latter the relation of length to breadth was as 2 to 3. The emarginate character of the normal petals is absent in them; the petals are, on the contrary, obtusely rounded. Their greatest breadth is in the middle. The narrowest petals that I have observed were three centimeters long and one broad. But as I have already stated there exists a complete series of transitions be- tween these and the obcordate ones of strong flowers. Oval petals are by no means confined to O. laevi folia. They occur regularly on O. elliptic a. I have also some- times found them on weak shoots of 0. biennis. In the other characters 0. laevifolia is very much like O. Lamarckiana, not differing from it in any essen- tial features save those already mentioned. The plants are about the same size. So are the flowers and fruits and general habit. Nevertheless a bed of 0. laevifolia, even if it is some distance away can always be recog- nized from a group of Lamarckiana by characters which may be manifested to a greater or lesser degree but which always tend in the same direction. The color of the flowers, especially of the later ones, is usually a little paler; the buds a little thinner, the bracts of the inflor- escence a little narrower and the whole plant more deli- cate and neat. During the first few years of its cultivation I used to allow O. laevifolia to cross freely with O. Lamarckiana for reasons which I have mentioned above (§6). But since 1894 I have excluded the visits of insects by en- Oenothera Brevistylis. 315 closing the flowers in parchment bags ; and fertiHzed them with their own pollen. Since that time the species has proved absolutely constant; and each year I choose the best examples with the smoothest leaves as seed-parents. § II. OENOTHERA BREVISTYLIS. This form has been thoroughly investigated and de- scribed by Julius Pohl.^ I have used it mainly in hy- bridization experiments, in which it behaves in a different way from all other species in the group of the evening primroses. I shall deal here only with its external char- acters, with its first discovery in the field and with its con- stancy. This species, which is very easily recognized dur- mg its flowering period, has never arisen in my cultures. In the rosette stage and in fact at any time before it flowers, it is difficult to distinguish. Its more rounded leaves give it a slightly different appearance; and in hy- brid cultures it is often possible before any stems have been developed to predict whether there will be many or few brevistylis. But it is not until the flower buds appear on the stem that the difference between it and other forms becomes clearly discernible, and that one can record them with safety. The young inflorescence forms a rosette of rounded leaves on the top of the stem in O. brevistylis and of pointed ones in O. Lamarckiana. Shortly after- wards the buds appear; they are shorter, thicker and blunter than the slender conical ones of the parent spe- cies. Then the flowers open, just as large and just as beautiful as in Lamarck's Evening Primrose. At first sight it looks as if they had neither a style nor stigma; * Julius Pohl, Ueher Variationsweite bei Oenothera Lamarckiana, Oester. bgt. Zeitschrift, Jahrgang 1895, Nos. 5 and 6, Tafel X. 316 Origin of Each Species Considered Separately. but closer inspection reveals them hidden away in the tube at the base of the corolla. Hence the name O. brevi- stylis or short-styled Evening Primrose. The length of the style varies very much; the stigmas sometimes lie right inside the tube, sometimes they stand a full centi- meter out of it. But there is a great gap between the longest styles of 0. brevistylis and the shortest ones of O. Lamar ckiana. When the flowers are through blooming they wither down as far as the fruit but are not thrown off as they are in 0. Laniarckiana, but remain attached for considerable time to the unripe fruit. The plants can be recognized from afar by this character, and even perhaps still more readily by the smallness of their fruits. In their fully developed state these are hardly larger than the ovaries of open flowers ; they remam bent outwards, pressed against the bract and almost hidden between the broad auricles at its base. From a distance it looks as if the plant had never been fertilized: LamarcHawa on the other hand does not hide its great, fine, erect fruits between the bracts. (Plate I.) In the fruiting period Brevistylis plants can therefore be more easily recognized than those in blossom; but as a rule brevistylis keeps up flowering later into the autumn than O. Laniarckiana. The stigmas are developed in an unusual way; for instead of being stout and cylindrical they are flattened and leaf-like. They retain the abundance of pollen that is brought to them by bumblebees, permit the develop- ment of pollen tubes which elongate in the usual way, and reach the ovary in numbers but fertilize only very few ovules. Many plants set no seed at all, others very little. Oenothera Brevistylis. 317 The ovary extends a little above the insertion of the corolla up into the style. O. brevistylis was the first sub-species of Oenothera Lamarckiana which I discovered. I found it in August of the first year of my investigations, 1886, when, as already stated (p. 266) it occupied a little corner in the northeast of the field. There were two individuals, one where the plants grew thickest, the other on a spot about one hundred paces away. Both were well developed, flowering from many shoots and, as far as I could judge, biennial. I found them on the 25th of August. They caught my eye from quite a distance by the almost com- plete absence of any fruit on them. This character made it easy to be certain that only these two had been short- styled when in flower, for all the others had set normal fruits. In 1889, the part of the field in which these two short- styled plants stood was well cleaned and dug up; never- theless, towards the end of July of that year I found a group of 12 short-styled individuals nearly in the middle of the field on a spot where not a single Oenothera grew in 1886. The new species has since maintained itself on this spot and it has been observed there nearly every year. In the summer of 1894 I saw six plants there in flower; in August of 1898 they were fairly numerous, but since then they have appeared only sporadically. Before 1895 I thought O. hrevistyUs incapable of setting seed; for I regarded it as being solely male. In 1895 I collected lots of fruits and got a meagre quantity of seed which seemed to me to be empty, so that I did not sow it in the following spring, but when in the fol- lowing autumn I had gone over my cultures thoroughly I came to the conclusion that it might be worth while to 318 Origin of Each Species Considered Separately. sow the seed. Of the whole quantity of seed borne by 200 fruits there germinated a httle over 300 seeds. That is 1 to 2 seeds per fruit. The mother plants had grown amongst other kinds and were fertilized by bumblebees and therefore largely crossed. Nevertheless of the 83 flowering plants to which this seed gave rise 69 or 83 % were O. hrevistylis. This result encouraged me to try artificial self-fertili- zation in parchment bags. For this purpose I chose in 1897 those plants whose stigmas projected farthest out of the tubes; for I had satisfied myself that as a rule these furnished the largest fruits. I harvested seeds from five plants. In 1898 I sowed the seed of each plant separately. Nearly all the young plants flowered between August and October; they were all, without exception, short-styled. Altogether there were 175; some in flower and some only with buds, in which, however, I was able to observe the length of the style. Oenothera hrevistylis, therefore, when self-fertilized, is absolutely constant in spite of its comparative sterility. B. THE CONSTANT NEW SPECIES. § 12. OENOTHERA GIGAS. (Plate II.) Oenothera gigas is at once the finest and rarest new species that has arisen in my cultures. Whereas most of the new forms are weaker than the parent species this one is almost in every respect stronger and bigger and more heavily built. A comparison of Plates I and II will show at a glance the nature of the difference be- tween gigas and Lamarckiana ; both represent the top of Oenothera Gigas. 319 the main stem in September, by which time the lower fruits are fully grown. The top of the plant still bears a head of flowers and buds. Figures 60 and 61 repre- sent the same two forms at the beginning of the flower- ing period. In warm weather the flowers of the Evening Prim- roses open in the evening usually at a rate of 2 or 3 a day, seldom more and sometimes fewer according to the weather. They are pollinated by bumblebees and by Noctuidae {Pliisia gamma, Agrofis segetiim and others) and as a rule wither during the night. The beauty of the flowers has completely disappeared by the following- morning. It is only in cool or even cold weather that the flowers remain open till the following day; but even then they seldom last on into the evening. The opening of the flowers has been described by E. RozE.^ The event is a very remarkable one. Early on a beautiful summer evening, when the plants bear nothing but buds and dead flowers, while one may be busied with other operations in the garden, one looks round and suddenly sees every plant in blossom. Half an hour suffices to change the whole aspect of the gar- den. The process of opening is in preparation all day. The buds become yellow ; their anthers have completely de- hisced. The tops of the sepals are still joined together to form an entire cap which however becomes split lower down during the course of the day. The petals gradually swell until at last they veritably burst the calyx open, throw the sepals backwards and unfold their free ends. The whole thing happens in a few minutes or seconds. * E. RozE. U epanouissement de la Heur de VOenothera suaveolcns Desf. Bull. Soc. bot. France. T. XLII, 8 Nov. 1895, p. 575. 320 Origin of Each Species Considered Separately. The petals now stand out in the form of a cross; their inner halves being still rolled up together. But it is not Fig. 60. Oenothera gigas. Top of a stem just beginning to flower. A petal has been removed from the flower a. h, a withering flower. long before these unfold and set the anthers and the style free. My new species agree in all these points with the Oenothera Gigas. 321 parent species and its related forms, 0. biennis and so forth. One of the most distinctive features of Oenothera gigas hes in the breadth of the petals. The swollen Fig. 6i. Oenothera Lamarckiana. Top of a stem begin- ning to flower, a, the lowest flower withered and fallen down on the bract. character of the buds is due to this feature; as also is the cup-like shape of the base of the open flowers. The petals in this species, as in Lamarck's Evening Primrose, 322 Origin of Each Species Considered Separately. are obcordate and more or less deeply emarginate at their broad apices. In both species the petals are about 3 cm. long; but in O. Lamarchiana they are 5 cm. broad, whilst in O. gigas they are 6 cm, I did not find any other differences worth mentioning in the absolute or relative dimensions of the flowers. The size of the flowers of both species gradually de- creases as autumn comes on, a fact which must be borne in mind when we are looking for constant differences between the two. The same is true of the calyx tube and of the tip of the calyx, of the height of the stigma and of the anthers and so forth. Speaking broadly gigas is more compact than Lamarckiana ; and though its flowers do not exceed those of its parent species in number, they form a denser and therefore more beautiful head on the stem. The fruits of 0. gigas are very different from those of O Lamarckiana : they are half as long but about as stout. The seeds are on this account less numerous; but they are larger and heavier. Oenothera gigas is stronger than the other species in almost every respect. This is seen most strikingly in the girth of the stem as shown in Plates I and II and in figures 60 and 61. The stem is stronger right from its base, and for that reason grows more vertically upwards — a feature which greatly facilitates the recognition of the young plants. In the flowering region the diameter of the stem is almost twice as large as it is in O. La- marckiana, in which it is at most 5-6 mm. : in gigas it is often 10 mm. The whole stem is much more thickly beset with leaves and the leaves themselves are broader, more nu- merous, and more or less recurved. The great number of leaves is due to the shortness of the internodes ; in the Oenothera Gig as. 323 flowering part of the stem I found the length of the internode between two nearly ripe fruits to be barely 0 . 5 cm. Moreover the leaves are broader, the bracts bigger and, for this reason, the whole fruit-bearing spike less naked. Fig. 62. Full grown radical leaves in August to show the difference in breadth. L, Oenothera Lamarckiana; G, Oenothera gigas. I recognized the species, when I first saw it in my cultures in 1896, by its short stumpy fruits crowded to- gether and also by its cup-shaped flowers. 324 Origin of Each Species Considered Separately, Although ^z^a^-plants can be recognized before they flower it is difficult to give an accurate description of their leaves because they exhibit a high degree of indi- vidual variability — much greater in fact than do those of the parent species. The greater breadth is the chief difference; the length and general shape are about the same. The leaves, moreover, of gigas are more crumpled (compare page 310 and Fig. 62). But the breadth which is usually a matter of 4-6 cm. sometimes sinks to Fig. 6s. Oenothera gigas. A young plant in June, a few days before transplanting, (V2). 2 cm. without however destroying the characteristic look of the species. The leaves of the stem are usually set on a shorter stalk and are more deeply toothed than in O. Lamarckiana. The branches, of which a great many develop, remain sessile in the axils of the leaves as short densely foliate spikes which tend to make the foliage on the stem much thicker, just as in O. ohlonga (Fig. 71). The difference between the young rosettes in June (when they are usually planted out) is very striking. Oenothera Gigas. 325 The cotyledons are at that time still present; but dying off; or perhaps already dead. Figures 63 and 64 repre- sent two plants at this age reduced to the same scale (%). The gigas rosettes are compact, round and stout; the Lamarckianas are looser, their leaves have longer petioles Fig. 64. Oenothera Lamarckiana. A young plant in June, a few days before transplanting (V2) ; c, the cotyledon- ary leaves. and therefore make less use of the space of ground at their disposal.^ Oenothera gigas has only appeared once in my cul- tures— a single specimen in 1895. The event has already ^ Miss Anne M. Lutz discovered another highly interesting dif- ference between O. Lamarckiana and O. gigas The former has 14 chromosomes in its nuclei, like O. biennis and other species, but the nuclei of O. gigas have twice as many, viz., 28. Cf. Science, Vol. 26, Aug. 2, 1907, p. 151., (Note of 1908.) 326 Origin of Each Species Considered Separately. been described in § 3 (p. 227). It was evident that it was a constant species directly its seeds germinated. As Fig. 65. Seedlings of Oenothera Lamarckiana (L) and O. gigas (G). c, the cotyledons. Magnitied; the natural size shown in the middle. soon as they have acquired their first and second leaves the seedlings can be distinguished from those of the parent species with perfect ease (Figure 65). Their leaves are not only broader but more or less markedly cordate at their base. The latter character is gradu- ally lost in the succeeding leaves but the broader base persists for some time longer as a convenient mark of identification (Fig. 66). This character made it possible for me to demon- strate the constancy of the new species in the second, third and fourth generations (1897, 1899, 1900) without having to grow more than 20 to 40 plants to maturity. Fig. 66. Older seedlings of O. Lamarckiana (L) and O. gi^as (G). c, cotyledons, red. to "A- Oenothera Riihrincrvis. Z27 0. gigas has appeared twice again, but not directly from O. Lamar ckiana. It appeared once in 1898 from the seeds of a plant of O. sublinearis which had itself arisen from the Lamarckiana-iam'ily. It appeared again in 1899 from a cross made between 0. lata and O. Jiir- tella, a new species which did not arise from my mutating families but turned up among the seeds which I had bought. I succeeded in getting the first of these ^z^a^-plants to flower, but it was annual and did not flower till the beginning of October, too late for the seed to ripen. I therefore compared the plant very carefully with the other plants of that species which I had growing at that time, and which were raised from gigas-SQeds; it agreed with them in all essentials. The plant which mutated from 0. lata died as a rosette and never developed a stem. § 13. OENOTHERA RUBRINERVIS. This form unlike Oenothera gigas, is one of the com- monest of my new species. It has arisen, altogether, 66 times from O. Lamarckiana or from other families or cultures. It is hardly necessary to state that among the ancestors of these mutants, as far as I have had them under observation, there have been no examples of riibri- nervis. And, as the genealogical tables given above show, most of the mutations arose in families which had been under observation for many generations. The 66 mutated plants belonged to a single type. They did not differ more from each other than the members of a culture raised from the seeds of a single one of them. Each of the characters, which have al- 328 Origin of Each Species Considered Separately. ready been briefly described on p. 230 and will be treated of in detail shortly, was present on every individual plant; and, as far as investigated, the characters did. not differ from plant to plant. Once the young rosette is recognized its future pecu- liarities can be predicted, as in the example of the muta- tion of which Fig. 48 is a photograph, which was also photographed again when it was in flower (Fig. 49) (See pp. 280 and 282). I have often planted the mutants singly or in groups so soon as I have recognized them in order to be able to follow their further development during the course of the summer. It is very important to note that the various char- acters, the red coloring, the brittleness, the narrow leaves, the hairy appearance, and so forth, have never appeared separately. It is obviously out of the question that this association can be ascribed to chance, seeing that it has occurred in 66 cases. There can be no doubt that there is some sort of a connection between them. This conclusion receives strong support from the fact that even in the oft"spring of crosses the ruhrinervis characters remain associated, as I have observed in numer- ous cases. And exactly the same is true not only of the newly arisen mutations but of the offspring of crosses made with them. Every species has its "type" according to which its whole nature is altered; this ''type" affects its whole organization in such a way that hardly a char- acter or an organ is untouched by it. This hidden connection between characters which are invariably associated together needs an explanation. Two possibilities present themselves. First, it is conceivable that all these visible characters are only expressions of a single change, and that a mutation is brought about by Oenothera Rubrinervis. 329 the appearance of a single new elementary character. On the other hand it might be supposed that in mutation the elements of the species are changed by groups. There Fig. 67. Oenothera rubrinervis. An entire flowering plant, 1900. Fourth generation of a rubrinervis-iamily which arose in 1895 from seeds of O. Lamarckiana as shown in the pedigree on page 262, that is to say from the second generation of Lamarckiana in that culture. is abundant a priori and a posteriori evidence for the view that the characters of plants are often associated in groups in such a way that whole series of them react 330 Origin of Each Species Considered Separately. as one unit to external stimuli, and also that in hybridi- zation and other breeding experiments these characters behave as if combined into inseparable groups.^ If we should ultimately succeed in splitting up the group of riihrinervis-z\'\2ir2iQ.tQvs into its component units we should of course demonstrate its compound nature. But until this has been done it seems to me both simpler, and better in accord with the facts, to adopt the view that the sum total of the characters is the expression of a single elementary "unit." How it comes about that this ''unit" can make the walls of the bast-fibres thin, the leaves narrow and gray- green, the veins and fruits reddish, is a question which cannot be answered at present. But chemical combina- tions also possess many attributes the interdependence of which one is far from being always able to explain. I shall not go further into this question now ; but before I leave it I wish to insist on the fact that the whole so-called ''habit" of a species can be so much altered by a mutation that, during its whole life and in every organ it differs from the species from which it arose. If we refer to the pedigrees and tables of mutations given in sections 1-8 we shall find the cases of O. rubri- nerz'is which are recorded in the following table. In it we see that one 0. rithrinervis occurs in about every 1000 seedlings. Besides this, O. ruhrinervis arose twelve times in other cultures which were either lateral branches of the pedigrees referred to, or had arisen from crosses. I have summarized these in Table II. The proportion of 0. ruhrinervis to the whole num- ^ Intracellulore Pangenesis, pp. 21, 33, ec. • \ Oenothera Rubrmervis. 331 ber of seedlings will be seen to be much greater than in the first table. It amounts here to about 6 per thousand. But it must be mentioned that only those cases are in- INDIVIDUALS OF OENOTHERA RUBRINERVIS WHICH HAVE ARISEN BY MUTATION. I SOURCE YEAR ^ T u- S 1890, 1895 ) (J. I^amarcKtana . . . . -j loqc; icqy r A branch of the same family 1895, 1896 O. laevifolia 1889, 1894 O.lata 1900 O, oblonga 1897 O. Laniarckiana X O. nanella 1897 O. lata X O. naftella . . . 1895, 1900 O. Lamarckiana from the field 1889 SEEDLINGS TOTAL RUBRINERVIS 33,800 32 10,000 9 4 2,000 3 45 1 1,051 2 222 2 1 Total 54 INDIVIDUALS OF OENOTHERA RUBRINERVIS WHICH HAVE ARISEN BY MUTATION. II SOURCE O, Lamarckiana, a biennial culture O. lata which mutated from O Lam., first generation . O. lata X O. Lamarckiana O. lata X O. brevistylis . O. nanella X O. brevistylis O. scintillans X O. nanella O. Laynarckiana arisen from the cross O. Lam. X O. scintillans eluded in which the species in question actually occurred and that a figure which properly represents the proportion of 0. rubrinerz'is cannot be obtained without including YEAR SEEDLINGS TOTAL RUBRINERVIS 1897 164 2 1896 326 4 598, 1900 750 2 1896 266 1 1895 270 1 1898 95 1 1900 80 1 Total 1951 12 V 2>2>2 Origin of Each Species Considered Separately. all the cultures irrespective of whether they contained it or not. On tlie latter estimate the number would prob- ably sink to 0.1 %, if not lower. It has already been stated that 0. nibrinervis can be recognized as quite a young plant. Pans or boxes con- taining nothing but 0. nibrinervis can be identified very A' .A G'P^ Fig. 68. Seedlings of Oenothera ruhrinervis at various ages ; c, the cotyledons ; A, with the first two leaves at the beginning of May; A' , the natural size of the same. B, 14 days older. C , Rosette, towards the end of June, just before transplanting, from a pan in which the seed- lings were growing very close. Compare Fig. 64, p. 325 and Figs. 65 and 66, p. 326. early, a good deal earlier than mutants standing amongst other species (Fig. 48 on page 280). But in these the narrow leaves with their red veins and gray felt-like surface, the almost complete absence of crumples and the brittleness especially of the stalk, clearly distinguish this form from 0. Lamar ckiana and the rest, (Compare Oenothera Riibrinervis. 333 Fig. 68 with the similar ones of O. Lamarckiana Figs. 64-66). The narrow form of the leaves is well brought out in Figs. 52 and 54. The older the plants become the greater becomes the difference and the more certain the diagnosis. As a rule I have removed the mutants at an age when they have about twice as many leaves as the rosette figured at Fig. 68 C. The plants shown there are, of course, not mutants but are raised from seeds of 0. ruhrinervis and selected from the crop as the most typical examples of that species. As the plants get older the veins of the leaves lose their pale red color more or less ; but this depends on how they are grown and on the amount of sun they get. On the other hand, with age the red pigment be- comes more evident in the inflorescences, the flowers and the unripe fruits, thereby contributing greatly to the characteristic look of the species. The young internodes of the stem are suffused with red, and this color is par- ticularly pronounced in the swollen bases of the larger hairs. The tips of the calyx are spotted with red and the flowers become much darker when they wither than those of O. Lamarckiana, reminding one of those species the leaves of which become red when they wither, such as O. stricta, 0. missouriensis, and particularly the white O. acanlis. The fruits are marked with four broad, dark red, longitudinal stripes, one along the middle of each valve. But in this case the redness varies accord- ing to the position of the fruits and from individual to individual within apparently wide limits; sometimes, in- deed, the stripes are very difficult to find. This red pigment occurs also in 0. Lamarckiana and particularly in the unripe fruits ; but very indistinctly : 334 Origin of Each Species Considered Separately, whilst in 0. nihrinervis the red stripes are handsome and striking. To turn now to the general structure of the plant; it has a greater tendency to develop lateral branches from the main stem and, in connection with this fact doubtless, fewer from the rosette (compare Figs, 49 and 67 with Fig. 55). But this feature is greatly affected by the manner of cultivation. Fig. 69. Oenothera ruhrinervis. A, Transverse section of the stem ; m. pith ; p, inner phloem ; h, wood ; b, bast bun- dles between the outer phloem and the bark ; B, part of such a bundle highly magnified; C, the same of O. La- mar ckiana. This form can be distinguished at some little distance, from the common Evening Primrose by the general habit of its inflorescence and flowers ; but it is very difhcult to find differentiating characters which can be described. Plate I might, if it did not lack red pigment, pass as well for a ruhrinervis as a Lamarckiana (see Fig. 43, p. 232). The whitish gray color, which is seen in a more pro- nounced state in O. alhida, is not really due, as it appears to be, to the greater hairiness of the plant ; but is brought about by the swollen surfaces of the cells of the epidermis Oenothera Ruhrinervis. 335 which have not grown out in the form of hairs. This swelhng is very shght in 0. Lamarckiana. It has already been stated in § 3 of this Part that one of the most characteristic features of 0. ruhrinervis is the brittleness of its stem. The latter as well as the petioles of the leaves are very fragile and break off at the merest touch. The cause of this is the weak develop- ment of the hard bast. It is only biennial plants or very strong annual ones that break in late autumn in the way that 0. Lamarckiana does when the hard bast is torn. A transverse section^ of the stem of a plant about a meter high, taken in August, shows the bast-fibres in a discontinuous ring around the outer side of the wood and inside the bark, as shown in Fig. 69 A. If we com- pare such a section with a similar one of 0. Lamarckiana we do not at first see any difference. In both plants the sclerenchymatous strands are about equally developed. But if we examine a single strand under a higher power we find that in O. Lamarckiana it is better developed in the radial and less so in the tangential direction than in 0. ruhrinervis. The most important difference however lies in the thickness of the walls of the individual cells which, as Figs. 69 B and C show, are about half as thick in O. ruhrinervis as they are in the parent species. There is great difference between individuals in re- spect to this character depending on whether they have plenty of room to grow in, or are crowded together; or whether they are sown early or late. Weak plants never entirely lack the bundles, though the individual cells of the bundles are fewer and more tangentially arranged. They often retain these characters until they ripen their fruits. ^ For the general anatomy of the stem see Francis Ramsay, On the Stem Anatomy of Certain Onagraceae, Minnesota Botanical Stud- ies, Bull. No. 9, Nov. 1896, p. 674. 336 Origin of Each Species Considered Separately. In late autumn there appears on the inner side of the sclerenchym ring a thin layer of cork which must of course have been laid down much earlier and possibly stands in some causal re- lation to the external characters of the plant. This species is char- acterized by an apparent inability to stretch its stem — so to speak — which is particularly no- ticeable in weak plants. This character is, in all probability, due to the weakness of the bast- fibres we have just de- scribed. Fig. 70 repre- sents a young plant grown in a pot, about the beginning of July, and illustrates this feature very well. The stem is not straight but bent in a zig-zag fashion; in such a way that the bends occur at the nodes and the leaves are in- serted in their outer con- vex sides. These bends do not straighten out with subsequent growth ; in fact they are often even more pronounced on the fruiting plants. The stronger the stem is, the less is this character developed; but I have, Fig. 70. Oenothera rnhrinervis. — Young annual plant, 30 cm. high, about V3 natural size. To show the zigzag course of the brittle stem. Oenothera Ohlonga. ZZ7 nevertheless, found it on perfectly healthy annual plants whose main stems have been heavily laden with fruit. I have already recorded experiments on the con- stancy of O. rtihrinervis § 3 (p. 232) and § 5 (p. 274). These experiments show that plants raised from the seed of mutated individuals are exactly like their parents, and that the characters we have described for the parents reappear in the children in exactly the same degree. O. ruhrinervis itself is very slightly mutable and seems to confine itself to throwing off lata and leptocarpa, as al- ready shown on page 273. § 14. OENOTHERA OBLONGA. (Plate VI.) 0. ohlonga has arisen much more frequently than O. rnbrinerz'is both from 0. Laniarckiana itself and other species and crosses. I have seen it arise altogether about 700 times from one form or another of known and pure ancestry. The various cultures in which it arose comprised about 70,000 seedlings. We might therefore almost speak of coefficients of mutation ; which in the case of this species would be about 1 '^/c, in the case of O. ruhrinervis 0.1 % and in that of 0. gigas 0.01 %. What is the cause of these differences? They cannot be ascribed to defective observation. I first saw 0. oh- longa in 1895 when my cultures were very extensive; in previous years they were probably there, but escaped my observation. Their young rosettes are as easy to recog- nize as those of O ruhrinervis; and often somewhat earlier, as rosettes with six leaves. But evidently this 338 Origin of Each Species Considered Separately. fact cannot explain the difference in the mutation-co- efficients. Fig. 71. Oenothera oblonga. Upper and middle section of a plant in September to show the peculiar type of branching with rosette-like lateral branches, (Compare Fig. 67 on page 329.) Reduced to Vs natural size. The smaller figures similarly reduced, a, a flower ; a petal is removed and shown separately at b; c, a flower without the corolla, showing the stamens which are bent down- wards at the base but upwards again towards their tips, and the style with the four stigmata; d, ripe fruits; e, one of their bracts. These differences in the "mutation coefficients" hold good, too, of the individual families. Not exactly of course, but to such an extent that in large sowings the Oenothe?'a Oblonga. 339 ohlonga mutants are almost always considerably more numerous than the ruhrinervis ones. The observations extend over six years (1895-1900), which is probably only a small section of the whole mutation period. Never- theless, the evidence seems to justify the conclusion that the various new species arise from the parent form, at any rate for a certain period, in definite and constant proportions which vary from species to species. This consideration seems to me to lead to two im- portant points. First, the probability that 0. Lamarckiana is able to produce other mutations in even smaller pro- portions, such as one in a million; in which case there would not be much chance of their appearing in my cul- tures. In other words if one could make the whole ex- periment ten or a hundred times as extensive, one would be very likely to get more mutations and amongst them, possibly, some better than those which have already ap- peared. 0. laevifoUa and 0. hrcvistylis might then arise again. The second point relates to the causes of these pro- portions. Is it possible to interfere with and alter the "mutation coefficients" ? Is there any hope of increasing the proportion of the rarer species?^ And when a method of doing this will be invented will it be possible to obtain mutations which are at present presumably too rare to appear ? An experimental study of the process of mutation during the mutation period may even put into our hands the power to bring about the inception of such a period ; or in other words the power to make an immutable spe- cies mutable. ^ See the case in § < (on page 264) where, as a result of defective germination the proportion of mutants arose to 40 %. 340 Origin of Each Species Considered Separately. But let us return to the figures which seem to justify these hopes. I shall first give the values for the chief families and then those for their lateral branches. INDIVIDUALS OF OENOTHERA OBLONGA WHICH HAVE ARISEN BY MUTATION. I A. FROM 0. LAMARCKIANA. Main family . . . 1895 14,000 176 1.3 ... 1896 8,000 135 1.7 A collateral family 1895 10,000 69 0.7 Biennial culture . 1897 1,660 31 1.9 Totals 33,660 411 1.2 B. FROM 0. LATA. Lafa-fa.mi\y . , , . 1900 2,000 7 0.3 Z.a/tz-cultures . 1895-1898 2,350 28 1.2 Totals 4,350 35 0.8 C. FROM O. NANELLA. O. nanella . . . 1897 760 1 0.1 Although the /a/a- family of 1900 is considerably be- low the average, the percentages in groups A and B con- form pretty closely to a general proportion of about 1 % ; whilst the figure for O. nanella affords a good example of the rule that new species mutate less than 0. La- marckiana or than 0/ lata fertilized with Lamarckiana pollen. About the same proportion is maintained in crosses. The following table embodies results already described, together with some to be referred to later on. Oenothera Ohlonga. 341 DATE TOTAL OBLONGA . 1897-1899 8283 38 is 1898 293 4 . 1895-1900 1586 14 . 1895-1899 498 6 1895 127 4 . 1895 1500 4 1898 95 3 1896 30 2 fis 1897 200 8 Totals 12,612 83 — 0.7 % INDIVIDUALS OF OENOTHERA OBLONGA WHICH HAVE ARISEN BY MUTATION. II FROM CROSSES. SOURCE O. Lamarckiana X O nanella O. Lamarckiana X O. brevistyli O. lata X O. nanella .. . . O. lata X O. brevistylis ■. . O. lata X O. laevi folia . . O. rubri7iervis X O. nanella- O. scintillans X O. nanella , O. Lafnarckiana X O. bietmis O. Lamarckiana X O. suaveolens To obtain the above figures I have sometimes re- corded the seedhngs when they had from 6 to 8 leaves but at other times later, according to the different years and various other circumstances. In many cases I have transplanted them in order to observe them during the whole summer. Fig. 72 gives an idea of the stage at which the seedlings were recorded, and may be com- pared with the parallel figures for the /a^a-families (Plate IV and Fig. 48, p. 280). We are concerned in Fig. 72 with a culture of O. Lamarckiana which was sown on the 14th of March 1900 and transplanted into wooden boxes on the 14th of April. The seeds had been harvested in 1895 from three plants enclosed in parchment bags to insure pure self-fertilization. Of the 188 seedlings raised, 4 were mutations of which two were albida and two ob- longa. By a lucky chance an example of each of the new forms stood quite close together; so that I was able to include them in the same photograph (Fig. 72). The 342 Origin of Each Species Considered Separately. plants are arranged in rows in the boxes : 0. albida can be recognized at once in the middle of the figure by its small size; just underneath it is the 0. ohlonga which can hardly be distinguished in the figure. I transplanted these two plants on to a separate bed to watch their further development. They grew up to strong rosettes which exhibited all the characters of the species to which they Fig. 72. A mutation in a culture of O. Lamarckiana. Ori- gin of O. albida and O. ohlonga. From a photograph taken at the end of May 1900. In the middle of the middle row is the little O. albida ; in the middle of the lower row O. ohlonga. The other seedlings are O. La- marckiana ; Vs natural size. belonged, clearly and beautifully; but were destroyed in the autumn by the caterpillars of Agrotis segetum. Seedlings of even less than 6 or 8 leaves can often be recognized ; but it is very difficult to describe the characters which render their identification possible. In Fig. 73 at A 3. very young seedling is shown with its Oenothera Oblonga. 343 first two leaves, and at 5 a rosette two months old. These are not mutants but plants grown from the seed of self- fertilized oblonga. These cultures came quite true to seed and exhibited a high degree of uniformity. The first two leaves, after the cotyledonary ones, are broad and with broad bases, markedly broader than those of O. Lamarckiana at the corresponding age (Fig. 65 L, p. 326). This can be seen in Fig. 72> A and 5 at 1 and 2 as well as in Fig. 72. There soon follow narrower leaves but the rate at which this decrease in breadth takes place is not constant. Fig. 72) B is more typical in this respect than the O. oblonga of Fig. 72, but the remaining seedlings Fig. yz- Seedlings of Oenothera oblonga. A, a few weeks old, magnified (2.5/1). B, two months old, 'A natural size, c, cotyledonary leaves. In B the leaves are marked in the order in which they appeared. in the same box behaved essentially in the same manner. I photographed many of them at the time, but do not think it worth while to reproduce the others as well. As the plants grow their characters become more pronounced, the leaves longer and narrower, the veins broader, paler and more prominent. In the third month growth takes place much faster, or at any rate produces a more noticeable increase than in the first two. At the end of that period the rosettes possess many leaves and are very strong and ready for the development of the stem (Fig. 74). If they do not do this they grow to a 344 Origin of Each Species Considered Separately. considerably greater size during the summer and increase the number of their leaves without, however, altering their general appearance. A very typical leaf with its venation will be found in Fig. 54 (See p. 295). If the rosettes develop stems in June and July they flower that summer. Such plants are very uniform in character, they are slender and yet firm, and branched either very little or not at all. Hardly any differences were discernible between 200 flowering plants growing Fig. 74, Oenothera ohlonga. A rosette with radical leaves at the end of June. together. The tallest plant began to flower when it was 60 cm. high, and flowered till the end of September when it had attained a height of one meter. It had a single lateral branch only 10 cm. long and with only two flow- ers on it ; all that there w^as in the way of lateral branches besides this was a series of rosette-like offshoots from the axils of the leaves along the middle region of the stem. The result of this is the very characteristic ensemble Oenothera Oblonga. 345 (Fig. 71 B) which is always found in all the plants of this species. The culture in question did not contain a single plant bearing flowers on a lateral stem, with the exception of the plant referred to. At the time when the plant is about to flower (Fig. 44, p. 233) the flowering spike is still densely clothed with leaves. Higher up, the bracts become shorter. The fruits likewise do not attain the size of those of 0. La- marckiana ; and we get in this way another very striking character which can be well seen by comparing Plate VI with Plate I. The ripe fruits hardly attain a third of the length of those of O. Lamarckiana. As a result of this, the seeds are often bad and developed in very small quantity so that all that can be hoped for is a very meagre harvest at the best. Biennial plants are much better in this respect; they are more robust and bear numerous, strong, though small, fruits which contain an abundance of seed. These fruits are not much longer than those of the annual plants but much stouter, rather like those of 0. lata. When grown under more favorable conditions the annual as well as the biennial plants develop a certain number of lateral stems from the axils of the radical leaves, such as have already been figured in the case of a mutation from the /a/a-family (Fig. 50, p. 284). But, even so, the main stem itself remains unbranched, a pecu- liarity which can best be seen by comparing such a plant with O. nihrinervis (Fig. 49, p. 282). There is not much to be said about the flowers and buds of O. oblonga (See Plate VI). They have the same form as those of the parent species; but in cor- respondence with the greater delicacy of the whole plant thev are a trifle smaller. 346 Origin of Each Species Considered Separately. I first harvested the seed of O. ohlonga in 1895; but as the plants had flowered too late they had not been artificially fertilized and the seed only gave quite a small percentage of ohlonga. But in 1896 I harvested self- fertilized seed partly from annual and partly from bi- ennial plants. The plants were all mutants, that is, had all arisen from 0. Lamarckiana of pure strain, in fact from the main line of descent of the Lamarckiana-iamily itself (See the genealogical table on p. 224). The bi- ennial plants had therefore three generations of pure Lamarckiana behind them; and the annual ones four. There were seven plants in the first and twelve in the second group. I sowed the seed in the middle of April 1897, and as the seed had not been sown too thick the seedlings dis- played their characteristic features as early as the middle of June, in the seed pans. The leaves with the exception of the first two broad ones (p. 342) were narrow and had long petioles; and exhibited the characteristic broad pale midveins. A comparison with cultures of the or- dinary Evening Primrose at the same stage of develop- ment made it certain at once that there were no La- marckianas among the ohlonga crops. The sowings were perfectly true, with the exception of four seedlings one of which became a ruhrinervis, one an elliptica and two alhida (p. 297). Besides this, one plant had a pitcher shaped leaf. I counted the seedlings for 17 out of the 19 seed parents separately; and, as I have already stated above (p. 235), they were all, with the exceptions named, ohlongas (1683 and 64; together, 1747 plants). In the same year I sowed the self -fertilized seed of three other mutants which had arisen, in order to find out whether the difference in their origin would cause Oenothera Ohlonga. 347 them to differ from the other mutants in the matter of constancy. I will describe the ancestry of the three plants separately. The first arose from the laevifolia-i^xmly, whose pedigree has been given on p. 273, in the following way : 1895-1896 O. oblonga I 1894 O. rubrinervis I 1893 O. rubrinervis y 5th gen., X O. natiella, 4th gen. Similar intermediate generations 1889 O. rubrinervis O. fianella 1888 O. Lamarckiafia I 1886-1887 O. laevifolia from Hilversum The first three generations in this pedigree as well as the rubrinervis of 1893 have already been referred to on p. 273. The 0. nanella w^as biennial in 1889, but since then has been annual. In the summer of 1893 I pollinated some castrated flowers of O. rubrinervis with the pollen of 0. nanella; in 1894 I obtained from these seeds some ordinary rubrinervis plants, which were fer- tilized with their own pollen and produced mainly O. rubrinervis. But amongst them were four oblonga (p. 341 ) of which I managed to bring one safely through the winter as a rosette. It flowered in 1896 in a parchment bag; 16 seedlings were raised from it and were subse- quently transplanted into pots ; they were all oblonga. The second oblong a-muidLni arose likewise from the laevifolia-fsimWy. It arose from a self- fertilized La- marckiana plant, in the main line of descent in 1894, a plant which had itself therefore arisen partly from laevi- folia and partly from Lamarckiana stock, but which, at 348 Origin of Each Species Considered Separately. least so far back as 1896, had ancestors of no other type. This ohlonga belonged to the 9th generation of the fam- ily. It was biennial ; was self-fertilized and gave rise to 297 seedlings. Amongst these I found a single 0. albida ; all the rest were 0. ohlonga. The third plant belonged to a lateral branch of the Lamarckiana-family. In the pedigree on p. 224 five lata plants will be found for 1888. One of them which had a fine ascidia flowered in 1889; but its seeds were not sown till 1894. This second generation was annual and left to be pollinated by insects. In 1895 I raised from its seeds 128 Lainarckiana, 18 lata, 3 nanella and 10 oh- longa. Of the latter one plant flowered in its second year, i. e., in 1896, in a parchment bag. Of its seeds 91 germinated and gave rise solely to ohlonga plants. These experiments show that the constancy of O. ohlonga, when it arises as a mutant, is independent of the character of its ancestors. These may ht Lamarckiana, laevifolia, rnhrinervis, nanella, pure or hybrid, but the ohlonga which arises from them is always pure from the first generation; except, of course, that it has inherited the mutability of its parent and has the capacity for giving rise to other types (alhida, rnhrinervis). The total number of plants recorded in the experi- ments of this year is 1747+16 + 297 + 91=2151. I have sown seeds of the same mother plants in subse- quent years, in 1899 and 1900 and always with the same result. I have so far not harvested any seed of the second generation because although the plants flowered freely they were all annual and so produced only imper- fect fruits. Oenothera Alhida. 349 § 15. OENOTHERA ALBIDA. (Plates III and IV.) A beautiful but delicate species which is very slow in growing as a seedling and is for that reason very easily recognized. See Fig. 48 on p. 280 and Fig. 72 on p. 342. These weak plants are at a great disadvantage when growing among the much stronger seedlings of the parent type; and it is only very rarely that I succeeded in getting them to flower. It was in 1895 (as has already been stated in § 3, p. 229) that I first succeeded in getting a rosette to sur- Fig. 75. Oenothera alhida. Young seedlings; A, with the first two leaves ; B, two months old. vive the winter ; it was on it that I first saw the flowers of the new species, but I got no seed from it. Before this I had seen alhida almost every year and in no in- considerable numbers, but thought they were merely sickly individuals and had taken no further account of them. That is why the records that follow are confined to the period 1895-1900. The young alhida plants are so delicate that it is only by the exercise of the greatest care that they can be kept 350 Origin of Each Species Considered Separately. aliv'C. I never found them at Hilversum; and even if they had ever succeeded in germinating there, they would most certainly have perished before developing a stem. This was exactly what happened in my experimental garden from the time my experiments began until 1896. These facts moreover show, as mentioned above (p. 229), that my first alhida mutants could not have had similar individuals in their an- cestry, neither as pollen parents nor as seed pa- rents. Even when 0. al- hida has set seed the difficulty of getting the seed to germinate is considerable; but the attempt to keep the young plants strong from the very begin- ning has succeeded. Some of them always remain weak and look just like the young mutants, others bear broader leaves and gradually grow to little rosettes .which are apparently just as strong as those of O. ohlonga at a like age (see Fig. 75). Moreover they differ from these very little in form at first (Fig. 72)). But their color is always, as their name implies, a whitish gray. For the first six weeks of their existence the leaves of these two species are about the same breadth; those of O. alhida however are a little blunter at the Fig. 76. Oenothera alhida. Young plant, 3V2 months old. Oenothera A lb id a. 351 tip. During the growth the leaves of the rosettes in- crease in breadth as a rule (Fig. 76) whilst those of the stem become narrow again (Fig. 54a on page 295). It is always easy to recognize albida mutants by the characters I have described. I have cultivated many of them beyond this stage, especially in 1895 and the fol- lowing years, in the hope of getting them to flower and set seed. And in this way I had ample opportunity of testing the accuracy of my diagnosis. The ease with which this species can be recognized as quite a young plant makes it a convenient one for the study of the relative frequency of its origin from O. Lamarckiana and other species. The result of this in- vestigation was that this frequency, this coefficient of mutation, turned out to be very different in different cases and to be subject to even greater fluctuations than those exhibited by the three species described above (0.01 % for O. gigas, 0.1 fo for rnbrinervis and 1 % for oblonga). The two tables that follow bring this out. I include in them figures that have already been given in §§ 2-5. INDIVIDUALS OF OENOTHERA ALBIDA WHICH HAVE ARISEN BY MUTATION. I SOURCE DATE TOTAL SEEDLINGS ALBIDA % ALBIDA O. Lamarckiana-isimWY . 1895-1899 28,500 56 0.2 O. Lamarckiatia, plants from crosses .... 1898 4,599 2 0.05 A lateral branch of the Lamarckia7ia-isim\\Y . 1895 10,000 255 2.5 O. lata 1900 2,000 42 2.1 O. lata 1896-1899 751 31 4.0 O. Lamarckiana, biennial 1896 164 15 9.0 1897 1341 1 0.1 1895-1900 1535 15 1.0 1900 IS-U 37 2.0 1900 636 2 0.3 1898 95 3 3.0 1900 743 13 2.0 352 Origin of Each Species Considered Separately. The proportion of albida mutants varies between 0.05 % and 9 %. This variabihty is also exhibited though to a con- siderably less extent by the proportions in whicli O. al- bida occurs among the offspring of various crosses. INDIVIDUALS OF OENOTHERA ALBIDA WHICH HAVE ARISEN BY MUTATION. II FROM CROSSES. CROSS DATE TOTAL O. ALBIDA % ALBIDA O. Lamarckiana X O. nanella O. lata X O. nanella . O. lata X O. rubrinervis . O. lata X O. scintillans . O. scUitillans X O. nanella O. lata X O. suaveolens . The mutants obtained in these two series of experi- ments amounted to 472 and agreed in all their charac- ters so far as these could be investigated. Flowering plants of 0. albida can be distinguished from all other subspecies of O. Lamarckiana and from this species itself just as easily as the seedlings and rosettes can. They do not attain even in late autumn a height of more than one meter; but as a rule they give rise, about the middle of their stem to a group of flowering branches, in the same way that O. rubrinerzns does. Their leaves are narrow (Fig. SAA, p. 295), pointed and very uneven; the crumples in them being more numerous and more pronounced than in the parent species (Fig. 57, transverse section of a leaf, see p. 310). The flowers are always somewhat smaller than those of Lamarckiana, as would be expected from, a greater delicacy of the species ; moreover they have a tendency Oenothera Leptocarpa. 353 to stand more upright, and not to open so wide as those of the parent species (compare Plate III with Plate I). In other respects they have the same structure, and the stigmas stand up well above the anthers. The color of the flower is a paler shade of yellow. The fruits do not attain the length or the stoutness of those of 0. La- marckiana, and as a rule set little seed. The gray color, which, like that of 0. rttbrinervis is not due to increased hirsuteness, but to the swelling of the outer wall of the ordinary epidermis cells, ex- hibits a high degree of individual variability, sometimes indeed to such an extent that doubt may arise as to the proper diagnosis, a doubt which however can always be dispelled by the examination of later stages. § i6. OENOTHERA LEPTOCARPA. The foregoing examples have shown us that muta- tions arise from Oenothera Lamarckiana in proportions which vary from about 1 % to less than 0.1 %. We have further seen that the same mutations recur regularly in the same mutation period. It follows from this that a careful study of such a period wnll soon reveal the commoner mutants which the species in question is producing. Then we have to look for the rarer ones ; and for this purpose much more extensive sowings must be made. If these rare mutations can be recognized as seedlings or at any rate as young rosettes, all that we have to do is to sow seed on a large scale, transplant any seedlings which exhibit any abnormality and throw away those which have not mutated. If this method is adopted 354 Origin of Each Species Considered Separately. many thousands of plants can be accommodated on a few square meters of ground up till the time when the mu- tants have to be planted out. But if the characters do not develop whilst the plant is young, it is a very different matter indeed. In this case 40 to 50 is the maximum number of plants that can flower on a square meter, and that is a high estimate. When this is the case the plants must be cultivated on an enormously extensive scale before we can enter- tain the smallest hope of seeing mutations. We become dependent to a large extent on chance as in the case of the first appearance of O. gig as. It is obviously for reasons of this kind that practically all my mutations are recognizable as seedlings, whilst the two new species which are found at Hilversum are indistinguishable in their young stages from young 0. Lamarckiana. Oenothera leptocarpa is the only exception to this rule, at least it is the only one among those which have arisen from the pure stock of O. Lamarckiana. Amongst the crops raised from crossed seeds there were occa- sional instances ; but it is often difficult in these cases to distinguish mutations from the ordinary products of crossing. 0. leptocarpa cannot, even in pure cultures, be dis- tinguished from O. Lamarckiana either as a seedling or as a rosette or even at the period when it is first devel- oping its stem. I have once or twice transplanted sup- posed mutations as young plants and found them to be O. leptocarpa. But as a rule I have not recognized them until just before they flowered. For these reasons little can be said with certainty about the frequency with which this form appears. The Oenothera Leptocarpa. 355 origin of two examples of this species will be found recorded in the pedigree of a branch of the Lamar ckiana- family which appears on page 262. It occurred in a culture of about 10,000 seedlings in 1895, of which about a thousand flowered. This indicates a frequency of about 0.2 %. I have noted the appearance of single individuals both before and after this, but have not the data from which to calculate the frequency of their oc- currence. 0. leptocarpa has arisen from O. riihrinervis, as well as from 0. Lamar ckiana, but not from any other new species. After I had become familiar with this fact I found them fairly frequently. A character peculiar to 0. leptocarpa is that it flowers very late — a character which has greatly diminished the prospect of discovering the species. For as soon as the plants on a bed begin to flower, seed-parents are chosen for self-fertilization. And I generally remove a certain number of the plants surrounding these either to allow them to grow more freely or to have plenty of room for my operations. This involves the destruction of the weaker plants with which the late flowering leptocarpa are easily confused. The riihrinervis cultures were usually made with a rather special end in view and therefore consisted of no more plants than were wanted as seed parents. The plants were grown for example for the purpose of mak- ing certain crosses which I had decided upon or for the experiments with tricotyly which I have already men- tioned, and in these latter of course selection took place directly the seed came up. So that the proportion in which leptocarpa appears in such cultures is no indica- tion at all of what the mutation-coefficient of that species really is. 356 Origin of Each Species Considered Separately. For example in 1895 I found 5 leptocarpa amongst 44 tricotylous nihrinervis (p. 273) and in 1897 I found one amongst 20. In 1900 I had planted out 24 apparent Lamarckiana plants which had sprung from a cross which I had made in 1899 between O. nihrinervis and O. nanella. When they came to flower, however, it turned out that only half of them were really Lamarckiana and that the rest were leptocarpa. In the same year two leptocarpa appeared among the 90 offspring of a cross between 0. nihrinervis and 0. Lamarckiana. The most characteristic features of O. leptocarpa are its late flowering and its long slender fruits. The late flowering is not the result of arrested growth for the plants are just as strong and as tall as others when these are about to flower; but it is due to the fact that after they have reached this stage they continue to grow vegetatively for some weeks to come. The first flower- bearing node is therefore considerably higher in lepto- carpa than in other forms and the spike of flowers standing well above those of the other plants on the bed enables us to detect the species immediately. The stem is, moreover, rather flaccid so that the flowering spike hangs over to one side. The flowers and buds do not differ in any essential feature from those of 0. La- marckiana; the buds are, just before they open, a slightly brighter green with less yellow in them. The fruits and bracts on the other hand are quite different. The bracts are broader at their base, more triangular and more flattened, whilst those of 0. Lamarckiana are often more or less bent and wavy along the midrib. They are pressed much more closely against the stem which they almost completely enclose in a mantle, as it were: instead of hanging down they stand up fairly straight. Finally Oenothera Leptocarpa. 357 they are covered with numerous small pits which tend to alter the color of the leaf. The fruits are long and thin, and therefore quite dif- ferent from those of O. rubrinervis. They seldom ripen because the plant flowers so late. In November 1896 I measured the length and breadth of the first five ripe, or at any rate full grown, fruits on a number of plants of 0. leptocarpa which occupied two beds. I divided the breadth by the length and employed the quotient as a measure of the thickness. The values which I got were quite definite ; the mean thickness lay between 1 5 and 1 7 whereas that of Lamarckiana ranges between 22 and 24. Thus we see that the carpels of O. leptocarpa are about % as thick as those of the parent species. The following are the values which I obtained : BREADTH — TJITATT^TrT? 1896 OF INDTVIDTIAT S LENGTH — ' J. 'I KJ A B 12 0 1 13 1 3 14 5 3 15 6 8 16 11 2 17 15 3 18 13 5 19 3 1 20 5 1 21 2 0 22 2 1 23 1 0 24 0 0 Totals 64 28 mean 17 15 The culture A was from the seeds of the two lepto- carpa mentioned on p. 262; B from an artificially self- 358 Origin of Each Species Considered Separately. pollinated individual in a parallel culture in the same year. It only remains to be stated that in these two cultures the 0. leptocarpa came true to seed. In fact this can be clearly seen from the above two columns of figures; the individuals measured obviously form a definite group, although their curve of individual variability naturally overlaps that of O. Lamarckiana whose mode is at 22-24. The curves are transgressive, as is so often the case in closely allied species.^ The thickest fruits of O. lepto- carpa are thicker than the thinnest of 0. Larnarckiana, so that if we only had this character to go by we should sometimes be unable to distinguish the tw^o species. Culture A consisted of 300 and B of 150 plants and all of them, even those which did not ripen their fruits exhibited the characters peculiar to O. leptocarpa with the single exception of two O. nanella. Some of the plants were perfectly pure leptocarpa whilst others approached the characters of the parent species to a certain degree. For, all the characters of the species exhibit individual variability just as we have seen the thickness of the fruits to do. In spite of this transgressive variability the constancy of the new species was proved by the cultures. There was no real reversion. § 17. OENOTHERA SEMILATA. This species has only appeared thrice in my cultures ; and every time from 0. lata. One appeared in 1894, the other two in two independent cultures in 1895. They looked very much like 0. lata except that the characters ^ On this point see § 25 of this Part. Oenothera Semilata. 359 of that species were only slightly developed. Hence the name semilata. The 1894 plant was broken in a storm. One of the 1895 ones flowered well but at first set no fruit. It was not until November when it had attained a height of 2 meters that some good fruits were de- veloped, but the oncoming winter prevented the ripen- ing of the seed. I was more fortunate with the third plant. It had arisen in 1895 from the first lata-family; and had there- fore O. lata as mother and grandmother, and 0. La- marckiana as father and grandfather. See the pedigree on p. 285. At first they only differed but little from the real lata of the same culture, the buds were however slightly thicker, the inflorescence looser and longer, the leaves narrower and slightly more rounded at the tip. But when the flowers opened it was found that the anthers produced apparently good pollen although not so much as is produced by O. Lamarckiana. I then enclosed the plant in a parchment cover and selfed the flowers. I also pollinated two pure latas with the pollen of this plant. The pollen proved to be quite good, for in both cases the plants yielded a good harvest of seed. I sowed the self-fertilized seeds of the semilata plant in 1897. The resulting culture consisted of 276 plants which flowered and 82 which did not. There occurred amongst them three dwarfs (O. nanella), three lata plants which flowered, and a rosette which evidently be- longed to the same species. The nanella were obviously mutants, the lata either this or perhaps reversions. The remaining plants clearly exhibited the characters of semi- lata and justify the establishment of this form as a con- stant species. But I did not consider the experiment important enough to continue. 360 Origin of Each Species Considered Separately. The above mentioned cross 0. lata X 0. semilata did not give any particularly remarkable result; amongst 105 seedlings there were 39 lata, 2 nanella, 2 ohlonga^ and 1 alb id a, whilst all the rest were 0. Lamar ckiana. These forms and the proportions in which they occur are the same as those which 0. lata produces when crossed with other species. These figures give little support to the supposition, which is improbable on other grounds, that O. semilata is a hybrid or perhaps an intermediate form between 0. lata and 0. Lamarckiana. § i8. OENOTHERA NANELLA. (OENOTHERA LA- MARCKIANA NANELLA.) In view of the great importance which attaches to a satisfactory distinction between species and varieties it seems worth while to go a little closely into the differ- ence between Oenothera nanella^ and the other new spe- cies.^ The new species, other than nanella, which have arisen in my experimental garden find no analogues either in other species of the same genus or anywhere else in the vegetable kingdom. Each constitutes a new and distinct type and is, without question, to be regarded as an elementary species. Varieties are distinguished from these in popular '^Oenothera nanella, or the Dwarf Evening Primrose, often called the dwarf for short, is a constant form. The term dwarf is often used to signify the smallest individuals, presented by fluctuating variability, which are of course of an entirely different nature. For information on such dwarfs see P. Gauchery, Recherches sur le nanisme vegetal, Ann. sci. nat. hot., 8 Serie, T. IX, 1899, pp. 61-156; and also D. Clos, Du nanisme darts le regne vegetal, Acad. Sciences Toulouse, T. XI, 1389. ^ For further details see the second volume of this work. Oenothera Nanclla. 361 estimation first as being derived forms and secondly by the supposed fact that they do not come true to seed but from time to time revert to the type of the species. This latter view has long been shown to be baseless ; for many varieties are just as constant as the best species. Varieties are re- ally distinguished by the fact that the same vari- ation recurs in a great number of species and genera. The type is not new but appears under a variety of forms. Let us apply this to our dwarf Oenothera. Dwarf varieties are as numerous as, for exam- ple, glabrous ones. The following are some well- known examples, Tage- tes patiila nana, Tagetes signata nana, Scabiosa atro purpurea nana, Pap- aver somniferiiin nanum, Dianthus caryophylliis nanus, Dianthus barba- tus nanus, Cheiranthus cheiri nanus, Matfhiola incana nana, Calliopsis bicolor nana, Cuphea purpurea nana, Impatlens Balsamina nana and many others.^ Most of them are very popular garden-flowers. * See List in Carriere, Production et Fixation des Varietes, p. lO. Fig. yy. Oenothera nanella. Entire plants with flowers and almost fully grown fruits. Va nat. size. 362 Origin of Each Species Considered Separately. From the systematic point of view therefore our dwarf should be called Oenothera Laniarckiana nana or as it is particularly small, 0. Lam. nanella. But from the experimental point of view it behaves just like the other elementary species ; for it is, as already stated in § 3, absolutely true to seed. And as the name O. nanella cannot refer to anything else I shall usually employ it.-^ If we look a little more closely into it we shall find other grounds for regarding our dwarf as an elementary species. In the first place it is by no means a miniature edition, as it were, of 0. Lamarckiana. Fig. 78. Oenothera nanella. A, a seedling with two leaves ; c, the cotyledons. B, an older seedling showing the long-stalked leaves or flag-leaves, of the atavistic period, which appear next after the first leaves. On the contrary it is, like the other new species, dif- ferent from it in almost all its characters. It cannot be mistaken for a weak plant of the parent species at any time. Or to express it more emphatically, if we reduce pictures of Lamarckiana and dwarfs to exactly the same size we find that we can distinguish them by perfectly definite characters. ^ I recall in this connection Darwin's aphorism : Varieties are only small species. Oenothera Nanclla. 363 The dwarfs can be recognized not only as early as the rosette stage but even when the first leaf is developed (Fig. 7^ A). This first leaf is broader and has a broader base and a much shorter petiole than that of 0. La- inarckiana. The same is true of the second leaf. The result of this is a compact appearance in the young plant which makes it possible to record them in the seed pans, if the seeds have been sown so far apart that the seed- lings only just touch one another. Of course there are often one or two doubtful individuals left over, as for example when the seedlings are crowded together in groups ; but the doubt can always be removed by growing the plants in question a few stages further. The stage we have just described is followed by an atavistic period. The dwarf characters disappear and it looks as if the little plants aspired to become tall La- marckianas. There appear two, three, or four narrow leaves set on long petioles (Fig. 7SB, v. v. v.); they conceal the two first leaves which are much smaller, and so determine the general appearance of the plant for a short time. But this stage is soon shown to be a transi- tory one by the appearance, in the center of these leaves, of the compact rosette of the regular dwarf type. (Fig. 78 B, and Fig. 79 A). This so-called atavistic period is very common in seedlings.^ We are all acquainted with the fact that seedlings of species of Acacia with phyllodes have pin- nate and doubly pinnate leaves ; a fact which enables us to derive the species in question from doubly-pinnate ancestors. The seedlings of Ulex, Sarothaninus and those Papilionaceae which lack, or have rudimentary, fo- ^ See the excellent summary of these phenomena in Goebel's Organographie, I, 1898, pp. 121-151. 364 Origin of Each Species Considered Separately. liage behave in the same way.-^ Another striking example is afforded by the decussate arrangement of the leaves of young trees of Eucalyptus Glohidus, a species which in adult life has long-stalked leaves arranged on a different plan.- Siiun lati folium and Berula angustifolia have in adult life simply pinnate leaves but in youth the broad compound leaves which are characteristic of other Um- belli ferae, and therefore evidently are like those of their ancestors. There are numerous other examples^ of spe- cies which exhibit the characters of the systematic group to which they belong as special characters of their early stages. These are the truest cases of atavism. The d\NdiVi-0 enothera is another example of a species which behaves in this way. With this difference, that in this case the ancestry is known by direct observation whilst in the other cases it has only been deduced from a comparative study. But the important point is that in this respect 0. nanella behaves as an ordinary species, or rather, what is much more important, that the best sys- tematic species behave in the same way in respect of this form of atavism, as elementary forms which have just arisen from the parent t3^pe. It is usually during this "atavistic" stage that the fate of the plant — whether annual or biennial — is de- cided. If the former, the stem begins to be formed al- most immediatel}^ ; the elongate leaves are a kind of prep- aration for this, for the leaves which clothe the lower part of the stem are of this form as is shown in the left ^J. Reinke, Untersuchungcn iibcr die Assimilationsorgane der Leguininosen, Jahrb. fiir wissensch. Botanik. Bd XXX, Heft I uiid 4, 1 896- 1 897. ^ F. Delpino, Teoria generale dclla HUotassi, Geneva, 1883, p. 242. ' For the Conifers see L. Beissner, Handhuch der Nadelhoh- kunde, 1891. Oenothera Nanella. 365 figure in Fig. 45 on page 236. If the rosette is to become biennial and if tlie conditions of growth are favorable which practically means, if the rosette has plenty of room to grow in, it begins to develop broader and shorter stalked leaves and is recognizable at once, and from any distance, as a dwarf rosette. The leaves are often not much longer than 7-8 cm. at this age whereas the radical leaves of Lamarckiana often attain a length of 30 cm. or more. This atavistic stage is, however, more often succeeded by a rosette stage which lasts well on into June but, if the plant is going to be an annual, comes to an end then. During this period the leaves are again very broad and attached to the short stem of the plant by a broad base. Their form is often triangular, the leaves being almost as broad as they are long. If the plants have plenty of room, the outer leaves are pressed close against the ground. The outer leaves at this stage have quite short stalks (Fig. 79 A), the inner ones however are almost sessile, almost ensheathing the others with their broad bases. A full-grown leaf of a rosette of this age, with its petiole is shown in Fig. 52 on page 293 at n. But if the plants are growing so thickly that they are cramped for room, their whole appearance becomes quite different but none the less recognizable (Fig. 79 B). The leaves which make their appearance after the "atavistic" stage (v. V.) stand more or less erect, are somewhat nar- rower and have longer petioles but are still set on the stem by a broad base. The result is that the stalks seem to be twisted in a curious way which is not brought out clearly in the figure, but which is so characteristic a feature of the young plant that it is by this character that the young dwarfs are usually first identified ; and they differ from 366 Origin of Each Species Considered Separately. individuals of 0. Laniarckiana of the same age in other respects as well. (See Fig. 64 on page 325.) I have recorded the young plants in any one of the four stages (Figs. 78 and 79) according to circum- stances. The further apart the seeds are sown the sooner can the recording be done. But even when they are sown thin a few seeds occasionally fall close together ; so that we find groups of seedlings which cannot be identified until long after the others have been recorded and re- moved. It is often from 4 to 6 weeks before the last individuals have fully developed their characteristics. Fig. 79. Oenothera nanella. Young rosettes in May and June. A, from seeds sown thin; B, from seeds sown thick ; V. v. the long-stalked leaves of the atavistic period. Indeed I have often had to transplant the seedlings be- fore I could be certain about them : when I did this I gave them ample room and grew them for about a month more in the boxes. If I recognized a plant as a dwarf-mutant in a culture of another species I kept it until it had at- tained the stage shown in Fig. 79 A ; and usually trans- planted it to watch its further development. If on the other hand it was merely a question of finding out whether any Lamarckianas occurred in sowings of 0. nanella (as they often did after fertilization in the open) Oenothera Nanella. 367 the recording was usually done at an earlier age. For it is obvious that the earlier this can be done the greater is the number of individuals that can be dealt with. When the seeds are sown in beds and not in boxes, as they usually were at first, we must of course await either the full development of the rosette or if they be- come annuals, the production of a stem. The characters described have enabled me to obtain the figures already given, as to the repeated appearance of O. nanella from O. Lamar ckiana and from other new species. I propose to give these figures again together with results obtained in certain other cultures in order to convey some idea of the frequency of nan^/Za-mutations. The fact that they appear every year, and in numbers which become greater in proportion as the sowings are more extensive is proved by the tables given in §§ 2-7; so that I shall have no occasion to refer to it again. INDIVIDUALS OF O. NANELLA V/HICH HAVE ARISEN BY MUTATION. I. FROM OENOTHERA LAMARCKIANA. THE ORIGIN OF THE YEAR LAMARCKIANAS: Lainarckiana-idcrmiy . 1889-1899 A branch of the same . 1895 Laemfolia-tamily . . 1889 Various crosses (p. 300) 1898 O. scintillans .... 1897-1898 A biennial culture , . 1897 Culture of plants with variegated leaves . . 1899 Totals 70,154 340 0.5 The proportion of dwarfs produced by Lamarckiana is — if we neglect the laevifolia-ia.mi\y where it is possible that other factors may have come into play — a fairly TOTAL OF SEEDLINGS NANELLA NAN. % 50,000 158 0.3 10.000 111 1.1 400 12 3.0 4,599 26 0.6 1,654 15 0.9 1,529 9 0.6 1,972 9 0.5 368 Origin of Each Species Considered Separately. constant one; moreover It seems to make no difference whether the Lamarckianas are of pure strain or the off- spring of crosses. This conclusion is supported by the proportions in which O. nanella occurs in crops raised directly from crosses, that is, in the first generation after the cross. The foregoing table referred to the second generation from artificial crosses, or from free crossing as in the case of O. laczi folia. INDIVIDUALS OF O. NANELLA WHICH HAVE ARISEN BY MUTATION. II. FROM CROSSES. CROSS YEAR TOTAL OF SEEDLINGS NANELLA NANELLA % O. Lam. X 0. biennis 1900 80 1 1.0 O, lata X O. biefinis 1899 299 2 0.7 O. Lam. X O. brevistylis 1898 293 5 1.7 O. Lam. X O.gigas 1899 100 2 2.0 O. Lam. X 0. scintillans 1899 112 1 1.0 O. lata X 0. Lam. 1900 2000 3 0.2 O. lata X 0. Lam. 1895-1900 2387 26 1.1 O. lata X 0. brevistylis 1896-1899 425 6 1.4 Totals 5696 46 0.8 If we compare these figures with those already given for other species we find a striking agreement between them and those for 0. ohlonga (about 1 %) and we may therefore regard 0. nanella as one of the commoner forms. It may also be regarded as parallel in this respect with 0. lata which will be described afterward (§ 22) and possibly with O. albida which however appears in varying proportions. 0. riihrinervis, O. gigas, and O. scintillans on the other hand form a quite distinct group of rarer mutations, whilst O. semilata and the other less important types form a third group of still rarer deviations. Oenothera Nanella. 369 O. nanella has arisen from other new species in about the same proportions; from leptocarpa in 1896 in a pro- portion of 0.4 %, from O. scintillans in various experi- ments from 1896-1899 also in a proportion of OA fo (There were 29 nanellas amongst 7872 seecUings). The progeny of nanella mutants come true to seed. I have observed over 400 examples of this species which have arisen directly from other forms. Together they obviously constitute a species which can at once be rec- ognized by many characters, although every one of them was different from its parents and ancestors. I have already in § 3, p. 238, given the most im- portant facts relating to the constancy of this form. It only remains to amplify the brief account of the experi- ments given above. I have made four series of experiments on the con- stancy of 0. nanella. I began the first of them in 1889 with the twelve mutants from the laevifolia-isimily men- tioned above (273). As I was not familiar with parch- ment bags at that time I was not able to guard my plants against the visits of insects although I always grew them on a bed which was isolated as much as possible. But even so, the dwarf type proved heritable in a very high degree. I harvested the first seeds in 1890 as the plants did not flower till the second year. I raised 20 plants, of which 18 were dwarfs; they flowered the same sum- mer and set plenty of seed. This seed (about 6 ccm.) I sowed on a bed of about 4 square meters. The culture consisted almost entirely of dwarfs. After this the plants flowered regularly in the first summer so that I obtained the fourth generation in 1893 and the fifth in 1894. The third consisted of 400 plants which were practically all dwarfs; I fertilized some of these with their own pollen 370 Origin of Each Species Considered Separately. by enclosing them in bags. As a result of this, absolute purity of the cultures was attained in 1894. A culture of 440 which flowered during August and September consisted entirely of dwarfs. I have not continued this experiment further because it seemed to me more important to work with the mu- tants themselves and to test the constancy of the first generation. In 1895 I used for this purpose some nanellas which had just arisen from the Lamarckiana-isimily, and from a branch of it. I fertilized 12 of the former and 8 of the latter with their own pollen excluding the visits of insects from them. I collected the seeds of each plant separately, sowed them in the following spring, and, after a month, transplanted all the seedlings without ex- ception, into wooden boxes, in manured soil, where they would have plenty of room to develop into rosettes like that shown in Fig. 79 A (p. 366). Some of them which were too close together grew like the type shown in Fig. 79 B ; I removed the plants which surrounded these in order that they might have room to expand their leaves in. The recording, as a result of this, took place at dif- ferent times but all during the month of June. The twenty seed parents of 1895 were raised from the seeds of nine separate plants of Lamar ckiana of wdiich five belonged to the third (p. 224) and four to the second generation (p. 262). The twenty mutants themselves therefore belonged to the fourth and third generations. In the following tables I denote the grandparent by Lam., the parent or mutant by Nan., and the seedlings raised from the seeds of these latter by S. The letters A-E refer to the five Lainarckiana-^A^ints of the third generation, L-O to those of the second generation ; their Oenothera Nanella. 371 children are the nanella mutants (nan.) whose seeds I sowed. The number of offspring from each separate parent are recorded in the tables. OENOTHERA NANELLA. OFFSPRING OF MUTANTS FROM THE THIRD THIRD SECOND LAMARCKIANA-GENERATION LAM. NAN. s. LAM. NAN. s. LAM. NAN. s. A No. 1 277 c No. 1 30 L No. 1 55 A " 2 124 ( ( " 2 21 ( t " 2 99 B 89 D " 1 80 i { " 3 302 ( ( " 2 66 E " 1 38 1 1 " 4 22 ' i " 3 292 t ( " 2 71 M " 1 30 ( ( " 4 68 N " 1 339 < ( " 5 34 O " 1 " 2 105 321 Total 950 Total 240 Total 1273 Altogether there were 2463 seedlings which w^ere all without exception O. nanella. These results seem to me to justify the belief that the remaining nanella-muta.nts of 1895 would also, if I had collected their seed and sown it, have proved constant. One thing which I learnt from these extensive sow- ings was that the dwarfs were recognizable, and could therefore be recorded, at a much earlier stage than I had imagined before — viz., in the pans, before the first trans- planting. Now, it is just the transplanting in such ex- periments which is the greatest labor and it is impossible to hand it over to an assistant on account of the danger of possible mistakes, so that this discovery opened up the possibility of testing the constancy on a much larger scale. I used for this purpose the nanellas, referred to on page 262, which came up in 1896 from seeds which had 372 Origin of Each Species Considered Separately. remained a year in the soil. Such plants also occurred in the main culture referred to on page 224, although they are not referred to there. I selfed 38 plants, in bags, with their own pollen. They were all mutants from Lamarckiana, some with three, some with two genera- tions of tall ancestors. I saved and sowed their seed separately ; and recorded the seedlings at the stage shown at Fig. 78 B. Doubtful ones were allowed to grow a little further. The average number of seedlings from each of the 20 seed-parents was 500, the maximum was 860, and only in three cases was it less than 100. The total number of seedlings was 18649; they were without exception dwarfs. Three of them were also ob- long a and one also elliptica. Thus the seed-parents of the second experiment proved themselves to be, like the 20 of the first, perfectly con- stant. It seemed to me important to test the constancy of nanellas from other sources. I chose for this purpose two plants from a scintillans-isimily. This family arose from the lateral branch of the Lamarckiana group (p. 262) and indeed from the only individual referred to there. This was biennial and flowered in 1896. I sowed some of the self-fertilized seed of this in 1898 and selfed the scinfillans plants again with their own pollen. From the seed thus produced I obtained nine examples of nanella, which I transplanted and selfed. Only two of them however set seed. They had two generations of scin- tillans behind them, and behind these two generations of Lamarckiana. The plants had become very weak; and the harvest was a meagre one. Only 64 seeds germinated ; but they were all nanella. This shows that the dwarfs even when Oenothera Nanclla. 373 they arise from another new species, exhibit in the first generation not only the same characters but are as true to seed, as those which arise directly from Lamarckiana. For testing the constancy of this form under self- fertilization in subsequent generations I used the second of the above mentioned experiments as a starting-point (p. 371 ). Some of the 2463 plants mentioned there were chosen as seed parents and self-fertilized. The seeds gathered from 4 of them were sown in 1897 ; they gave respectively 94, 135, 154 and 164 seedlings — 547 in all — which proved without exception to be dwarfs when they were recorded as large rosettes m July. I allowed about 100 of these to flower and fertilized some of them with their own pollen. In 1898 I raised from the seed thus produced the fourth nanella generation which again was perfectly constant, and allowed about 100 specimens to flower. The fifth and sixth generations (1899 and 1900) also came perfectly true to seed. Of the total number of seedlings — about 400 in 1900 — I allowed about 70 to flower and used about 30 of these as seed parents. Thus from the third generation to the sixth, embra- cing in all over a thousand plants, there occurred no single instance of atavism. The new species must therefore be regarded as perfectly constant. The constancy of nanella is however incomplete in the sense that it has inherited the capacity of mutating, from the parent species. For it gives rise to individuals which though obviously nanellas also betray the charac- ters of some of the other new species. And, conversely, it occasionally happens that dwarfs arise from other new species and then bear the characters of both types together. In this way we get species of the second order, which correspond to the cultivated varie- 374 Origin of Each Species Considered Separately. ties of the second and third order as described above. ^ Combinations of this kind occur both in pure cultures and in the offspring of crosses. The following are the cases which I have observed so far. Commonest of all were dwarfs which also bore the characters of lata, developed to their full extent. I no- ticed the first in 1892 amongst my nanellas, which were at that time, as I have already stated, constant in every other respect. There were three plants which, like the rest, were annual. They flowered amongst the others and were fertilized with their pollen. They produced fruits which however contained little seed. They attained a height of 25 cm. and could be recognized even before they flowered as naneUa-lata. Their broad rounded leaves, the compact inflorescence with broad bracts, their thick swollen buds and the crumpled petals of their flow- ers exactly resembled those of true lata. But their seeds, resulting from fertilization by nanella, gave rise to ordi- nary nanella only In the sum.mer of 1896 I had another example of nanella-lata from seeds of self-fertilized nanella. It agreed exactly with those observed in 1892. In 1898 and 1899 the same combination appeared amongst the offspring of two crosses (1)0. Lamarckiana XO. nanella, (II) O. lata X O. nanella. In the first of these (1898) there were two examples amongst about 100 dwarfs, in the other (1899) only one amongst 133 nanellas and 79 latas. The second culture was under- taken solely with the object of bringing about the com- bination of the two forms by crossing. This object was attained, the characteristic features of the two pa- ^ Compare for example Scahiosa afropurpurea nana purpurea and other cases on p. 197. Oenothera Nanella. 375 rents being fully developed; but only in a single indi- vidual. Beside nanella-lata I have observed the following combinations : COMPOUND TYPES OF OENOTHERA NANELLA. FROM SEEDS OF COMBINATION O. Lamarckiana X O. nanella O. 7ianella-oblonga 1898 O. lata X O. nanella O. nanella-albida 1899 " " " " O. nanella-ellipHca 1899 " " " " O. 7ia7iella-sciniillans 1899 O. nanella O. nanella-oblonga 1897 O. scintillans O. scintillans-nanella 1899 O. gigas O. gigas-nanella 1897 O. Lamarckiana O. nanella-ellipiica 1889 This list is sufficient to show that the dwarf-character can be associated with the characters of the various other new species. The characters of these others may also be associated with one another although this is very rarely the case. This difference between 0. nanella and other species is doubtless intimately connected with the varietal, as opposed to specific, character of 0. nanella ^vhich marks it off from all the other new species. If we regard nanella as a variety we should expect it to arise from any of the new species just as much as from O. Lamarckiana itself. And it should be noted in this connection that, of the 6 combinations mentioned, only one arose from the seeds of nanella. I shall now give a description of the flow^ers of the dw^arf (Fig. 77, p. 361). These are remarkably large compared with the size of the plants, especially in the case of vigorous biennial individuals. On the latter they attained to very nearly the size of the flowers of O. Lamarckiana. On plants which flower in the first year they are usually much 376 Origin of Each Species Considered Separately, smaller in correspondence with the greater weakness of the whole plant. The petals commonly measure only 2% X 4 cm., as opposed to the 3X5 cm. of Lamarckiana. On annual plants some of the flowers are often in- completely developed. As a rule only one or two flow- ers on a plant are affected in this way. Sometimes there is very little pollen, sometimes none at all; very often it happens that the stigmas cannot separate, and remain close together. This stig- matic group is often quite small and . some- times blackens and shriv- els up before pollination. Or again, the style may be so short that it does not protrude from the corolla.^ A striking abnormal- ity is an oblique position of the buds on the calyx tube (Fig. 80). The calyx is bent sideways in such a manner that the bud is at right an- gles to the calyx tube. The opening of the calyx is interfered with by this and indeed often takes place abnormally or incompletely. The petals do not unfold properly and the sexual organs are more or less sterile. These abnormalities are usually to be found on the lower flowers when the plant has not attained a height _^ These and other malformations of the dwarfs are often due to a disease, and as such, to a large degree dependent on outer circum- stances. (Note of 1908.) Fig. 80. Oenothera nanella. Buds at the top of the stem. At the side are shown the commonest malfor- mations of such buds. Oenothera Scintiltans. 377 of more than 10-15 centimeters. But even the common Lamarckiana often produces some abnormal flowers amongst the lower ones. If the nanella survives this pe- riod and if it becomes markedly stronger it forms, first, a shorter or longer intermediate piece and then a fine head of large flowers. This is borne by the barren look- ing, bracteated, flowerless part of the stem, well above the lower half of the inflorescence. But it is by no means all plants that are strong enough to reach this state. The best way of raising fine plants of nanella is to make them biennial by sowing the seed late. C. THE INCONSTANT SPECIES. § 19. OENOTHERA SCINTILLANS. (Plate V.) As far as we know, species in nature are constant. This is true also of elementary species, and of most so- called varieties. It is true that the older systematists, such as Koch, Spach and many others, believed to be able to distinguish varieties from species by their in- constancy. But they seldom took the trouble to exclude the visits of insects in the numerous experiments they made. If we take this precaution many varieties prove to be as constant as species. The universal belief in the constancy of species has led us to regard this quality as one of the attributes of a species. From this standpoint, it would seem a con- tradiction in terms to speak of an inconstant species. But such a contradiction need only trouble adherents of the current theory of selection. The mutation theory can remove even this difficulty. Lack of constancy is 378 Origin of Each Species Considered Separately. obviously one of the most unfavorable characters that a species can possess ; and the theory of selection which can only explain the origin of favorable characters can- not account for the existence of unfavorable ones. According to the mutation theory a species, even if it is so weak that it can hardly maintain itself, and can only just reproduce itself, is capable of existing, for a time, alongside the parent species. Oenothera hrevistylis, which hardly sets any seed and yet has maintained itself amongst the Lamarckianas at Hilversum since 1887, proves the correctness of this view. At some future date no doubt, if the struggle for existence becomes keener, it w^ill give way to Lamarckiana or be vanquished in the struggle with other plants whilst Lamarckiana may survive. But if the conditions of life remain as they have been up to now, there is at least the possibility that O. hrevistylis may continue to exist alongside La- marckiana.^ This difficulty can be avoided by confining the term species to those forms which have emerged victorious from the struggle for existence. But such a limitation of the meannig of the term would of course be perfectly arbitrary and only serve to further confuse a problem already sufficiently difficult. The doctrine of mutation on the other hand makes it easy to see how species may arise and yet be disquali- fied for survival for any length of time. Mutability produces deviations in all directions (I, § 26, p. 198) ; it is absolutely uninfluenced by the greater or lesser utilit}^ of the changes it produces. It simply produces varia- tions, leaving it to the struggle for existence to decide whether they are in the right direction or not. But the ^ It still occurred in that locality in 1907. (Note of 1908.) Oenothera Scintillans. 379 event is dependent not only on the quality of the varia- tion but also on the environment in which it is placed. The variations which pass through the sieve of the struggle for existence are not different from, but merely part of, those which are put into it. The mutation theory admits of the production of such forms as will sooner or later for some reason or other, perish without having contributed materially to the flora or fauna of a district. The causes of such dis- appearance are mainly three: (1) sterility, or at any rate insufficient fertility; (2) constitutional delicacy; (3) inability to breed true. Nor is there any a priori ground for supposing that more ''fit" species arise than ''unfit." There have arisen in my cultures besides robust forms like 0. gigas and O. riihrinervis, and weak ones like O. ohlonga and 0. alhida, a series of forms which were either sterile, or were fertile but did not come true to seed. I should have called them transitory species, were it not that all species are transitory. I now refer to the former group as infertile and to the latter as inconstant species. Neither of these types can last long in nature. They must obviously be excluded from amongst the species with which the ordinary investigation of nature familiar- izes us. It is only when one can witness a period of mutation that there is any chance of seeing such forms. I propose to deal now with some types of inconstant species and shall begin with the one I have investigated most thoroughly. This is Oenothera scintillans which is figured on Plate V and in Fig. 47 on page 244. I have already stated, in § 3 of this section, that the seeds of this species pro- 380 Origin of Each Species Considered Separately. duce three different forms, 0. scintillans, O. ohlonga, and O. Laniarckiana even after it has been carefully fertilized with its own pollen and the visits of insects have been effectually excluded. The proportion in which O. scin- tillans is reproduced is in some cases about 35-40 % and in others about 70 %. Before we can estimate the effects of this incon- stancy we must know what happens in subsequent gen- erations. I shall afterwards give the details of some ex- periments which show that the 0. ohlonga and 0. La- marckiana thus produced are as constant as those given off directly from the main stem of the Laniarckiana-isim- ily. The scintillans on the other hand behave like their parent, their offspring segregating in the same way. What will be the composition of the successive gen- erations? We will suppose that the plants are self- fertilized, that no selection takes place and we will put the proportion of scintillans in each generation at about one-third. And we will limit the extent of the genera- tions to a thousand plants each. The contents of suc- cessive generations will, then, obviously be :^ . \ SCINTILLANS LAMARCKIANA + OBLONGA 1st Generation 333 667 2nd 111 222 -f- 667 — 889 3rd 37 74 + 889 — 963 4th 12 25 + 963 = 988 5th 4 8 4- 988 — 996 6th 1 3 4- 996 = 999 7th 0 1000 Therefore in a batch of about 1000 plants all the scintillans would have died out after seven generations without the operation of any selective process. In the case before us however the process would be hastened ^The xth generation will contain (Va)^ scintillans. Oenothera Scintillans. 381 by a very definite selection resulting from the fact that scintillans is much more delicate than Lamarckiana. It is now sufficiently clear that a species which pro- duces besides offspring like itself other constant types must inevitably disappear sooner or later. If the constant types appear in a smaller proportion than we have considered so far, in each generation, as in the case of 0. scintillans producing 70 % of its kind (p. 246) it will take longer for the form to disappear; but disappear it must.^ It is only by excelling its con- stant offspring in individual strength that it can ever stand a chance of surviving altogether. If it did this it would be in the position in which 0. Lamarckiana finds itself now with regard to the new species arising from it. These facts give a simple explanation of the absence (or perhaps rather the great rarity?) of inconstant spe- cies in nature. For it is not necessary to assume that such do not arise or even that they do not arise often. The proof that they cannot maintain themselves is suf- ficient. Left to themselves the}^ will be reduced in a very few years to a hundredth or even thousandth part of the total of their offspring, and they will very soon be lost altogether. They can only continue to exist by being produced continuously or, at least, frequently by the parent species. The mutation theory renders the origin and disappear- ance of unfit types intelligible ; moreover the actual origin of such has been observed. These cases constitute an insuperable obstacle in the way of the theory of selec- tion. * The I2th generation will bring the form down to about i % ; and generally speaking the xth to (Vio)''. 382 O rig 171 of Each Species Considered Separately. From these general considerations I pass on to the special treatment of our first example O. scintillans. Fig. 47 and Plate V show the flowering spike of annual examples of the species ; the long, tapering budbearing internodes above the open flowers cannot fail to attract attention. This feature stamps the habit of the plant from July to late in the autumn : in most of the other species the buds do not rise much above the crown of flowers. Moreover the bracts in this region are pretty Fig. 8i. Oenothera scintillans. A, young plant with 6 leaves above the cotyledon. Bj young rosette at the age of two months. large so that the youngest part of the stem is fairly thickly clothed with leaves. The flowers are considerably smaller than in 0. La- mar ckiana, a circumstance almost certainly due to the general delicacy of the species. Otherwise, the struc- ture of the flowers resembles that of the parent species; for example the stigmas extend well beyond the anthers so that the flowers are generally cross-fertilized. Oenothera Scintillans. 383 The development of the anthers and the pollen is to a high degree dependent on external conditions. The pol- len is sometimes plentiful, sometimes scanty, and at other times entirely absent. These variations occur on one and the same plant and seem to depend chiefly on the temperature, inasmuch as the anthers degenerate under the influence of hot weather. It is in consequence of this circumstance that I have lost many fruits by en- closing flowers in parchment bags (to insure pure self- fertilization) in the full sunlight. Fig. 82. Oenothera scintillans. A rosette ot radical leaves, at the end of June. The annual plants are only very slightly branched, and begin to flower when they are only Yo meter high. The lateral branches spring from just underneath the flowering zone, and on them isolated flowers appear towards the end of September or even later. Biennial plants are usually more branched, and if the heart of the rosette happens to have frozen in the winter a circlet of 384 Origin of Each Species Considered Separately. secondary stems is formed. Biennial plants are in every respect stronger and bear larger fruits with better seeds. But the most characteristic feature of this species is its smooth shining dark-green narrow leaves. The young rosettes are recognizable by this char- acter (Figs. 81 and 82), and can easily be distinguished by means of it from the species by which they happen to be surrounded (Fig. 52, p. 293). The leaves are not very narrow at first ; in fact they do not become so until the rosettes are 2 or 3 months old. This feature gradually becomes more pronounced during the summer whether the plant remains in the rosette stage or develops a stem. The midrib of the leaf is broad and like the leaf stalk is of so pale a green that it might almost be called white, and has not a trace of red color in it. The leaves of the full grown rosette have long petioles and are about four times as long as they are broad or even narrower. There are no unevennesses on the blade nor is there that pale green bloom on them which is characteristic of O. alhida and 0. rtibrinervis ; they are almost absolutely smooth and very different from those of Lamar ckiana in their dark green color. Indeed, scintillans bears very little resemblance to its parent species except in its flowers. The leaves of the stem (Fig. 54, p. 295) resemble those of the rosette in all essential points and so do not require any special description. In regard to the mode of its origin 0. scintillans re- sembles O. gigas and O. semilata in being one of the rarest types. It has only arisen 14 times altogether as a mutation. Although most of these instances have al- ready been described it is worth while summarizing them all here. 1 (2) 14,000 0 8,000 2 (2) 1,800 0 10,000 1 (2) 3,000 0 164 1 (1) 300 0 Oenothera Scintillans. 385 OENOTHERA SCINTILLANS. INDIVIDUALS THAT HAVE ARISEN BY MUTATION. SOURCE YEAR ^°^^^ °^ O. SCINTILLANS P'^ODUCING SEEDLINGS RIPE FRUIT O. lata 1888 I 1895 The Laniarckiana-ioimWy - 1896 ( 1897 Lateral branch of this family 1895 O. Lam. , a subsidiary culture 1897 O. lata 1898 O. lata X O. biennis . . . 1899 As the last column of the table shows I only succeeded in getting ripe fruits from five of these mutants, of which four set seed in the second year (2) and only one in the first ( 1 ) . The rest died as rosettes or at any rate be- fore they fruited. The percentage composition of the cultures raised from these seeds has already been given on pp. 244-246, but will be described in greater detail now. I shall begin with the oldest. It appeared in 1888 in the /a/a- family referred to on p. 288; it was biennial and flowered luxuriantly in July 1889 but was left to be fertilized promiscuously amongst a crowd of Lainarc- kianas. It had all the characters which were afterwards observed both in its offspring and in the other mutants. I sowed its seed partly in 1890 partly in 1894 and ob- tained, in both years, both annual and biennial plants. The rosettes of 1894 flowered in 1895; the plants were self- fertilized in parchment bags. There were 14 healthy plants, bearing hardly any branches : their fruits were small, and did not afford more than 1 to 3 cubic centi- meters of seed per plant. The seeds were sown on sep- 386 Origin of Each Species Considered Separately. arate beds and the young plants recorded at the end of June. As the seeds were sown thin and as a great many did not come up, the seedHngs stood far apart and had plenty of room in which to develop their characteristic features. I counted :^ PER SEED PARENT TOTAL IN \ Seedlings 16—52 399 O. scintillans 2— 9 62 15 O. Laniarckiana 7—36 268 68 O. oblofiga 1—11 60 15 O. lata 0— 2 8 2 O nanella 0— 1 1 This experiment showed that each of the 14 seed pa- rents gave rise to the three chief forms, when self- fertilized. They did this moreover, so far as the small numbers enable us to judge, in not very widely different proportions. In this experiment the original mutant was fertilized by insects ; but all the subsequent mutants which appeared were enclosed in parchment bags, as soon as they began to flower, and artificially self-fertilized. The first to be treated thus was the scintillans, mentioned on p. 262, which appeared in 1896 in a branch of the Lamarckiana- family. Six stems were developed from the axils of its radical leaves and a quantity of seed was set. I also succeeded in taking cuttings from the remaining branches of the rosette : they survived the winter and flowered in the following year. I sowed the 1896 seed partly in 1897 and partly in 1898, in the former year both in pans and in a bed in the garden. The three crops raised in this way were composed as follows :^ ^ See the table on p. 244. ^ See the second table on p. 245. Oenothera Scintillans. 387 1897 1897 1898 IN PANS IN THE GARDEN IN PANS Number of seedlings 572 275 165 Oo scintillans 36 % 34 % 36 % O. Lamarckiana 52 % 52 % 60 % O. oblonga 10 % 13 % 3 % O lata 1 % 1 % 1 % O. nanella 1 % 0 0 In the summer of 1897 I selfed five of the La- marckianas in this culture with their own pollen. Each bore from 12-13 cm. of seed, of which some was sown next year in the garden and in pans. 117 seeds germinated in the garden and 1079 in the pans. There was not a single example of scintillans amongst these. The major- ity of them were Lamarckianas, with a considerable ad- mixture of mutants. In the garden these were, 4 O. ruhrinervis, 3 O. lata, 1 0. nanella, 1 O. alhida and 2 O. oblonga ; in the pans the only mutants were 7 ex- amples of O. nanella. The Laniarckianas, therefore which are produced by 0. scintillans, have the same constancy as the original Lamarckiana, that is to say their grand-parents, but also exhibit the same degree of mutability. The point is that a continued segregation into La- marckiana, scintillans and oblonga is not witnessed in the seedlings from the Laniarckianas extracted from scin- tillans as it was in the original scintillans. Of the mutants mentioned O. rubrinervis, 0. lata and O. nanella flowered the same summer. The very important question now presented itself, how the scintillans plants in this generation behaved on self-fertilization. To answer this question I enclosed over 50 plants of the 1898 culture in bags, harvested their seed separately and sowed it. All the seedlings 388 Origin of Each Species Considered Separately. were transplanted, under glass at first, in order to give them plenty of room for the full development of the rosette. The rosettes were counted between the ages of 2 and 3 months, the process lasting from the middle of May till the middle of June (see Fig. 82). Alto- gether about 5850 rosettes produced by 42 plants were recorded. The crops which contained less than 50 seed- lings from a given parent were recorded but not included in those from which the following percentages were counted. The average number of seedlings per parent plant in the cases dealt with is therefore about 140. The number of scintillans naturally varied from crop to crop, as a result, doubtless, of the smallness of the number of seedlings counted. I have determined the percentage values for each parent, and arranged them in groups of 1-10 %, 10-20 % and so on. I found: NUMBER OF SEED PARENTS 7 % 1 19 % 1 21—30 % 9 31—40 % 12 41—50 % 15 51 55 % 4 This gives an average of 40 %, a figure which agrees with the coefficient for the grandmother (36 %) closely enough. The oblongas in this experiment varied between 0-12 per seed parent. There were 197 of them altogether, i. e., they formed about 3 % of the population. The rest were, with the exception of about 1 % O. lata and O. nanella, all O. Lamarckiana. We have therefore on the average : Oenothera Scintillans. 389 SECOND GENERATION FIRST GENERATION O. scintillans 40 % 36 % O. Lamarckiana 56 % 60 % O. ohlonga 3% 3% O. lata and nanella 1 % \ % The agreement between the two succeeding genera- tions is as great as can be expected in an experiment of this kind. There were four seedparents with 52, 52, 54 and 55% scintillans in, respectively 111, 61, 161 and 95 seedlings. The ^aH//7/a/z^-producing capacity seems very variable, but the figures would perhaps deviate less if greater numbers had been grown. The deviation seems, however, to lie within the limits of individual varia- bility. In 1896 six plants of O. scintillans arose in the main line of the Lamarckiana-family (p. 224 and 385). I succeeded in bringing two of these through the winter and in getting them to flower in 1897. Their pollination was artificial and not disturbed by the agency of insects. The amount of seed set was small, varying from % to 2 ccm. per plant : is was sown, separately for each seed parent, in March 1898. One seed parent gave rise to 365 seedlings which in- cluded the same types in the same proportion as in the previous experiment.-^ The other yielded a total of only about 200, which was, however, made up quite differ- ently.^ 69 % of the population were scintillans, that is to say twice as much as in the previous experiments. The relative number of ohlonga was also doubled and amounted to 21 %. The number of Laniarckianas was correspondingly low, amounting to no more than 8 %, * See the numbers on the lower table on p. 245. ' P. 246. 390 Origin of Each Species Considered Separately. whilst that of the other mutants (0. lata, 0. nanella, etc.) remained at about 2 %. There is therefore in 0. scinfillans a highly fluctuating degree of the hereditary capacity, if by this term we may denote the different proportions in which 0. scintillans is able to produce offspring like itself. The hereditary capacity can either be very small, or about 35-40 %, or as much as 69 % ; the latter figure being about twice as large as the formen In the case of the former this capacity was essentially the same in the second generation as it was in the first, and in the case of the latter the difference was also very little. This is proved b}^ the result of the continuation of the experiment we are dealing with (See p. 246). In 1898 about 30 plants were self-fertilized; they yielded a poor harvest. The crop raised from 26 of them consisted of about 2200 plants, i. e., about 90 per seed parent. The hereditarv values for each of these are arrans^ed in the following list in groups, as before. % SCINTILLANS NUMBER OF MOTHER PLANTS 66—69 % 2 71—74 % 2 76—80 % 5 81—85 % 6 86—90 % 9 92—93 % 2 The average is 84 % and therefore even higher than the 69 % of the previous generation. The average composition of the whole culture from the 26 parent plants of 1898 was: 0. scintillans 84 % 0. Lamarckiana 13 % 0. oblonga 2 % 0. lata 1 % Oenothera Scintillans. 391 The amount of 0. ohlonga has greatly decreased, whilst that of Lamarckiana has somewhat increased (see p. 246). In the summer of 1899 I again selfed a whole series of plants in this culture. I selected these from the off- spring of two plants which had produced respectively 87 % and 90 % scintillans and seemed, therefore most likely to breed true. I only fertilized scintillans plants. The yield, however, was very poor; 10 seed parents only giving more than 60 seedlings each. These could be recorded in June and showed high hereditary coefficients : MOTHER NUMBER OF SEEDLINGS % SCINTILLANS 1 146 86 2 122 91 3 113 76 4 112 92 5 98 89 6 96 87 7 77 83 8 75 80 9 74 81 10 68 74 The whole crop raised from the seeds of the 29 plants gave: NUMBER OF SEEDLINGS % O. scintillans 1126 79 O. Lamarckiana 93 6 O. oblonga 209 15 Total 1428 These figures agree almost exactly with the mean value of the culture in the previous generation in spite of the fact that I began the culture with two seed parents with exceptionally high ^n/i^/7/a7Z.y-producing capacity, viz., ^7 % and 90 %, and irrespective of the fact that 392 Origin of Each Species Considered Separately. the proportions for the other mutants seem to have been inverted. This result tends to prove the correctness of the conclusion, suggested above (p. 389), that the devia- tions from the mean hereditary coefficient are phenom- ena of fluctuating variability and have nothing to do with mutation. The fifth mutant of O. scinfillans from which I was able to get seed arose in 1898 from the /a/a-family de- scribed on page 285. There was only one plant which unlike all the previous ones developed a stem very early and flowered in the first summer. It was self-fertilized in a bag, set little seed and gave rise in 1899 to 148 identifiable plants of which 37 % were scintillans^ This is another example of the hereditary coefficient exhibited bv two of the three other mutants which were tested — a particularly interesting case because the origin of this scintiUans was quite different from that of the others and the plant was an annual. I shall now summarize these coefficients. SOURCE YEAR OF MUTATION SCINTILLANS PLANTS 2nd gen. 3rd gen. 4th gen. O. lata 1888 — IS % O. lata 1898 37 % O. Lamarckiana 1895 34—36 % 40 % O. Lamarckiana 1896 39 % O. Lamarckiana 1896 69 % 84 % 79 % These figures seem to be arranged in groups of 15 %, 34-40 %, and 69-84 %. It would obviously be very im- portant to determine more of these figures in the case of a large number of ^cm////a/w-mutants ; if this were done the groups will probably turn out to be more vari- ^ See the first table on p. 245. Oenothera Elliptica. 393 able, or even wholly illusory. Perhaps we might even get a constant race of scintillans. § 20. OENOTHERA ELLIPTICA. Almost every year there appear amongst my plants isolated individuals with very narrow leaves. There are three types of such. First those in which the narrow- ness is the result of some malformation. Sometimes one half of the leaf in this case is more reduced than the other and the leaf is consequently more or less deformed. Plants of this kind sooner or later return to the normal type of 0. Lamarckiana, and behave afterwards just like this. The narrowness is presumably in this case a pathological phenomenon ; I shall not deal further with it. The two other types are constant and maintain the character throughout life. One of the forms has long leaves which are broadest in the middle and gradually taper off to the tip and to the stalk. I call this form 0. elliptica. The other, a much rarer form, has linear, almost grasslike, leaves and will be described in the next section under the name of O. suhlinearis. The seedlings of O. elliptica are recognizable at a very early age (Fig. 83 B, to be compared with Figs. 64-66, pp. 325-326). Its leaves have long petioles and are very narrow, seldom attaining a breadth of more than 0.5-0.7 cm. for a length of 8-10 centimeters. One result of this is that they assimilate much less carbonic acid than 0. Lamar ckiana, so that they are weak and very easily overgrown by their normal neighbors. But even when they are transplanted early and treated with every possible care they grow very slowly. The plant shown in Fig. 83 B was photographed in July. 394 Origin of Each Species Considered Separately. The great majority of the new mutants of this species stayed in the rosette stage their first year; but they were so dehcate that I did not succeed in wintering them. Fig. 83. Oenothera elliptica. A, twig of an adult plant, (1895) ; B, a seedling of 1893; C. radical leaf of a full grown rosette. Others developed stems but did not flower. I have only seen flowers on ten plants altogether and only obtained seed from five of these. Oenothera Elliptica. 395 j^r ■f*!."^'. \ \ Even when they flowered the plants were still deli- cate: their leaves retained the long narrow shape (Fig. 83 ^). The plants do not as a rule attain a great height but are profusely branched and are so unlike an Oeno- thera Lamarckiana that they do not look as if they could be any relation to it. So unlike, indeed, are they that the seedlings ran the risk of being taken for weeds and thrown away.^ But the flowers reveal its kinship with O .Lamarckiana at once. They are large and fine, much larger indeed for so weak a species than our experience of O. oblonga, O. scintillans and others would lead us to expect. They have the same structure as those of the parent species; the stigma extends well above the anthers and so cannot be fertilized without the help of insects or of the experi- menter. The shape of the petals however is different, as will be made sufficiently evident by a comparison of Fig. 84 with Fig. 42 on page 218. The petals of O. Lamarckiana are broader than long, indented at the tip and so more or less obcordate. In the open flower their margins overlap so that a closed cup is formed. The petals of 0. elliptica are elliptical ; ^ This circumstance considerably increases the work in my ex- perimental garden. Weeding ought only be done by assistants who can assign individual plants to their species and can be trusted to spare unknown forms. For the rarer a mutant is the more likely is it to be taken for a weed. For this reason, I have usually done this work myself. Fig. 84. Oenothera elliptica. An open flower, to show the rounded tips of the petals, (1895). 396 Origin of Each Species Considered Separately. their greatest breadth is about their middle or sHghtly above this and they are rounded at the tip. They are very hke the autumn flowers of 0. laevifolia. (Fig. 59, p. 312), only that they have this shape from the be- ginning of the flowering period. And just as the shape of the petals in laevifolia in autumn may be ascribed to the diminished supply of nutriment at that time of the year, so there is very probably some causal relation be- tween the shape of the petals and the narrowness of the leaves in 0. elliptic a. The pollen w^as frequently empty, but this also hap- pens occasionally in other species as in O. scintillans and even in O. gigas. It is quite normal for numerous species of Oenothera to have a large proportion of sterile pollen, as for example in O. biennis L. and O. niuricata L. The fruits of elliptica were small, and contained little seed. The origin of 0. elliptica has alread}^ been noted in the pedigrees of the various families (Part II, §§1-7). Here is a summary of these cases : FAMILY YEAR NUMBER OF O. ELLIPTICA O. Lmnarckiana, a branch of the main family, 1895, 1896 8 O. laevifolia 1889, 1891, 1893, 1894 7 O. lata 1900 1 O. lata 1890 2 I have not entered the occurrence of 0. elliptica in the pedigree of the La/warc^fana- family (p. 224) ; its occurrence in the various years in which it appeared was as follows : YEAR NUMBER OF O. ELLIPTICA 2nd Generation 1888 2 3rd " 1890 2 6th '• 1896 7 Oenothera Elliptica. 397 O. elliptica occasionally appeared in other cultures. Here are some examples. OENOTHERA ELLIPTICA. INDIVIDUALS THAT HAVE ORIGINATED BY MUTATION. NUMBER OF SEEDLINGS SOURCE YEAR TOTAL O. ELLIPTICA 3200 6 O. Lamarckiana (subsidiary cul- j 1889, 1891, tures in the laevifolia-f SLmily) I 1893, 1894 O.Lamarcktana{iTom. O.sciniillans) 1898 1080 2 O.oblonga 1896 1680 1 O. Lamarckiana X O. nanella 1899 3815 1 O. Lamarckiana X O. brevistylis 1898 290 1 O. Lamarckiana X O. suaveolens Desf. 1897 200 1 Totals 10265 12 This is a proportion of about 1 in a thousand. It appeared in similar proportions in other cultures. Alto- gether rather more than 50 mutants have arisen. This species flowered in 1890 (1), 1891 (1), 1895 ( 3 ) , 1 896 ( 3 ) , 1 897 ( 1 ) , that is, only tentimes altogether. I obtained seeds from the three plants of 1895, and from those of 1896 and 1899; in all of which cases self-fertili- zation had taken place under bags. The first plant of 1895 set seed abundantly, and gave rise to some hundreds of seedlings, wdiich grew to fine rosettes but proved however to be ordinary Lamarckiana. Many of them flowered in their first summer, but others passed through the winter as rosettes. The second mutant had about 500 offspring; one of these was an 0. elliptica which had become a fine rosette by the middle of August but was then killed by a cater- pillar in the soil. The remaining seedlings were normal Lamarckianas. 398 Origin of Each Species Considered Separately. The third plant of 1895 set little seed and only gave rise to 27 seedlings not one of which was an elliptica. The mutant of 1896 was an extraordinarily beautiful plant with narrow leaves and narrow elliptical petals, and altogether absolutely unlike an ordinary Oenothera. Its fruits were long and thin and contained but few fertile seeds. 32 seeds germinated; 27 of the plants they gave rise to were O. Lamarckiana, the remaining 5 were 0. elliptica — that is about 15 %. These five plants devel- oped stems, but did not flower till November : they were exactly like their parent. Their leaves did not exceed 2-3 cm. in breadth, the petals were elliptical and without the emargination at the tip. They did not set seed. The last mutant which bore seed was a plant which arose in 1899 from seed of 0. scintillans. It appeared in the culture of 5850 rosettes (p. 388) which gave 40 % O. scintillans in the third generation. This culture con- tained only one 0. elliptica, which, as it was transplanted early, grew up into a plant which branched profusely and flowered freely but was of rather low growth. Its leaves were very narrow but its flowers relatively large. The breadth of the petals on this plant was highly vari- able. Its fruits were slender and contained but little seed. Abut 100 seeds germinated, but gave rise solely to ro- settes of O. Lamarckiana. To sum up: the hereditary coefficient for O. elliptica was 0 in three cases, 1 per 500 in one case, and about 15 % in the remaining one. The first three plants had only a few hundred offspring between them, and this fact in itself may be sufficient to account for the non-appear- ance of cllipticas amongst them. If this is really the case the last two mutants (with 0.2-15 %) may provisionally be regarded as representing the normal. Oenothera Sublinearis. 399 § 21. OENOTHERA SUBLINEARIS. This form differs from the last named chiefly by its grass-hke leaves which are very narrow and of equal breadth along their whole length (Figs. 85 and 86). The foliage leaves are longer and markedly narrower ; the Fig. 85. Oenothera sublinearis. Two annual plants, at the end of August 1900. A, with out, and B with flower buds. Fig. 86. Oeno- thera sublinearis A radical leaf, 1895. foliage on the stem is dense and not scanty ; the fruits are short, and not slender as in O. ellipfiea. Although I have had very few examples of this species so far, it is evi- 400 Origin of Each Species Considered Separately. dently a genuine well-characterized type; the herbarium specimens and photographs I have kept of the first that appeared agree perfectly with the mutants which have arisen since. The flowers presented no differences from those of O. elliptica. They are the same size, that is, are some- what smaller than those of 0. Lamarckiana, but large when the weakness of the species is taken into considera- tion. The petals are not obcordate but narrower at the extremity, and rounded or sometimes even pointed at the tip. The stamens and stigmas resemble those of the parent species. ^^^ ^^^^ Examples of 0. snblinearis ap- i^^^l ji^^^^k peared in my cultures from year to \ WKKKU year, but the majority of them ' '^ / / ^^y^ perished as young rosettes. Only ^X^^y^ ^our plants grew beyond this stage ^ ^m and only one of these afforded fer- Fig. 87; Oeiiotherasub- ^ile seed, which when sown pro- linearis. retals with a . ^ stamen, July, 1896. duced the new form in the pro- From the same bien- ^^^4.: ^r m «9/ t ^.i • 4. nial plant as Fig. 86. portion of 10 %.^ In this respect it falls therefore into the same cat- egory as O. scintillans and 0. elliptica. The history and fate of the four mutants which pro- duced stems must now be briefly described. I shall be- gin with the single plant which set seed. This plant arose from seed of the Lamarckiana- family which had been sown in 1895, but had remained in the ground for a year. It was recognized in June 1896 as a peculiar form and transplanted separately. It was biennial and flowered in 1897 on its numerous lat- eral branches which however bore only a few flowers each. The whole plant was short and stunted, and its Oenothera Siihlinearis. 401 flowers were relatively large. They were self- fertilized in parchment bags. But the harvest was very scanty. Only 31 seeds germinated, and these were transplanted with the greatest care and cultivated further. The composition of the progeny was by far the most varied that I have observed ; there occurred : 19 0. Lamarckiana 1 0. alhlda 3 0. subline aris 3 0. suhovata 1 0. lata 10. gig as 1 O. nanella 2 0. ohlonga I weeded out the Lamarckianas as strong rosettes at the end of June, when there could no longer be any doubt as to their identity. The O. snhlinearis and 0. subovata remained in the rosette stage and died in the winter. All the remaining plants flowered, some in August and September, and some (0. gigas) in Novem- ber of the same year. Their identity with plants grown from the seed of mutants of the same name was fully established especially in the case of the rarer forms 0. albida and 0. gigas. This extraordinary richness in mutants is probably connected in some way with the smallness of the harvest as was believed to be the case in the experiment described on page 264. This highly important point needs further investigation. The second plant belonged likewise to the Laniarc- kiana-fd.m\\y ; it appeared in 1895 and flowered in 1896. One of its first year's radical leaves is shown in Fig. 86, two of the petals which it bore in 1897 are shown in Fig. 87. The plant was pale green and so weak that there seemed very little chance of its surviving the win- ter. But it flowered rather well : there were about a 402 Origin of Each Species Considered Separately. dozen flowers on two stems which arose from the axils of its radical leaves. These attained a height of about half a meter. But in spite of all the trouble I took I got no fertile seed from it. The third mutant arose in 1900 in the first Lata- family, as already recorded in the genealogical table on p. 285. It is figured in Fig. 85 B. It was planted out in June, grew well, but remained short and did not branch. It bore large flowers and small fruits and was cut off at the end of August to be photographed. The fourth mutant (Fig. 85^) arose from a cross between O. rnbrinervis and O. nanella, which was made in 1899. It developed an unbranched stem, which at- tained a length of about half a meter, in its first year, but it did not flower. D. THE STERILE SPECIES. § 22. OENOTHERA LATA. One of the most difficult questions which the muta- tionist has to answer is that which refers to the nature of the fundamental process, involved in mutation the visible results of which are the peculiarities and char- acters by means of which the new form is distinguished from the parent species. I have already laid stress on the fact, which has not escaped the notice of the best workers in this field, that elementary species are not dis- tinguished from one another by one character only, as varieties are, but by almost all their organs and charac- ters. This is not only true of the elementary species, in a state of nature, which have been described by Jordan^ Gandoger, Thuret, De Bary, Rosen and many others but also of those which have arisen in my cultures. Oenothera Lata. 403 I hold that all the new characters of a mutant are manifestations of a single change that has taken place within it. Morphological proof of this thesis can as yet hardly be produced, but physiologically it follows of necessity, in my opinion, from the fact that these char- acters are always associated and, so far as our experience goes, cannot be separated. Fig. 88. Oenothera lata. A lateral branch at the end of August just opening its first flower. Oenothera lata is perhaps the most beautiful example. I have already described the characters peculiar to it in § 3 on pages 239-243 (Fig. 46), but I propose now to elaborate that description. In the first place it is one of 404 Origin of Each Species Considered Separately. the commonest mutants, and at the same time one of the most easily recognizable in its early stages. It appeared 229 times in the main line of the Lamar ckiana-id.m\\y (p. 224), in the branch family 171 times, in the laevi- /o/m-family 9 times and very frequently in other cultures too. I have cultivated many such mutants until they flowered and set seed ; in every case they conformed ex- actly to a common type. No separation of the characters of the species has been observed. Oenothera semilata (§ 17) which ap- peared at first to be an instance of this, turned out to be a distinct form. The characters of the species can be regarded as distinct "groups," better in the case of O. lata, than in the case of any other species. Each ''group" obviously constitutes a unit, but how the existence of the separate ''groups" is brought about by the same cause is as yet unknown. Examples of these "groups" are, the form of the leaves, the thick flower-buds, the lack of pollen, the abnormal growth of the pistil, and the short fruits w^ith relatively few seeds. Let us look at the leaves; they are crumpled, and round at the tip ; the edge is too small for the area of the leaf which is therefore much bent. The bracts are much broader at the base than they are in the parent species. The apices of the large branches and the smaller lateral branches form peculiar little rosettes. A complete de- scription would extend over a whole page of print and need many figures. (Fig. 89.) Nevertheless it is certain that all these units are intimately bound up with each other and that they must owe their existence to the pres- ence of a single factor. Perhaps this factor is the abnormally luxuriant super- Oenothera Lata. 405 ficial growth of the leaf parenchyma in proportion to that of the nerves; but perhaps we must seek deeper for it. It is not, however, easy to see how the same cause can make the stigma abnormal, the fruits small and the pollen sterile. On the other hand if we suppose that each Fig. 89. Oenothera lata. A, a radical leaf. B, the bract, from the axil of which the lowest flower arose. C, apex of a small lateral branch. A', B', C, the corresponding parts of O. Lamarckiana diminished the same amount. of these characters is due to a distinct cause we have no means of accounting for the fact that they always appear together and never one at a time : for this clearly could not be due to chance. I imagine that the cause of every such a mutation 406 Origin of Each Species Considered Separately. is a single one ; though its real nature is not yet apparent to us. But this much is clear, that it is not, at least as a rule, manifested as a single quality. And in this respect a mutant differs from a variety, in which single qualities like color, hairiness and so forth form the diagnostic character. In the mutant this quality can only be mani- fested in connection with the older characters of the plant; and the total expression of this cause must there- fore depend partly on the older characters and only partly on the new factor itself. If we look at it in this way we can easily imagine how a single internal change can bring about the ab- normal development of the leaf parenchyma, of the pollen cells of the anthers, of the petals, the fruits and the stig- mas, and in this way produce the broad crumpled leaves, the sterility of the pollen, the thickness of the buds and the abnormal stigmas. This is of course only an idea. I mention it partly to simplify the problem and partly because it may indi- cate the lines along which the problem may be dealt with empirically. In order to make my position clearer I will now tentatively indicate the parallel which exists between this idea and certain phenomena of parasitism. It is generally admitted and, indeed, there can be little doubt that the wonderfully definite and complicated structure of the Cynipid-galls, with their nutrient tissue, their stone-cells and their spongy tannin-containing outer par- enchyma^ whose thickness is adapted to the length of the ovipositors of parasites and Inquilinae, cannot be the result of a single chemical stimulus. But it is an entirely different matter with the cases of virescence since these ^ M. W. Beyerinck, Bcobachtungen i'lher die crsten Entwiche- lungsphascn ciniger Cynipidcn-Gallen. Verh. d. K. Akad. d. Wet., Amsterdam, 1882. Oenothera Lata. 407 are evidently only useful to the parasites in quite a gen- eral way. The virescence of Lysiiuachia vulgaris, which is occasioned by a Phytoptns, affords perhaps the most beautiful example of a complete series of transitions from flowers to leafy branches.^ This change is obviously the object of the stimulus given by the Acarine and it ob- viously does not matter whether the number of the changed petals of the corolla varies or not. Neverthe- less these and other monstrosities accompanying the vi- rescence are by no means rare. The case of the virescence on the galls of Anlax Hieracii, in the flowerheads of Hieraciiim viilgatwn, H. tanhellatiim, etc., which have been studied by Treub, is very instructive.^ These galls are usually situated in the stems far away from the flower, but in rare cases they occur in the flowerhead itself. When this happens the flowers are affected by a whole series of the most re- markable malformations which begin by the calyx pro- ducing little green sepals. These changes are obviously of no use to the Cynipids which live inside the galls ; for the Aidax larvae grow just as well if no inflorescences are borne on the galls. Galls not rarely evoke monstrosities of this kind, provided of course that the potentiality for these mon- strous growths already exists. I have found, for ex- ample, a stem of Hieraciimi vulgatiim which was normal below the gall but, above it, was broadly fasciated. In the summer of 1887 I saw several stems of Enpatoriiim cannahmum bearing, about their middle, galls of Pter- ophorns rnicrodactyhis : below these the leaves were green ^ A. B. Frank, PUanzcnkrankheitcn, 1880, p. 691. ^ M. Treub, Notice sur Vaigrette des Composccs. a propos d'linc monstruosite de rHicracium iimbcUatum, Archives Neerlandaises d. sc. phys. et nat., T. VIII, p. i and Plate I. 408 Origin of Each Species Considered Separately. but above them they were variegated. The gall stimulus, therefore, does not exert its influence only on those qual- ities which are necessary for the formation of the galls, but on others as well. The effect of a single mutation on the most diverse and important, as well as on subsidiary, qualities may be of a similar nature as that of a gall stimulus. But if it is difficult to discover the chemico-physiological nature of the gall stimulus ; it is wellnigh impossible to penetrate into the mystery of the chemical nature of a primary mutation. Let us now make a more detailed study of the char- acters of our Oenothera lata and let us begin with the stamens. The anatomical structure of these has been in- vestigated by Prof. J. PoHL^ (Fig. 90), partly on the plants of my first /a /a- family (p. 285) in 1894, partly on a larger culture which I raised in that year from the seeds of the second /a/a-family (seeds of 1889 and of 1890, see p. 288) and partly on isolated new mutations. The structure of the stamens was the same in all cases; and was, therefore, independent of the ancestry of the plant. Pollen formation takes place in Oenothera Lam arc- kiana and O. lata in the ordinary way ; the mother cells enclosed in the tapetum each divide into two daughter cells and each of these into two granddaughter cells. The loculus increases in capacity by the dissolution of the tapetum ; and the further development of the pollen- grains takes place in the fluid which now surrounds them. The ripe pollen of 0. Laniarckiana consists of two forms ^Julius Pohl, Ueher Variationsweite von Oenothera Lamarc- kiuna; Oesterr. Botan. Zeitschr., 1895, Nos. 5 and 6, and Plate X. — See also R. R. Gates^ Pollen Development in Hybrids of Oenothera lata; in Botanical Gazette, T. 43, p. 81. (Note of 1908.) Oenothera Lata. 409 of grains, about 70% large normal grains,^ the rest being small grains poor in protoplasm. The pollen of 0. lata on the other hand consists of crumpled, distorted grains which form every conceivable transition between abso- lutely empty sacks and apparently normally developed grains. But the empty ones and the almost empty ones are in the majority ; the apparently well developed ones occurring only sparingly amongst them. Moreover the viscin threads, which in O. Lamar ckiana connect all the pollen-grains together so that they form a sticky mass, appear to be absent. The open an- thers feel dry and if they are touched with the finger, bits of sticky masses of pollen do not adhere to it. If we follow the develop- ment of the anthers in O. lata in a series of buds of increas- ing size we find that develop- ment is normal up to about the stage of tetrad - formation. Shortly after this stage disso- lution of the tapetum takes place and there are found floating in the lumen, besides apparently normally developed tetrahedral pollen grains, some quite round, others invaginated on one side. I devoted a great deal of time in 1894 to trying to fertilize O. lata with its own pollen, trusting that the few apparently good pollen grains which I had found would be able to effect fertilization. I plastered as much Fig. 90. Oenothera lata. Transverse section of an anther, showing the large cells of the tapetum. After J, PoHL, Oesterr. Bot. Zeit- schrift, 1895, Plate 10, Fig. 28. ^Figured by Luerssen" in Pringsheim's Jahrhiich., Vol. VII, pp. 35-42, and Plate IV, Figs. 1-14 (Pollen of Oenothera biennis). 410 Origin of Each Species Considered Separately. pollen as I could on the stigmas of a few flowers ; but all to no purpose. Then, as it was difficult to liberate the pollen direct from the anthers on to the stigma, I teased it out with needles on to a glass slide, collected it in a lump and transferred it direct from this to the stigma — but, again, all to no purpose. After this I pollinated the flowers, without castrating them, with the pollen of a remotely related plant, be- longing in fact to another subgenus; O. odorata.^ I have obtained fertile crosses between this form and 0. Lamarckiana, O. biennis and O. miiricata. It ought, then, to be able to fertilize O. lata also; but my attempts to effect this were practically without result although I pollinated many flowers on four plants. Only a single seed germinated ; and this produced a hybrid plant. This experiment also shows that self-fertilization did not occur. Besides this I have pollinated castrated and non- castrated flowers of O. lata with the pollen of O. La- marckiana. I have also tried the effect of putting very little Laniarckiana-poWcn on the stigmas of uncastrated flowers in the hope that perhaps I might induce self- fertilization in that way. All these experiments gave exactly the same result; about 15-20% of the seeds gave O. lata, the rest 0. Lamarckiana. I conclude from these and from a number of other experiments that the pollen of O. lata, in spite of the presence of occasional apparently good grains, is never- theless absolutely sterile. One result of the establishment of this fact is that the castration of flowers of 0. lata in hybridization experiments becomes unnecessary. ^ From the subgenus Oenothera (Euoenothera) ; whilst O. La- marckiana etc. belong to the subgenus Onagra. Oenothera Lata. 411 If there had been evident signs of the existence of pollen in individual flowers among the numerous mutants and their offspring which I. have artificially fertilized during the course of the last six years, I must certainly have seen it. But" this has never been my lot.^ The stigmas have also been figured and described by PoHL.^ They differ from those of 0. Lamarckiana by their tendency to be confluent with one another and with the style. Their number is variable, as in the parent species, where 4 is the normal ; but numbers up to 8 are Fig. 91. Oenothera lata. Young seedlings. A, showing the cotyledons and the two first leaves. A', natural size. B, with 7-8 leaves (Vs) two months old, seen from above. The tear in the leaf to the right was caused by trying to bend the leaf flat. not rare. As result of the concrescence just mentioned, there arise in O. lata curious hand-shaped deformities. The individual fingers of these hands are sometimes free but sometimes fused to their tips. This deformity goes hand in hand with a shortening and thickening and also with a crumpling of the individual stigmata. The capa- city for taking pollen and for permitting the normal development of the pollen tube, doe§ not, however, seem to have been impaired by these malformations. ^ As already stated. I have lately raised a hybrid of 0. lata which produces some fertile pollen, which I have now in cultivation. See Section I, §3. (Note of 1908.) ^ Julius Pohl, /. c. p. 8 and Fig. 27. 412 Origin of Each Species Considered Separately. The fruits are short and thick and contain Httle seed. They attain scarcely half the length of those of O. La- mar ckiana but are almost as stout as these. I first observed the above mentioned deviations from the type of O. Lamarckiana in 1887 on the very first mutants, as far as they could be discovered without mi- // / Fig. 92. Oenothera lata. Rosette with radical leaves. Aged about 3 months. croscopical investigation ; and since then I have observed them every year not only in the new mutants but also in their progeny. Like O. Lamarckiana, 0. lata is both annual and bi- ennial. But I grow it as an annual by preference. The first two leaves which appear after the cotyledons plainly Oenothera Lata. 413 reveal the identity of the species (Fig. 91 A). The tips of the leaves are rounded and not pointed, which makes them shorter. About a month after germination this character is so clearly expressed that I choose this stage for sorting out the Lamarckianas from the latas in the results of crosses. This form of the leaf is maintained through the whole life of the rosette (Fig. 92) and on the lower part of the stem. The wrinkling and distortion which so detract from the beauty of the leaves in O Lamarckiana are much more pronounced in 0. lata; and are very rarely absent (Figs. 57 and 58, pp. 310 and 311). This feature may be brought about by the relativel}^ small margin of the leaf. On the whole, the abnormal breadth of the leaves is maintained up the stem even to the tops of the inflores- cences and branches (Fig. 89). But, as in the case of Lamarckiana itself, the leaves become gradually more pointed and narrower the further up they are. Our figure (Fig. 89 A, A') brings this out very clearly; a fine point is seen on the otherwise rounded tip of the leaf. If we look at the lowest leaf which bears a flower in its axil or an immature fruit and compare it with a corresponding leaf on a Lamarckiana plant (Fig. 89 B, B'), we shall find that the relative breadths are as 4 to 3. Higher up in the inflorescence this difference in- creases ; the leaves, from whose axils the flowers which open in August arise, are about twice as broad as the corresponding ones in the parent species. And if we look at a small branch from above it looks like a thick rosette of broad leaves (Fig. 89 C) whereas in such a view of Lamarckiana the leaves are reduced to small and narrow bracts forming a kind of rosette of pointed leaves. 414 Origin of Each Species Considered Separately. The tops of the flowering branches are also densely clothed with leaves (Fig. 88). The remarkable thickness of the buds is clearly brought out in our figures (Fig. 46 on p. 241). The petals have not sufficient room for development in the thick but short bud: they acquire in this way folds and wrinkles which, even when the flower is fully open, are never completely lost. As a result of this the flowers are always rather unattractive, and not nearly so large and bright and widely opened as in the parent species. The stem and branches in O. lata are weak, often bent downwards with the tops heavily laden and usually needing a stake to prevent their falling over. The lateral branches flowering in September are frequently seen to hang downwards; thereby heightening the characteristic appearance of the species. The plants do not as a rule attain a great height; seldom more than half that of O. Lamarckiana. With all these peculiarities O. lata is perhaps that one of the new species which differs most widely from the parent form. Moreover it can be recognized in its ear- liest youth no less certainly than easily (Plate IV and Fig. 48 on p. 280). Doubtless as a result of these cir- cumstances it Avas the first mutant which I noticed, and the only one wdiich I found in my first crop of 1887. Since that date it has appeared every year as a mutation. And as it can be seen so early in each crop and as, therefore, it is not likely that it will be overlooked to any large extent, the proportions in which it appears may be regarded as established on a sufficiently firm basis to admit of a comparison between the small differ- ences in its "mutation coefficients" (See p. 338). I found that these numbers vary considerably, often sink- Oenothera Lata. 415 ing to 0.01% or less, or mounting to 2% or more. Ex- ternal conditions therefore probably affect the propor- tion in which O. lata arises from O. Lainarckiana. What these conditions are is a subject for future enquiry. Perhaps they exert their influence only during the ripen- ing of the seed or during germination (See p. 263) but probably they come into play at or before fertilization. In order to give some idea of the range of variations of these ''mutation coefficients" I give a list containing the figures which have already been given (§§ 2-7) to- gether with some new observations. INDIVIDUALS OF OENOTHERA LATA WHICH HAVE ARISEN BY MUTATION. I. FROM O.LAMARCKIANA. LAMARCKIANA FROM: DATE TOTAL SEEDLINGS O. LATA % LATA Main line of descent, p. 224 1888-1890 25,000 8 0.03 " " '• " 1895 14,000 73 0.5 " " " " 1896 8,000 142 1.8 " " " " 1897-1899 3,500 6 0.2 Lateral branch, p. 262 1895 10,000 168 1.7 An annual culture 1897 4,132 11 0.3 A biennial culture 1897 164 8 5.0 II. FROM CROSSES. O.Lam.x0.7ia7iella 1897-1899 8,283 22 0.3 O.Lam.xO.gigas 1899 100 2 2.0 O. Lam. y^O. biennis 1900 80 1 1.0 6>.Z,aw. (from crosses; p. 300) 1896 4,600 7 0.2 III. FROM OTHER FAMILIES. O.Lam.irom O.laevifolia 1889 400 3 0.8 O.laevifolia 1894 1,500 2 0.1 O.rubrifiervis 1894 96 2 2.0 O.scintillans 1896-1899 7,872 38 0.5 If we examine these figures closely we make rather an interesting discovery. A high figure (5%) is only 416 Origin of Each Species Considered Separately. given by a culture of strong biennial plants, carried out with the utmost care (1897). Three of the figures, viz., 1, 2, 2% are afforded by too small crops to be of any significance. On the other hand the cultures of 1895 and 1896 which involved 8000, 10,000 and 14,000 plants respectively, and may therefore be relied on, gave 0.5, 1.7 and 1.8% apiece. These, then, are the most reliable figures with which the rest, with the exception of the first named (5%) agree very well. The remaining fig- ures, which are 0.3—1—2—2—3—3—5—8 per 1000, were either obtained in earlier years or in special cul- tures. In the whole table there are 493 /a/a-mutants amongst about 130,000 seedlings, or about 0.4%. § 23. INCIPIENT SPECIES. According to the mutation theory, natural selection chooses between species ; some are eliminated by it, others permitted to increase and multiply. The new forms arising from a single parent species by mutation may be very numerous; they are often equally well equipped for the struggle for existence being distinguished from one another by characters which are unimportant in this respect, as in the familiar case of Draba verna. But should there have arisen from Draba verna, be- sides the elementary species now existing, others less fitted for the struggle for existence, they would most certainly have been eliminated sooner or later. And there is no reason for supposing that elementary species different from those now existing have not been pro- duced. I have seen almost every year in my cultures of Incipient Species. 417 Oenothera unfit mutations of this kind, and very often in considerable numbers. For the sake of completeness I shall describe some of them here. They are not dis- tinguished from the ''fit" new species by any sharp line of demarcation, and perhaps there exist amongst them some forms from which, by the help of better methods, it may some day become possible to obtain constant types. They are the beginnings of new species from which, for one reason or another, I have not succeeded in ob- taining the species themselves. For example it was only after a series of the most laborious experiments which extended over about 6 years that I was able to get O. alhida to flower and set seed (§ 15, p. 349). I shall therefore refer to these forms as "incipient species."^ My incipient species were more or less aberrant types and exhibited few points of resemblance with the new species hitherto described. Of late years I have devoted much energy to their cultivation, but with varying results. Many of them died as young rosettes; others formed fine thick clusters of root-leaves, but developed no stem. Some I was able to bring through the winter, others per- ished in their first year. Many of them flowered, in some cases as early as August, in others not till autumn. If the latter happens there is no prospect in our climate of ripening seed ; when the former was the case I always enclosed the flowers in parchment bags in order to insure self-fertilization. But as the pollen was usually sterile the operation was commonly fruitless. I then tried using the pollen of 0. Lainarckimia or that of some other new species but with no better success; the ovaries seemed incapable of being fertilized. Sterility is well known to be a highly variable char- ^ Ebauches d'espcces, of some French authors. 418 Origin of Each Species Considered Separately. acter. It certainly is in the Oenotheras. The older species 0. biennis, 0. muricata and 0. Lamarckiana al- ways produce, so far as we know, a pollen of which some part, often as much as one-third, consists of sterile grains. It would be very useful if some one would de- termine the degree of this fertility; it would without doubt follow Quetelet's law of fluctuating variability and would probably exhibit also partial variability in a high degree — since the percentage of sterile grains would be likely to be high on weak lateral branches.^ Sterile or almost sterile individuals may therefore appear from time to time. For example I once found a plant of Oeno- thera gig as, which, in spite of repeated attempts to fer- tilize it with its own pollen, set no seed. And Oenothera hrevistylis is much more often than not absolutely sterile, in spite of the full development of its pollen; and this sterility is closely correlated with the individual varia- bility in the size of the fruits. It may happen that a new species absolutely lacks pollen, as we have seen to be the case with O. lata. But it does not follow from that, that every new form that arises, if we find it first as a sterite plant, must, when it arises once more, be sterile again. The incipient species in my cultures were as a rule represented by solitary individuals. In rarer cases the new form was represented in the same crop by two or three seedlings; or was repeated in succeeding years. If nothing more than rosettes of radical leaves were pro- duced, absolute certainty as to the identity of the type was of course out of the question ; but it is always better ni cases like these to unite those which apparently belong ^ See amongst others A. Jencic, Untcrsuchiingcn i'lher den Pol- len hybridcr Pftanzen, Oesterr. Bot. Zeitschr., T. 50, 1900. Incipient Species. 419 together than to multiply new types indefinitely. For these unsuccessful incipient species have hardly any fur- ther signification than that of supporting the thesis of indiscriminate mutability. This is my chief reason for describing here in some detail a few such incipient species. For this purpose I select three of those which I have noticed, and shall call them for the sake of convenience by ordin- ary specific names. They are O. spatJiu- lata, of which I obtained only rosettes of radical leaves, O. stibovafa which flowered several times, but was always sterile, and O. fatna which though it branched freely has produced practically no flowers so far. Besides these three types there was a whole series of other forms which do not seem to me to be worth either naming or describing.^ Of 0. spathidata I obtained two ro- settes of root-leaves in the laevifolia-ia.m- ily in 1889, (p. 273) ; seven rosettes in 1890 in the main pedigree of the Lauiarc- kiana-f am'ily; and one in a lateral branch Fig- 93- Oow- of this famny m 1895 (p. ZoZ). lata. A rad- Repeated appearance in different and ^ rlt^dyp mutually independent families is thus es- tablished in the case of this rare new species. The plants of 1890 were fine strong rosettes about the end of June: they grew healthily through the whole summer but died in the winter without having developed a stem. I have kept some of their leaves and photographed them (Fig. ^ Some of my new species have not arisen from the pure stock of O. Lamarckiana but have arisen from crossed seeds of various an- cestries. Both sterile and fertile forms have arisen in this way. 420 Origin of Each Species Considered Separately. 93). The petiole was very long, gradually merging into the blade of the leaf; the latter attaining its greatest breadth near its rounded apex. Another mutant with similar leaves flowered in the same summer; it had little flowers and empty anthers ; but I was not sure whether it belonged to the same type. I gave the name 0. fatna to a plant which arose in 1896 from the Lamar c- kiana-idiXmly : it proved a biennial and branched profusely in 1897. It bore numerous inflorescences with green bracts, but no flowers (Fig. 94). In the summer of 1896 I had isolated the rosette as a new form; the leaves were oval and obviously different from those of 0. Lamarckiana. It grew vig- orously in the second year, attained a height of about a meter, produced more branches than any other form I have seen and developed the most extraordi- nary profusion of inflorescences. It was not until late in the autumn that it began to develop normal flowerbuds, too late for them to open. I have observed isolated examples of similar plants on other rare occasions. My last example is Oenothera siib- ovata which first appeared in 1889; and, afterwards, as isolated examples in various cultures, from time to time. Four of these mutants have flowered ; the rest died in the rosette stage. • Fig. 94. Oeno- thera fatua. A branch in au- tumn with nu- merous flower- less bracts. Incipient Species. 421 Fig. 95 is a photograph, taken in the same way as those already given for other mutations, of a group of young plants which were raised from the seeds of Oeno- thera lata fertilized by 0. Lamarckiana, and were planted out in early spring, in rows, in boxes containing good Fig. 95. A mutation in the /ato-family. From a photograph taken on the 25th of May 1900. In the right column there can be seen at the top O. alhida, in the middle O. lata, and at the bottom O. albida again. In the second column O. Lamarckiana, O. oblonga (in the middle) and O. Lamarckiana. In the third row O. lata (below), O. oblonga, and O. subovata (at the top, very small). In the left row three O. lata, and in the middle, the largest rosette, O. Lamarckiana. The O. lata and O. Lamarck- iana are like the two parents, the rest have all arisen as mutations. soil. It shows some of the mutations which arose from the main stem of the /a/a-family (as given in detail on page 285), and happens to be a group in which many new forms are growing close together. The figure shows 422 Origin of Each Species Considered Separately. besides 0. lata and O. Laniarckiana two examples of O. albida, (to the right), recognizable by their small narrow leaves, and two of O. oblong a (in the middle) which can hardly be distinguished from the surrounding ex- amples of the parent species in the picture. The major- ity of the /a/a-plants photographed flowered afterwards. The albidas and oblongas grew as far as the rosette stage, but died in the course of the summer. The La- marckianas were not planted out. The O. subovata (the second from the left in the top row) was noticeable very soon after transplanting by the fact that it remained very small whilst the other plants grew vigorously. Its leaves were almost orbicular and were shortly petiolate. It was planted out at the end of April in a separate bed with the other mutants, and grew up to a strong full rosette with numerous ovate leaves with long petioles which distinguished it at once from all the other plants in the same bed. It died in the autumn. The two other examples of 0. subovata (mentioned on page 285) died in the latter part of the summer after they had formed thick rosettes similar to those al- ready described (see also Fig. 48 on page 280). I had, before this, observed one or two instances of rosettes with the same form of leaf and of the same general appearance. For example in 1895 I observed seven of them amongst 14,000 seedlings of the main trunk of the Lainarckiana-isLmily, that is to say 1 per 2000 (p. 224). They survived the autumn, but not the winter. Then there were the three mutants (mentioned on page 401) which arose in 1898, from the seeds of O. snblinearis and two others which arose from 0. scintil- Incipient Species. 423 lans. And last of all, two rosettes from seeds of Oeno- thera lata fertilized by O. biennis. I obtained altogether four flowering plants of O. siib- ovata, one in 1889 and 1895, and two in 1899. The former arose from Lamarck- iana-SQcds, was annual, freely branched but dwarfed. It de- veloped not only a main stem, but lateral ones from the axils of its radical leaves. Only a single one of these latter bore normal flowers like the parent species. The other lateral branches and the main stem however were quite sterile. In- stead of flowers in the axils of the leaves there were little green leafy shoots (Fig. 96) which gave the plants a most singular appearance.^ The mutant of 1895 arose from the Larnarckiana-isimily and flowered in August of its first year. The flowers were of the same form as those of the parent species ; but, in corre- spondence with the greater deli- cacy of the plant, smaller. Later on, however, it became stronger and the flowers which were borne on its lateral branches at the end of Sep- tember were quite as large as those of Laiiiarckiana. * This foliation was due to internal causes and not a pathological virescence like that which may be brought about by parasites (Pliy- toptus, plant-lice, etc.). I have occasionally seen examples of this Fig. 96. Oenothera sub- ovata. A barren stem, 1889. 424 Origin of Each Species Considered Separately. The two snhovata-y\2ints> of 1899 belonged to the /a/a-family. One arose from lata crossed by O. nanella, the other from lata crossed by O. Lamarckiana. Both were recognized while young plants; they developed stems and flowered freely. The first was weak and had relatively small flowers; the second was strong and bore flowers as fine as those of 0. Lamarckiana. Both were absolutely sterile, at first I fertilized them with their own pollen and later with foreign pollen, but in both cases without result. But I regard it as extremely probable that I shall some day succeed in getting a siihovata that will bear seed. And if siihovata', why not other forms which have not been detected, or even have not yet arisen in my cultures ? latter in my O^wof/i^ra-plants (Botan. Jaarhoek; Gent, 1896, p. 88). But the effects of it, especially the actual virescence of the flowers themselves bore no resemblance to the peculiarities of the plants described above. III. THE SYSTEMATIC VALUE OF THE NEW SPECIES. § 24. THE NATURE OF THE BOUNDARIES BETWEEN RELATED SPECIES. It follows directly from the doctrine of mutation that the species which arise by mutation are as sharply distinguished from one another as are neighboring spe- cies of recognized systematic value. The previous chapter has shown us that new species are, as a rule, from the very beginning as constant as other species. This constancy is manifested in two ways. In the first place the various individuals of the new spe- cies are absolutely alike in all their essential features: and this is true not only of the offspring of new muta- tions and of others which arise from the same parents, but also of the hosts of mutations which have arisen from widely separated and independent families of the same parent species. In the second place they come true to seed, they do not revert to the parent form. If the latter character is absent as in the case of Oenothera scintillans the new form cannot be capable of existence in nature and therefore cannot be compared with true wild species. The object of the present chapter is to show that the characters of the new species which have arisen in the 426 The Systematic Value of the New Species, genus Oenothera possess the same systematic value as those which distinguish the species of Linnaeus and later systematists. I shall confine myself to the nearest rel- atives of 0. Lamarckiana, that is to say, to the subgenus Onagra. In this way I shall have the advantage of deal- ing with very well known forms (0. biennis L., 0. muri- cata L., 0. suaveolens Desf., etc.) and of using them as subjects of comparison. The new species differ in some respects as much, in others more, in yet others less from one another and from the parent species than these recognized forms do amongst themselves. And first I must emphasize two things which make the treatment of this subject a difficult matter. I mean, our present wholly insufficient knowledge of the units of which the characters of organisms are built up, and secondly the phenomenon of transgressive variability.^ I believe that each new mutation is brought about by a single new quality (see page 403). This inner or primary quality then comes in contact with the qualities which are already present in the various organs and it is this interaction to which its particular external mani- festation is due. The nature of the outward and visible form therefore depends only partly on the mutation but partly also on the characters already existing in the organism. Or, in other words, the new species is marked not as a rule by a single new peculiarity but by the trans- formation of many or all of its organs, more or less. So long as we do not know the single causes in ques- tion we must compare these visible transformations in new mutations with the visible dififerences between old established species. Transgressive variability is one of the main supports * See page 56, and the following section. The Boundaries Betzveen Related Species. Ml of the current theory of selection. It makes it possible to pick out series of individuals which belong to related but different species and (being careful to choose suit- able characters) to arrange them in such a way that they form a perfectly continuous series from one end to the other. If no gaps have been left in such groups by the death of species or if there are sufficient species left, perfectly continuous series of this kind can be arranged of almost any desired length. But, as a rule, this can only be done by dealing with single characters and by not being particular about the number of individuals which go to form the successive steps in the series. An example will make my meaning clear. Oenothera Lamarckiana differs from 0. biennis by the beauty and size of its flowers. The two species can be distinguished at a great distance. The petals of the former are twice as long as those of the latter. But in both species the leno^th is variable and follows Ouetelet's law of indi- vidual variability, being in a high degree dependent on nutrition. Petal-length also exhibits partial variability and is especially low at the end of the flowering period when the plant is exhausted by bearing seed. The small- est flowers are found on the main stems of plants which are nearly over, on small lateral branches or on indi- vidually weak plants : the largest on well nourished plants just beginning to flower, or on vigorous lateral branches of large plants which through some accident or other have lost their main stem. This is true both of plants in the field and of those in the garden. If now we choose the largest flowers of 0. biennis and the smallest of 0. Lamarckiana, we shall find tiiat 428 The Systematic Value of the New Species. this character of petal length overlaps in the two cases. ^ For in these extreme cases the flowers of biennis are actually larger than those of Lamarckiana.^ If we have collected a number of flowers including such extremes, it is obviously an easy matter to arrange an unbroken series with the smallest of biennis at one end and the largest of Lamarckiana at the other. The limits between the two species cannot be detected in such series even by the practised eye. And yet 0. bie finis and O. Lamarckiana never merge into one another. If we wish to extend the series we can do so by adding the small-flowered 0. miiricata to it in exactly the same way.*^ And, if we leave relationship out of account, we could extend continuously down to Oenothera mimiti- flora with its flower no longer than a millimeter. Such series can be arranged for almost any character in the vegetable kingdom and in infinite variety.^ They con- fuse the limits drawn between related species as far as the several characters are concerned. If, in the classification of animals and plants, we fix our attention on a single character we shall always en- counter long unbroken series of this kind. The shells of snails and the wings of butterflies are examples. It * That this must generally be the case may be derived from the law of variability. Imagine two curves of variation drawn on the same abscissa. The greater the number of individuals included the further will the limits of the curves extend until they, first, touch and ultimately overlap. It is further obvious that the likelihood of this happening, even with a small number of individuals depends directly on the closeness of the tops of the curves (the mean values of the characters) and on the amplitude of the curve, or the degree of var- iabihty (Q). ^ Examples in the following section. ^ See the following section. * For example the narrowest leaves of Typha latifolia are nar- rower than the broadest leaves of T. angustifolia. The Boundaries Between Related Species. 429 is only when we compare other characters as well that we can distinguish the individual species. The object of exact inquiry should be first to collect as many of these continuous series as possible ; but second to anah^ze them into their component units. This analysis can be effected both by the statistical and by the experimental method. Let us examine the former first. A transgression of the limits is only exhibited by isolated and relatively few individuals; the vast majority belong to the mean type of the species. Therefore if we take care not to be too much on the lookout for transi- tional forms, or even, if we try to make our measure- ments as numerous as possible, the separate curves will become discernible. The result will be exactly that which we got from an investigation of the Oenothera flowers, namely curves with numerous apices, such as those with which the work of Bateson^ Ludwig and others has made us familiar. Each apex indicates a group of indi- viduals which belong together, to a type, or even to an elementary species. Transitional forms can then be recognized imme- diately by their rarity. It becomes obvious that the transitions are only apparent and that no real continuity between the different centers of variation exists. These curves show no more than that the edges of neighboring curves on the same abscissa may overlap. The simplest way of making the experimental method clear is to take the case of the Oenothera flowers again. We collect the seeds from the fruits of two flowers of equal size of which one is one of the largest flowers of Ocn. biennis, the other one of the smallest of Ocn. La- marckiana. There can be no doubt what the plants raised 430 The Systematic Value of the New Species. from these seeds will be like. The result can indeed be predicted pretty accurately by means of the law of re- gression (see pp. 72>, 120 et seq.). The seeds of the biennis-^ov^^r will give plants whose flowers revert to the mean of the type of hicnnis ; the seedlings of the La- marckiana-flower will revert to the normal of that spe- cies. In other words : if we are in doubt as to the nature of individuals which stand at the boundary between re- lated species, the offspring produced after the self-fertili- zation of the individuals in question will settle the diffi- culty. Two plants which are absolutely identical in re- spect of any particular character may be proved by their progeny to be fundamentally different. And if, as often happens, two related groups only differ in a single char- acter their extreme variants may be indistinguishable. Yet their seed will prove them to be intrinsically differ- ent. The study of the limits of species is by no means solely a descriptive one. Classifications based on no more than an examination of a series of forms have no more than a transitory value. ^ Statistical methods^ will re- veal where the boundaries are : experimental methods must be called in to decide in individual cases. § 25. TRANSGRESSIVE VARIABILITY. The general conclusions arrived at in the foregoing section may now be illustrated by numerical data. The determination of the nature of the limits between related species is one of the most difficult parts of the task of the ^ De Candolle, La Phytographie, p. 80 ^ C. B. Davenport, Statistical Methods with Special Reference to Biological Variation. Transgrcssive Variability. 431 systematist. The vast majority of systematic species are made and described on a few specimens; and where large numbers of individuals have been available the in- vestigator has contented himself with the general im- pression they produce on him. The result of this is that we have a knowledge of the typical forms of species but no exact idea of their limits. Statistical investigation, as we have already said, is necessary to determine what these limits are.^ Such in- vestigation teaches us not only what the mean of a char- acter is, but also the range of its variation. In the fore- going section we have seen that the deviations are often so large that neighboring curves sometimes overlap. This is the phenomenon which I call transgrcssive variabilitw Let us choose a particular example to make this phe- nomenon clear. We will select the common species O. biennis L. and O. ninricata L. which, as every one knows, can be easily distinguished by the size of their flowers. In the former species they are large and project horizontally from the stem; in the latter small and erect. But let us subject this familiar and obvious and con- venient distinction between two species made by Lin- naeus himself^ to statistical analysis.^ We measure the ^ Beautiful examples of transgressive curves are given in a zoo- logical article of P. P. C. Hoek^ Nciiere Lacks- uiid Maifischstudicn in Tydschrift d. Nederl. Dierk. Vereeniging. (2) VI, 3, S. 231-235. See further G. Duncker, On Variation in the Rostrum in Palac- monetes vulgaris. Americ. Naturalist, Vol. 34, No. 404, 1900 and of the same author: Variation uiid Asymtnctrie bci Plcuroncctus iicsiis L., Wiss. Meeresunters. Helgoland, Bd. Ill, Heft 2, 1900. ^ Spach in his monograph of this genus also separates these two forms as species (O. vulgaris, Spach. = O. biennis L. : O. chrysantho Spach. =^ O. muricata L.). ^The types indicated in this book by the names O. bioniis and O. muricata are those found all over Europe, which are probably the prototypes on which Linnaeus based his descriptions. In Amcr- 432 The Systematic Value of the New Species. sepals, the corolla, and the calyx-tube for a certain num- ber of flowers; a very great number is not necessary. We plot the measurements as in the table on page 433, writing opposite each length the number of flowers which possess it. In the case of O. biennis I measured one flower per plant; in O. muricata I did the same (I) but in the case of a few plants several per plant (II). All the examples of both species come from the same wild locality, a sandy spot near Zandvoort (Sept. 1894). See table p. 433. The measurements^ show in the first place that the mean length of calyx and corolla for O. muricata at that locality is about 14-15 mm. and for 0. biennis about 19-20 mm. They confirm the alleged difference between the two species. But they show, further, that this differ- ence is by no means such that all of the flowers of 0. biennis must necessarily be larger than those of 0. muri- cata or that in any given case mere size would settle the question of their specific identity. On the contrary the largest flowers of O. miiricata are larger than the smallest flowers of O. biennis. The mean differences are fixed and typical. But the extreme variants overlap — the ^variability is transgres- sive. But must we conclude from this that there is no boundary between the two species, that they merge into one another? Not at all. For the flowers were picked on undoubted muricata- and biennis--^\2ints>. Or we may express the case thus : the limits of the ica numerous other elementary species of both groups are found, but they have not yet been described or named. (Note of 1908.) ^ I have given above similar data for the fruits of Oenothera leptocarpa (See page 357). Trans gressive Variability. 433 LENGTH OF CALYX LENGTH OF COROLLA. Vlillimeters Muricata Biennis Millimeters Muricata Biennis I II I II 8 0 1 — 8 0 1 10 2 1 10 1 0 11 1 3 11 3 6 12 2 6 12 8 4 13 6 13 13 12 21 14 13 24 2 14 14 34 1 15 20 35 7 15 16 29 4 16 6 11 3 16 3 5 6 17 3 7 9 17 1 6 18 4 0 8 18 9 19 9 19 9 20 8 20 12 21 4 21 5 22 6 22 3 23 1 23 1 24 1 24 6 25 1 3 Numb. 25 1 26 of flowers 57 101 63 27 1 28 1 33 1 Flowers, 5; 101 65 Species overlap but they will not disappear. And, con- versely, the homogeneity of an unbroken series of forms cannot be taken as established until it can be shown that the forms are grouped round a single center. The exist- ence of two such centers may point to the existence of two distinct types, even though they seem to merge into one another. A comparison of the length of the calyx tube in the two forms leads to the same conclusion. The plants all come from the same locality, and the flowers were all plucked from the main stems. 434 The Systematic Value of the Nezu Species. LENGTH OF THE CALYX-TUBE. Millimeters O. miiricata O. bie?inis I II 21 0 1 22 1 0 23 0 0 — 24 1 2 • 25 3 6 26 6 8 27 8 11 28 8 15 29 12 15 30 10 17 31 4 15 1 32 3 7 2 33 1 1 2 34 0 1 2 35 1 1 0 36 0 1 1 37 0 38 3 39 3 40 4 41 3 42 3 43 1 Number of flowers 58 101 25 Large numbers of such tables could be made ; and the result would be well worth the labor if they demon- strated, as they most certainly would, that the universal- ity of the law of transgressive variability is no argu- ment against the independence and immutability of spe- cies. Let us now examine a second group of characters, namely those which do not separate related species or at any rate only to a very slight extent. In this case the mean values will either coincide or show slight differ- ences caused by external conditions. For example I found Transgressive Variability. 435 the seeds of 0. biennis and 0. muricata almost exactly alike as to shape and size in spite of the considerable difference between the seeds of a single fruit. The same is true of another well-known character, the relation be- tween the length of the corolla and that of the filaments. I measured this in ten flowers of 0. muricata and in twenty of O. biennis and found the mean for the former species to be 14.6 and 8.3; for the other 10.0 and 5.5 mm., in both therefore a proportion of 100 : 55. In O. Lainarckiana however this proportion was 100 : 44. The length of the fruits is dependent to a large extent on the conditions of life. If we examine plants which have been grown under diverse conditions, we find differ- ences in the lengths of the fruits and we get series of figures which bear a superficial resemblance to the fore- going ones but are due to different causes. See the table which follows. I measured the fruits in the ripe, prac- tically dry condition, using in each plant the lowest cap- sule on the main stem; the plants were collected in wild localities, but in various places. The differences have no specific value and are manifestly due to conditions of nutrition. An opposite result might have been obtained under reversed conditions. Space does not permit me to deal in the same way wn'th the variability of the new species which have arisen from Lainarckiana. The great majority of the characters are without doubt transgressively variable ; unbroken series could be easily made with the leaves of O. snblincaris at the one end and those of O. lata at the other; or with the fruits of O. oblonga at the one end and those of O. rubrinervis at the other. But whenever a sufficient number of individuals are dealt with, curves derived from this material are found not to be monocentric but 436 The Systematic Value of the New Species. LENGTHS OF THE FRUITS OF THREE SPECIES OF OENOTHERA IN MILLI- METERS. {O. muricata 1894, the rest in 1893.) Millimeters Larnarckiana Biennis Muricata 15 1 16 1 — 17 5 18 11 1 19 17 4 20 27 9 21 Zl 13 22 62 10 23 74 23 2 24 83 24 1 25 79 28 3 26 51 30 6 27 43 36 12 28 32 32 18 29 18 27 34 30 13 21 36 31 5 22 34 32 5 26 32 33 3 28 24 34 1 7 14 35 5 5 36 6 2 37 3 2 38 1 2 40 0 1 Number 568 356 228 polycentric, each individual species forming a perfectly definite group. The character of each group is given by the center of greatest frequency, independently of the apparent absence of definite limits. Oenothera Lamarckiana Seringe. 437 § 26. OENOTHERA LAMARCKIANA SERINGE. Oenothera Lamarckiana belongs to the subgenus Oma- gra which some authors separate into a distinct genus. ^ Its most important dis- tinguishing characters he in the seeds which are irregu- larly angular with ridges along the angles and relatively smooth spaces in between (Figs. 97 and 98). These characters make it possible to distinguish them easily from the seeds of all other subdivi- sions of the genus Oenothera,'^ Fig. 97. Transverse section through the seed of Oeno- thera Lamarckiana; cc, the cotyledons (there is no en- dosperm) ; 0, epidermis; s, areolar tissue with cavities ; h, hard layer; ff, wings of the seed. which are either smooth ex- cept for small indentations, or with a sort of crown at the upper end. The epidermis of the seed of Ona- ^The most important special literature on this group is the fol- lowing : E. Spach, Monographia Onagrearum, Nouv. Ann. Mus., IV, 3, 1835. S. Watson, Revision of the Extra-tropical North American Species of Oenothera. Proceed, Am. Acad, of Arts and Sci., Vol. VIII, 1868-1873. Engler und Prantl, Die natilrl. Pflanzenfam., Ill, 7, p. 199, where references to general literature will be found. The following works must also be mentioned : J. ToRREY and Asa Gray, Flora of North America, Vol. I. 1 838- 1 840, p. 492. A. S. Hitchcock, Les Oenotheracees du Kansas, 1898. H. Leveille, Monographic du genre Oenothera (as yet only partly published). Onagra is given as a subgenus in Endlicher, Genera Plantanim, p. 1 190 sub No. 61 15; as a genus in the Natilrliche Pftanzcnfamilicn of Engler und Prantl, /. c.,p. 214; and also in Britton and Brown. An Illustrated Flora of the Northern United States, Canada and the British Possessions, Vol. II, 1897, p. 475. ^The various subgenera of Oenothera are often accidentally mixed up in botanical gardens but the above mentioned characters of the seeds usually make it possible to sort them out again before sowing. 438 The Systematic Value of the New Species, gra which is smooth at first, grows much faster than the parts inside so that it acquires wrinkles and folds by the pressure of the surrounding seeds (Fig. 98). As a result of this there is great diversity in the form of the seeds of the same loculus. In this respect the seeds of the various species of Onagra are practically all alike. The fruit is an erect capsule which splits longitudi- nally and contains many seeds. The flowers have a long calyx-tube, are tetram- erous and apparently regular but exhibit as a matter of fact, a slight degree of zygo- morphy which is most pronounced in the fila- j ments. Prophylls are absent. The species which bore these characters were described and named by Linnaeus. They were O. biennis, L., O. miiricata L., and O. parvi flora L. To these were added later the well-known forms 0. siiaveolcns Dcsf. (=0. grandiflora Ait.) and 0. La- marckiana Ser. (^0. grandiflora Lamarck).^ Besides these there belong to the group in question a whole series of American forms which are little known in Europe. ^ Although the name 0. grandiHora has -good claim to priorit}' for both these species I do not propose to use it because it has already given rise to a great deal of confusion. See Nederl. Kriiidk. Ar chief. Aug. 1895. Fig. 98. Seeds. Lm, of O.Lamar chana, seen from the back; Lm', seen from the other side which has a sharp ridge on it; g, O. gigas; r, O. riibri- ncrvis; n, O. nanella; It, O. lata; a, O. alhida; s, O. scintillans, opened and empty ; h, the hard layer which surrounds the inner lumen. Oenothera Lamarckiana Seringe. 439 Later systematists have either lumped all these forms into one big species or have separated them up in differ- ent ways. When they did the former, they took O. biennis as a type. The other species were ranged round this as varieties. This was done for example by Torrey and Gray in their famous Flora of North America and by Watson in his monograph. This is an important point for our further discussions, as showing that a minute comparative study of the various forms points to their common origin from O. biennis. Spach however thinks differently. He separates six species of Onagra. Two of them include all the forms which interest us; the rest are rare and do not exist in Europe either as wild or as cultivated forms. These two species are (1) Onagra vulgaris Spach = Oenothera biennis L; but also including O. suaveolens Desf. and O. Lamarckiana Ser. and (2) Onagra chrysantha Spach, which is composed of O. muricata L., O. parz'iflora L., O. cruciata Nutt. and of a Var. latifolia with which I am not familiar. I think it is legitimate to conclude from this that the original Oenothera biennis has given rise to the remain- ing species mairfly along three lines: 1st by an increase in the size of the flowers {Lamarckiana), 2d by a de- crease in this size {chrysantha) and 3d without any change in it. The other features of the flowers are closely correlated with its size and, in fact, appear in great measure to be determined by it. The species of Onagra differ from one another not only in the structure of the flowers but in that of the leaves as well. Furthermore the fruits of 0. parz'iflora split by 8 apical valves instead of 4, whilst O. Lamarck- iana has an entirely different appearance. Further, less 440 The Systematic Value of the New Species. important, distinctions are afforded by the degree of hoariness and so forth. The following questions suggest themselves : Has Oenothera biennis been through a period of mutation similar to that through which O. Laniarckiana is going now? If it has, did it give rise to species which now exist in the same way as 0. Laniarckiana is doing now? Have the existing forms arisen directly from it or do they owe their origin to repeated mutations? Finally does there exist anywhere at the present moment a mu- table family of O. biennis which is perhaps giving off some of the forms which we know and others, perhaps, as well ? These and other questions must for the present be set aside for future investigation to answer. At any rate they serve to illustrate the theme of the present chapter, which is that the Onagra-group is pre- cisely parallel to the group of mutations which is being produced by 0. Laniarckiana. It is older and perhaps more extensive. But if the distinctions between species within the two groups can be shown to be of the same systematic value, the parallel between my new species and the older species of recognized position will strongly be supported by pure systematic evidence. Now that I have foreshadowed' the contents of the paragraphs that follow, let us go back to O. Laniarckiana.^ ^ Oenothera grandiflora Ait. = 0. suaveolens Desf. is often con- fused with Laniarckiana either under one of these names or as O. macrantha Hort. The facts are as follows It was first described by WiLLDENow in his Species Plantariim (Vol. II, 1799, p. 306) but seems to have been figured before that time by L'Heritier, Stirpes novae, Tom. II, Plate 4. De Candolle in his Prodromiis suggests that O. grandiUora Ait. and O. suaveolens Desf. may perhaps be dif- ferent species. Desfontaines who gives no description of it in his Tableau (ist edition 1804, p. 169; 2d edition 1815, p. 195) seems to regard the two names as synonyms. In the general Herbarium of the Museum of Natural History at Paris I found in the drawer for O. biennis a sheet of paper on which Oenothera Lamarckiana Seringe. 441 First I shall give a translation of Lamarck's own de- scription of the species. In order to understand this we must bear in mind that Lamarck was not comparing it with the forrris most closely related to it but with another form with very large flowers, O. longiflora, belonging to another subgenus. Moreover he had neither seen speci- mens from America, nor living plants either wnld or culti- vated. His description rests on dried specimens in the Paris Herbarium, which had been grown in the Jardin du Museum d'histoire naturelle. Lamarck^s words are : Leaves entire, oval-lanceo- late, petals not indentate, fruits glabrous. This species bears a general resemblance to Oenothera longiflora but can be distinguished from it by a number of obvious characters, in particular by its branched stem, its entire leaves and short and smooth fruits. Its stem is three to four feet high, cylindrical, almost glabrous, of a red- dish brown color with numerous projecting branches. The leaves are green, spirally arranged, oval-lanceolate, glabrous on both sides and entire; the lower leaves are petiolate and slightly toothed below. The bracts are nar- two stems of Oenothera grandiHora Ait. had been stuck. One of them bore this name in the handwriting of Michaux. At the side of the other Desfontaines had written Oenothera snaveolens Hort. Paris. Somebody else had written above it Oenothera grandiflora and Poiret EncycL, and there is written underneath it in the handwriting of Spach : Onagra vulgaris grandiHora Spach., which name, in Spach's Monograph (p. 353), is synonymous with O. grandiflora Lam. Spach therefore, we see, did not distinguish between these two grandiHoras although they are absokitely unHke one another. These two specimens are identical with the form frequently cul- tivated in gardens under the name of O. grandiflora Ait. z=: O. sua- veolens Desf. I have also often got it under the names of O. niae- rantha Hort. and O. odorata Hort. (the latter name is erroneous and is due to the French name Enothere odorante). My investigations in the Herbarium at Paris have convinced me of the" identity of the form I cultivate as O. suaz-eolens Desf. (O. macrantha Hort.) with the form described by Desfontaines. Both of them have flowers of the same size as those of O. biennis. 442 The Systematic Value of the New Species. rower, more pointed and sessile. The flowers form a broad terminal cluster; they arise singly from the axils of the bracts but are crowded close together. The calyx is yellow, the tube somewhat longer than the four lan- ceolate broad-based sepals, which are terminated by a short, fat, thread-like prolongation. The four petals are oval, very large, and rounded, almost as long as the calyx-tube and tapering down to a narrow base. The fruit is a short capsule ; it is cylindrical, glabrous, and truncate, square in section and is about one-third of the length of the calyx-tube.^ The original specimens described by Lamarck are still in the Herbarium of the Museum of Natural History at Paris and are marked there with the same number as they are in the Dictionnaire. I have carefully compared these specimens with the plants which I have cultivated in my experimental garden and have convinced myself of the identity of the two.^ The original specimens, however, by no means represent the mean type of the species in every respect and therefore the description does not exactly correspond to this type, particularly as regards the corolla and the fruits. The petals are ob- cordate but only slightly emarginate as compared with O. longiflora; the fruits are of the same form and size "^ Encyclopedic mcthodique, Botanique par Lamarck, Tome IV. Paris, An. IV (1796), pp. 550-554. Usually cited as Lam. Diet. ^ It appears that it was not Lamarck but Poiret who wrote the section on Oenothera in the Dictionnaire. The specimens in the Her- barium bear the note O. grandiflora written by Poiret. In the same Herbarium there is in the case for O. biennis, a specimen of Oeno- thera grandiUora Lam. from the collection of Father Pourret : both plants were given to the Museum in the year 1847 by Dr. Barbier. This plant was probably picked by Pourret in the garden of the Museum at the time of his visit to Paris in 1788. Later, Spach made the following note on this specimen : Onagra vulgaris grandi- flora Spach, which proves the identity of this name with O. La- marckiana. This plant also agrees exactly with the form I use in my cultures. Oenothera Lamarckiana Seringe. 443 as those of 0. biennis and agree moreover with these in the amonnt of hoariness.^ The subgenus Onagra, to which O. Lamarckiana be- longs, comprises North-American forms almost exclu- sively. The various forms growing wild in Europe have been imported from there. Oenothera biennis from Vir- ginia about 1614, Oenothera miiricata from Canada in 1789 by John Hunnemann, Oenothera siiaveolcns in 1778 by John Fothergill.^ The first two grow abun- dantly in the Netherlands, on the sand dunes which stretch along the coast, where each consists, so far as I am aware, of a single subspecies. They are widely dis- tributed throughout Europe. O. suave olens grows wild at the present time in many localities in the w^estern parts of France.^ The native country of O. Lamarckiana is unknown, but is probably Texas. ^ It only occurs wild with us when it escapes from gardens. One of the characters of 0. Lamarckiana is the sym- metrical floral structure, which is best seen in the sta- mens.^ The flowers project sideways from the stem, often almost horizontally. The stamens are inclined downwards at their base; the upper ones more than the low^er ones; the upper halves being more or less erect. *I refer the reader who is interested in a further discussion of the synonymy and wants a further account of the characters of these species to Sur V Introduction de I'Oenothera Lamarckiana dans Ics Pays-Bas, in Nederlandsch Kruidkundig Archief, Aug. 1895. ^W. T. AiTON, Hortus Kcwensis, 2d edition, Vol. II, 1810, p. 34i- ^GiLLOT, Sac. Bot. France, 1893, p. 197. See also Tome III, p. 437- *The strain which is now being cultivated in European gardens was introduced from Texas about i860. See Bcr. d. dcutsch. Bot. Gesellschaft, 1905, Bd. XXIII, p. 382. (Note of 1908). ^H VocHTiNG, Ueher Zygomorphie und deren Ursachcn, in Pringsh. Jahrb. f. wiss. Bot, Vol. XVII, 1886, p. 311. See Plate XVI, Fig. 14, in particular. For an account of geotropical curvations of Ocnothera-^owtrs, see also Hanstein, Beitrdgc cur allg. Morpho- logic, IV, 3, p. 151- 444 The Systematic Value of the New Species. VoCHTiNG found that this symmetrical bending was due to gravity. He fixed branches to a chnostat and observed that the flowers opened normally but that the filaments remained straight. Neither light nor darkness had any influence on these processes. The bending of the fila- ments takes place just before, or during the unfolding of the flower. If, about the middle of the day on the evening of which the flower will open, we open a bud we find the filaments perfectly straight.^ It follows from this that the degree of bending in the filament depends on the angle which the open flower makes with the perpendicular. The smaller the angle the less the bend. In all these respects our species behaves like O. bien- nis. On the other hand in O. miiricata and O. parviflora the filaments are not bent.^ The absence of bending in this case, however, directly depends on the fact that the flowers in these species instead of projecting sideways stand up erect. As a matter of fact, this bending is not entirely absent; I always found some signs of it even if they were only very slight ones. § 27. SYNOPSIS OF THE CHARACTERS OF THE NEW SPECIES. My new species, without exception, possess the gen- eral characters of the biennis-gvon^ to which 0. La- in arckiana belongs. According to Watson^s Monograph of the genus, the following are the characters of this group. •^' ^ See the Figure on page 218. ^ Bull. Soc. Bot France, T. Ill, p. 437. ^ Sereno Watson, Revision of the Extra-tropical North Amer- tcan Species of the Genus Oenothera, Proc. Amer. Acad, of Arts and Sciences, May, 13, 1873, Vol. VIII, pp. 573-618. Characters of the New Species. 445 Plants annual or biennial, forming an erect, for the most part branched stem. Flowers yellow, buds erect, surmounted by the four tips of the calyx. Anthers lin- ear, each inserted at its center upon filaments of equal length. Stigma formed of four or more long cylindrical parts, which are either free or more or less fused later- ally. Calyx tube narrow, slightly broader at the top. Fruits sessile, oblong, tapering upwards; seeds in two rows in each loculus ; the integument of the seeds too big for the kernel and therefore wrinkled. This list, taken in conjunction with the following tables, will suffice to show that the new species belong to the group in question both morphologically and sys- tematically. That they are more closely related to O. Lamarckiana than to 0. biennis, O, niuricata, 0. sua- veolens or the other species of this group described in systematic works, is shown, apart from their origin, by certain characters of the flowers. In the first place these are much larger than they are in the other forms ; and, in the second, they have a longer style. The style raises the stigma in the bud above the tips of the anthers. When the flower opens the four stigmas expand into the form of a cross, not touching the anthers, however, as a rule. In 0. biennis on the other hand the stigmas lie, in the bud, between the anthers and do not reach above them at the time of flowering. This state of affairs is very important from the point of view of fertilization. In O. biennis this takes place in the bud because the anthers dehisce a whole day before the flower opens; though the exact time of this is of course subject to some variation. This fact obviously makes the operation of castration preliminary to crossing much more difficult, because it has to be done on very 446 The Systematic Value of the New Species. young buds. On the other hand it faciHtates self-fertili- zation by rendering it superfluous to do more than exclude the visits of insects. A very different state of things obtains in Oen. Lamarckiana. Here castration can be easily and safely effected even in large buds ; but in self- fertilization the pollen has to be actually transferred. In this respect all the new species except 0. lata and O. Fig- 99- Ripe fruits shortly before drying, half natural size. L, Oenothera Lamarckiana; R, O. rubrinervis ; A, O. albida. hrevistylis and the sterile forms behave exactly like 0. Lamarckiana and not like O. biennis. The necessity of fertilizing, year after year, with my own hand every single flower from which I wanted to save seed has given me sufficient experience on this point. In the description of the species (§§ 10-23) I made occasional reference to atavistic phenomena. For ex- Characters of the New Species. 447 ample 0. nanella in its earliest stages forms a few long- stalked leaves; and occasional crumples are seen in the otherwise smooth leaves of O. laevifolia and 0. scintil- lans, etc. In this they behave like many species even in other families which reproduce in their early stages the characters of their ancestors (for example Acacia, Ulcx, Stum, etc.). I shall now attempt to set forth the characters of the Fig. 100. Ripe fruits shortly be- Fig. loi. Ripe fruits, shortly be- fore drying; half nat. size; the fore drying, half natural size, bracts have not yet fallen off. o, G, Oenothera gigas; Lt, O. lata. O. oblonga; s, O. scintillans. new species in synoptic tables in order to make it easier to compare them with the characters of the older species in the section which follows. And in order to express myself as simply as possible I shall regard the character of the parent-species O. Lamarckiana as the normal and compare the others with it. Further I propose to deal with the different organs 448 The Systematic Value of the New Species. and developmental stages separately in the tables; and I shall start with the seedlings at the age (2-3 months) at which they are usually sorted out and recorded. The first table therefore gives the characters which are used in this sorting. ANALYTICAL TABLE OF SEEDLINGS. I. Leaves stalked. A. Leaves of the same breadth or broader.^ 1. Of the same breadth and shape, not to be distinguished as seedlings. a) (Fig. 48, 51, 52, 64, 65, 66, 72, 95) 1. O. Lamarckiana. b) 2. O. bj-cvistylis. c) 3. (9. leptocarpa, 2. Broader, pointed, with many crum- ples. (Fig. 52, 63, 65, 66) \.O.gigas. 3. Broader, rounded at the tip with very- deep crumples, edge incurved. a) (Figs. 48, 51, 52, 91, 9^, 95) . 5. O. lata. b) 6. 6^. semilata. B= Leaves narrower. 1. Broadest in the middle. a) very long with long stalks, with narrow veins, almost smooth (Fig. 83) 1 . O. elliptica. b) small with broad leaf-stalk and broad principal veins, very smooth, shiny, dark green (Figs. 51,81,82) S. 0.sci7ihllans. 2. Of equal breadth over the greater part of their length. a) green. a) 1. Only slightly narrower, smooth without, or al- most without, crumples . 9. O. laevifolia. a) 2. Very narrow with broad leaf -stalks and broad veins which often are reddish; wrinkled (Figs. 48, 53, 72, 73, 74, 95) . . 10. O. oblonga. *"(than in Lamarckiana)" as also in the other analytical tables. Characters of the New Species. 449 b) whitish, b) 1. Crumples many, pointed, narrowing ofif into the stalk (Figs. 48, 72, 75, 76, 95) 11. O.albida. b) 2. Crumples few, narrowing off into the stalk, wavy, brittle, veins reddish — (Figs. 52, 68) .... 12. O. rubrinervis. b) 3. Crumples few, scarcely narrowing off into the stalk, almost grasslike . 13. O. sublinearis. II. Leaves sessile, short and broad, almost heartshaped, crumpled (Figs. 51, 52, 78, 79) 14. O. nanelLa. The new species can best be distinguished from one another by their so-called habit. This is, as in the case of O. Lamarckiana itself, largely dependent on external conditions. In the first place biennial plants are as a rule naturally stronger than annual ones. The former are sometimes more than two meters high; the latter often little more than a meter. In both cases the time of sow- ing makes a difference; the earlier the plants come up the more time they have for their full development. The height and amount of branching of the plants are largely dependent upon the amount of sunshine they get, and on whether they are growing close together or not. The result of this is that spurious differences, which are either indirectlv connected, or not connected at all, with real specific characters, may appear in comparing cul- tures of related species, and obscure the real differences. On the other hand genuine differences sometimes tend to become obliterated. But if uniformity of treatment is insured, beds of my new species have a perfectly distinct and different aspect and can be recognized with certainty, even at a distance. 450 The Systematic Value of the New Species. The following table mainly refers to annual plants in flower. ANALYTICAL TABLE OF FLOWERING PLANTS: HEIGHT AND MODE OF BRANCHING I. Of the same or nearly the same height (1 5 — 1.8 m). A. Flowering over in October. Stem erect, rigid 1. Of the same strength. a) Secondary stems strong, branches short, foliage lax (Fig. 55) ... 1. O. Lamar ckiana. b) Secondary stems weak, main stem branched; infrutescence lax; stem reddish, brittle, often wavy (Fig. 49, 67, 69, 70) 2. O . rubrinervis . 2. A little weaker. a) Leaves narrow, very much like O.Z«w. (Fig. 56) 3. O.laevifolia. b) Leaves broad, like O. lata but taller 4. O.semilata. 3. Very strong, stem stout and very erect, dense foliage, short internodes, branches short and rosette-like. In- florescence closer and fuller ... 5. O.gigas. B. Flowering continues till winter. Weak and drooping at that time. 1. Much branched; flowers many; group of buds above the flowers small . . 6. O.btevistylis. 2, Slightly branched; flowers rare; group of buds above the flowers very long 7. O.leptocarpa. IL Shorter (about a meter or less). A. Much branched. 1. Branches pressed close to stem; the whole plant rigid. Bud-bearing zone above the flowers long 8. C?. scintillans. 2. Branches projecting outwards, rigid. a) Main stem thick, projecting above the branches 9. O.albida. b) Short, weak 10. (9. elliptica. c) Usually very weak W. O. sublinearis. Characters of the New Species. 451 3. Branches weak, and so bent down- wards, top of plant also weak . . .12. O. lata. B. Almost unbranched, branches in the form of rosettes, stem very thin (Fig. 50, 71). . . U. O.oblonga. III. Dwarf, often flowering when only 10-20 cm. high (Fig. 45, 77) 14. O.nanella. In Oenothera Lamarckiana every part of the plant has a characteristic type of leaf from the seedling to the top of the inflorescence. The same is true of the new species which have arisen from it. The radical leaves of the full grown rosettes merge by imperceptible de- grees into the lower leaves of the stalk. As we ascend the stem, the leaves become gradually shorter and set on smaller stalks until we reach the inflorescence, at the bottom of which, or slightly later, they become almost sessile. In the young inflorescence they extend beyond the flowers, but, later on, become relatively small com- pared with them. The greatest breadth of the leaf, which at the bottom of the plant is about its middle, gradually shifts, as we ascend, to its base. In describing the leaves of the different new species we must therefore compare only such as are borne on the same part of the stem. ANALYTICAL TABLE OP THE LEAVES. I. Of normal breadth. A. Of normal length and form. 1. Pointed. a) (Figs. 62, 89) \. O. Lamarckiajia. b) 2. O. leptocarpa. 2. Rounded 3. O. brevisfylis. B. Roundish 4. (7. semilata. C. Short, sessile or with a short stalk; broad at the base; often auriculate or heart shaped 5. O. nanella. II. Broader. A. Of the same form, but very variable, teeth large, numerous, especially at the 452 The Systematic Value of the New Species. base. Those on the stem bent down- wards. (Fig. 54, 62) 6. O.gigas. B. Round, stumpy, slightly toothed, but usually with an incurved edge (Fig. 57, 58, 88, 89) 1. O. lata. III. A little narrower. A. Green. 1. Smooth, without crumples. a) Of normal length, flat 8. (9. laevi folia. b) Small, median vein broad, whitish (Fig. 54) . . . 9. 6>. sci7itillans. 2. Uneven; radical leaves narrow with a broad vein; leaves on the stem sessile and with a broad base (Fig. 54) . . 10. O. oblonga. B. Whitish. 1. Often with red veins; broadest in the middle, bracts folded longitudinally (Fig. 54) 11. (9. riibrinervis . 2. Sessile with a narrow base; only the lower leaves stalked (Figs. 54, 57) . 12. O. albida. IV. Very narrow. A. Lanceolate, long, often ten times as long as broad (Fig. 83) 13. 6>. elliptica. B. Almost linear, small (Fig. 85, 86) . . 14. O . subh7iearis . To turn now to the flowers ; I have already stated above that their size depends largely on the strength of the plant which bears them. They exhibit both indi- vidual and partial variability and follow Quetelet's law in these respects. A very striking fact is that their size gradually diminishes during the flowering period (which lasts from July till October) and that at the end of it they sometimes sink to % or even half their orig- inal size. This is obviously determined by the exhaustion of the plant by fructification ; for O. hrevistylis, which sets practically no seed and often goes on flowering until well into November, bears large and bright flowers even at that time. The flowers are smaller on the lat- eral branches if the main stem is laden with fruits. But Characters of the New Species. 453 if part, or all of this, has been cut off during early life (as is often done for the purpose of artificial fertiliza- tion) the lateral branches bear remarkably large and fine flowers. It follows from this that those new species which are of a delicate nature will have somewhat smaller flowers. ANALYTICAL TABLE OF FLOWERS, FRUITS AND SEEDS. (Figs. 98-101.) I. Flowers as large or larger, petals, on the average 3-4 cm. long (plants large). A. Fruits and seeds normal; buds thin; ta- pering to the top (Fig. 99). 1. Calyx and fruits green, sometimes slightly reddish (Fig. 61) . . . . 1 O. Laniarckiana 2. Calyx reddish, fruits striped with red, petals often more or less crumpled, broad, becoming darker as they fade 2. O rubrinervis. 3. Pale yellow; the later flowers with oval petals (Figs. 59, 60) . . . . 3. O laevifolia. B. Fruits short and thick (Fig. 101). 1. Seeds dark brown, large and plenti- ful; petals very broad; buds thick . 4 Ogigas. 2. Seeds large, scanty; buds fat; petals crumpled; anthers sterile (Fig. 46) . 5. O lata. 3. Almost the same, pollen fertile . . 6. 6> semilata. C. Fruits short and thin, flowers short- styled, ovary partly superior . . . 1 . O. brevistylis. D. Fruits long and thin. Flowering does not begin until late in the summer and lasts well into the autumn . . 8. O. leptocarpa. IL Flowers smaller, or very nearly as large; petals about 3 cm. long (plants short). A. Fruits long and thin; flowers much ex- panded; petals elliptical. a) Fig. 84 ^ O. elHpUca. b) Fig. 87 10. (7. sublinearis B. Fruits of almost normal size. a) Seeds plentiful, of almost nor- mal size. Buds often laterally twisted \\. O. nayiella. 454 The Systematic Value of the New Species. b) Fruits thinner, poor in seed; flowers pale yellow, corolla less expanded. (Fig. 88, 89) 12. O. albida. C. Fruits short and thick, of half the normal size or less. a) Flowers erect; seeds small; fruits smooth 13. (9. scintillans b) Flowers projecting sideways; fruits not so stout, poor in seed . . . 14 O. oblonga. The characters given in these tables are those which I have myself ordinarily employed in sorting and record- ing my plants. But there are also small differences which practice enables one to recognize with ease, and to emplo}^ with certainty. It is, however, almost impossible to express them in words. And the above mentioned circumstance that the degree of development of all the organs is highly correlated with the individual strength of the plant always makes descriptions appear incom- plete ; but, on the other hand, materially facilitates the discriminations of the living material. § 28. COMPARISON OF THE CHARACTERS OF THE OLD AND NEW SPECIES. The new species which have arisen in my experi- mental garden from Oenothera Laiuarckiana differ from one another in the same way as do the already known species of the bieiinis-group. I shall now endeavor to prove this important generalization by a detailed com- parison of the two groups. Unfortunately the difficulty of giving this proof is considerably enhanced by the in- completeness of the descriptions which have been given in the literature. The diagnoses are usually short, often based on single herbarium specimens about which we have no means of knowing in what characters they repre- Comparison of the Old and New Species. 455 sent the mean of the type and in what they deviate from it, and if so, to what extent. There is practically no in- formation about the seedlings ; and this would have been particularly valuable in this case. And so forth. These gaps in the literature can of course best be filled up by growing the species in question : and for many years I have cultivated the forms which grow wild with us and some other ones, on a large scale and under different conditions. In 1895 I procured in ex- change from the botanical gardens all the available sam- ples of seed of the subgenus Onagra and sowed as many of these as I could manage. And since then I have taken every opportunity that offered, of procuring Onagra- seeds. I am, of course, most familiar with the forms which grow wild with us, 0. miiricata and O. biennis; but I only possess one form of each of these. ^ I am familiar with 0. suavcolens which is widely distributed over France and have two subspecies of it; with 0. hirsutis- sinia (O. biennis hirsutissima Torrey and Gray) ; with 0. parz'iflora L. and 0. cntciata A'utt. ; and with some others. I am only acquainted with figures or herbarium specimens of O. spectabilis Spach (0. corynibosa), O. elata Kunth, 0. media Link, 0. erosa Lchni., etc. But they are intermediate in character, so far as it is pos- sible to judge, between the two species mentioned first ; in fact they bridge over the gap between these two to a large extent. For these reasons I shall confine myself almost en- tirely to the comparison of the new species with 0. biennis, O. miiricata, 0. Lamarckiana, O. crnciafa and * Probably the types, used by Linnaeus for bis descriptions. Compare Note on page 431. (1908.) 4S6 The Systematic Value of the New Species. O. suavcolcns. This will be sufficient to show that the differences between the former are greater than those between the latter. A study of the other old species would obviously only serve to bear out this conclusion. Let us begin with the seedlings. They fall into two groups. O. biennis and O. Lamarckiana have broad leaves (Fig. 102 A), O. muricata, 0. cruciata and 0. suave olens narrow ones (Fig. 102 B). These differences can be seen particularly well in very young rosettes ; but when the leaves begin to grow quickly as they do in June they all become longer and their distinguishing feature, therefore, less striking (Fig. 103), Fig. 102. Seedlings. A, of Oenothera biennis L. ; B, of O. muricata L., two months old. only however to become quite clear again later on. I have often grown rosettes of various new and old spe- cies in rows, close to one another, in order to compare 10-20 or more individuals of the same age and under the same conditions. O muricata and 0. scintillans differ most widely from the normal in the narrowness of their leaves ; in both of them the leaves are smooth and shiny ; in the former however they are pale green and long, in the latter dark green and short. O. ruhrinervis, O. suaveolens and O. hirsutissima have wavy crumples and Comparison of the Old and New Species. 457 pale leaves. They look very much alike when young, but the first of them can be distinguished earlier and with greater certainty from its neighbors (in hybrid crops for example). Rosettes of 0. gigas are much larger and stronger than those of O. Lamarckiana ; these are about as vigorous as O. biennis, but their leaves are not smooth like those of O. hiennis but uneven. O. elliptica is often scarcely distinguishable from O. cruciata; 0. suhlinearis Fig. 103. Full grown leaves of young rosettes in June at the age of 3 months. B, O. biennis; M, O. muricata; S, O. suaveolcns. has the narrowest leaves of the whole group. Between this latter and 0. gigas the various old and new species form a perfect series of transitional forms. Although single individuals or their figures convey only a very imperfect impression of a species, I invite the reader to compare Figs. 102 and 103 with those, which have already been given, of rosettes and leaves. 458 The Systematic Value of the Nezv Species. First with the groups of leaves from rosettes in June (Fig. 52, p. 293 and Fig. 53, p. 294) ; and then with the rosettes of O. gigas (Fig. 63, p. 324), O. lata (Fig. 92, p. 412); O. scintillans (Fig. 82, p. 383), O. oblonga (Fig. 74, p. 344) and so forth. The radical leaves of the full grown rosettes and the leaves on the stem behave in the same way. Those of O. biennis and 0. Lamarckiana hardlv differ at all in Fig. 104. Radical leaves of full grown rosettes. B, of OenotJiera biennis; L, of O. Lamarckiana. The spots on the leaves are brown in life. form (Fig. 104 B and L). Those of the former are smooth with few crumples, with red main nerves, and often a number of scattered brown spots; the latter are very much wrinkled, without red pigment or at most with no more than isolated red spots. In form. 0. gigas differs somewhat more (Fig. 62 on page 323) and 0. lata more still (Fig. 58, p. 311, and Fig. 89, p. 405). Comparison of the Old and New Species. 459 At Fig. 105, p. 460 will be seen a group of stem leaves for comparison with the corresponding ones in Fig. 54 on page 295. The differences are obviously of the same order. In the case of O. cruciata and 0. nniri- cata (Fig. 105, p and in) they are most pronounced; and still more so in O. elliptica and O. sublinearis, which are not included in Fig. 54. With regard to ''habit" the majority of the older species do not differ much from one another. 0. nniri- cata has usually stronger lateral branches than 0. biennis; O. Lamar ckiana has a longer spike than either. 0. cru- ciata is shorter than O. biennis, which, however, 0. siiaveolens and O. hirsutissima very much resemble, though they are less robust. All these comparisons are of course made between plants under similar conditions of cultivation. Under such conditions O. rubrinervis, 0. gig as, O. laevi folia and 0. brevistylis do not differ so much from Lamar ckiana as do the shorter forms which have an entirely different habit. Amongst these O. lata is broad, close and compact whilst O. oblonga and 0. scintillans with their narrow leaves have a rigid and thin stem which branches only slightly or not at all. The glaucous color of O. mtiricata is characteristic of this species; the green of 0. albida is paler than, and that of 0. rubrinerms about the same as that of O. siia- veolens and O. hirsutissima. These four forms are very much alike, apart from their flowers and fruits. With regard to the flowers the differences are much greater between the older species than they are between the new ones. The flowers are small in O. miiricata, 0. parviflora and 0. cruciata: medium in O. bienjiis, O. snaveolens and O. hirsutissima, and very large in O. Lamarckiana. In the first group they are erect, and their 460 The Systematic Value of the New Species. stamens therefore not bent (see. p. 444) ; in the two latter groups they project outwards and the androecium is mod- ified correspondingly. In Lamarckiana the stigma ex- Fig. 105. Stem-leaves of Oenothera biennis (b) ; O. sua- veolens (s) ; O. hirsutissima (h) ; O. cniciata (p) ; O. muricata (m) ; to be compared with Fig 54 on p. 295. tends beyond the anthers ; it reaches the same level in the other forms. Comparison of the Old and New Species. 461 In all these and other details the new species have the flowers of Lamarckiana. But during the last two years my mutants have overstepped even this limit; one hav- ing appeared with hicnnis-^ow^vs and one with ;//z/n'ca/a- flowers, not however in pure but in crossed strains. A curious form must be mentioned here, Oenothera cruciata Nuttall which is described by some systematists as a species, but regarded by others as a variety of O. parviflora, from which it differs by its small linear petals but in no other respect. It is therefore more closely allied to O. parviflora than any two of my new species are to one another. Lastly let us look at the fruits. The older species resemble each other with the exception of O. parviflora, in which the capsule is described as opening by eight valves instead of four. The other alleged differences such as cylindrical or conical form, greater or less degree of hoariness, length and thickness, etc. are subject to a very great extent to individual fluctuations and do not seem to constitute differences of specific rank. On the other hand it is just in the characters of the fruits and seeds that the new species differ most amongst one another, as the table on page 453 and Figures 98 (group of seeds, p. 438) and 99-101 (fruits, pp. 446- 447) clearly show. We can sum up by saying that the known systematic species of the subgenus Onagra differ from one another in essentially the same way as do the forms which have arisen from 0. Lamarckiana. The two groups are pre- cisely analogous. The relation between the group of Owflf^ra-species and O. biennis is the same as that between the group of Lamarckiana-mwtdints and OenotJiera La- marckiana itself. IV. ON THE LATENT CAPACITY FOR MUTA- TION. § 29. REPEATED MUTATIONS ARE THE RESULT OF THE SAME INNER CAUSES. Hitherto I have confined myself to a mere descrip- tion of the phenomena of mutation in the genus Oeno- thera such as I have directly observed them. . Our task now is to form some idea of the causes of these phe- nomena. The attempt to deal with this problem is not only a perfectly legitimate one but the reader would be justified in complaining that my work were incomplete if I did not attempt to deal with it. The solution of this problem must, however, be sought among the facts themselves. And for this purpose I shall divide my argument into two parts. The causes which can be dealt with most easily are, naturally, those which operated throughout my experiment, that is the internal and external causes of each of the several muta- tions. But to provide an answer to the questions : what is the cause of the whole phenomenon, and to what is the initiation of the mutation period due Pisa task of a very different nature. And I propose to postpone this to the last section of this chapter. Repeated Mutations Due to the Same Causes. 463 The facts, summarized in this and tlie previous sec- tion, of the repeated reappearance of the mutations ob- served in my cultures evidently admit of one explanation only, viz., that the potentiality for each mutation is pres- ent in a latent condition m the apparently normal indi- viduals in my cultures. Let us take as an example the Lamarckiana-id.m\\y (p. 224), of v^hich I have grov^n a great number of suc- cessive generations. The first sow^ing gave two mutations {lata and nanella) ; the following generation gave them again, and one other besides. The seeds for this second sowing were gathered from 6 seed-parents which had flowered far away from other Oenotheras and therefore can only have been fertilized by one another. They were obviously chosen without any indication whether they would be more likely to produce mutations than the re- maining individuals of the first generation. That these six seed-parents reproduced the same mutations as their parents proves that there existed in them some heritable character in a latent condition. The same is true of subsequent generations and of the other families in my cultures. Each time the same mutations arose from apparently normal individuals. The capacity for giving rise to these must therefore have been inherited in the latent condition. If a latent capacity of this kind is not assumed the following three facts become absolutely inexplicable. First, the circumstance that the same mutation ap- pears in the same crop in two or more of several indi- viduals, whether the crop arises from the seeds of one or several seed parents. Secondly, the oft cited fact (Part II, p. 272 etc.) that the appearance of mutations seems to depend almost 464 On the Latent Capacity for Mutation. exclusively on the extensiveness of the crop. Whenever I, had the opportunity of sowing on a large scale, either with seeds from the field at Hilversum (1889) or, in my own families, with the seeds of a few seed parents, espe- cially in the year 1895 (p. 224 and p. 262), a large num- ber of mutations appeared. Their rarity in other years and cultures can therefore only be attributed to the small scale on which the sowings were made; for on a few square meters we cannot expect to get many mutations if the seeds are not sown very close and the crops are not examined every day. Thirdly, the small number of the different mutations which appeared. By no means does every conceivable deviation occur. Thus there arose no white flowers, no glabrous or unbranched individuals, no linear petals,^ no trace of petalomany or apetaly and so forth. Even of the two new species which w^ere found in the field at Hilversum, 0. hrevistylis and 0. lacvifolia, not a single example occurred in my cultures. We are led to the same conclusion by a consideration of the more or less incompletely developed individuals^ of the new species which sometimes seem to constitute transitional forms. For these arise in my cultures not before the mutants, but simultaneously or more com- monly only after them. Each mutation is as completely developed when it first appears as afterwards. When a mutation is grown through many generations and on a very large scale its various representatives conform to exactly the same type. I possess photographs and de- scriptions of my mutations from the first year of their appearance and find that nothing has been added to, or ^ "Forma cruciata" as found in Oenothera cruciafa Nuff. and some others. Repeated Mutations Due to the Same Causes. 465 taken away from, their type. I have often hacl /a/a-plants from two or three sources, e. g., the 1st, 2d and 5th gen- erations, growing side by side in my garden; they were quite indistinguishable from one another. Intermediate forms seem to be associated more with some mutations than with others. Rarest with O. nanella, they are commonest with 0. laevifolia. Sometimes the intermediate forms repeat the new type of their species more or less completely in the lateral branches which arise from the axils of the rosette leaves (as for example an O. laevifolia which exhibited excessive crumpling on the leaves borne on the main stem). In this case they may be regarded as individuals in which the typical char- acter of the species is more or less latent at first. Thus these apparent transitional forms are not the steps by means of which the new species has attained its full development. They are rather the imperfect copies of a perfect picture which already exists. They are, in a word, the extreme variants of the perfectly constant new type (see §§24 and 25). It is in this very respect that the newly formed spe- cies behave in a diametrically opposite way to the races built up by the accumulation of fluctuating characters (Part I, § 7, p. 71) ; and it is this fact which justifies their title to specific rank. The general conclusion of this argument is : At the beginning of my observations, in the year i886, the characters of the nezv species, zvhich appeared later in my cultures, were already present in the plants in the field at Hilversum in a latent condition. They remained in that condition for many generations, both there and in my cultures, and only appeared from time to time, espe- cially in large sowings. 466 On the Latent Capacity for Mutation. I regard this conclusion as thoroughly proved in the case of the commoner species which appear in measur- able proportions (e. g., 1% or 0.1%). Whether or no it also holds good for the rarer ones or for those which did not appear till late must be regarded as of no con- cern for the present. But if the existence of this capacity in a latent con- dition in 1886 is demonstrated by my cultures, it follows that all or most of these new species existed in a latent condition before that date. This latent capacity to mutate, i. e., to produce a series of definite and identical mutations, is therefore a heritable character in my Oenothera Lainarckiana. The particular factor for every given mutation must ob- viously exist separately. And it must be supposed that the various mutations, although they belong to the same group or period, are nevertheless independent of one another. As far as observation goes, this potentiality is always inherited by all individuals. Of course many sowings have given no mutants, and in other sowings certain mutants have been lacking. But this may always have been the result of the smallness of the scale on which the experiment was carried out (whether this was because the available quantity of seed w^as insufficient or that a small culture was all that was necessary for the imme- diate object of the experiment). In larger crops all the commoner mutants appeared as a rule. For large cul- tures like these the seeds of four or even 10-20 seed- parents were needed. In these cases I have always sown the seeds from each parent separately and it has never happened that no mutants appeared amongst the progeny Repeated Mutations Due to the Same Causes. 467 of any single seed-parent. If some mutants were absent, others were more numerous to make u]) for it. The power to mutate is also inherited by the new species. We have already seen several examples of this in § 8 and later in §§ 10-23. For instance O. seiutillaiis is very mutable: it produces pretty regularlv 10-207t' ohlonga; about %% O. lata and about %% 0. nauella (p. 244). O. ohlonga, O. nanella, O. leptocavpa and others gave also rise pretty regularly to the other muta- tional forms in proportions not very different from those in which they are produced by 0. Lauiarekiana itself (§8). And the same is true of crosses, for example between 0. lata and 0. nanclla, 0. ruhrinervis and O. nanella and so forth. Therefore, when a character passes from it^ latent to its active condition, all or apparently all of the other characters latent in it remain so. They are not lost in the process. The question arises : are they ever lost ? 0. hrevistylis and 0. laevi folia seem to afford an answer to this question. They grew in 1887 in the field at Hilversum, they are not known anywhere else and, what is more to the point, they have not been observed as mutants a single time in my cultures, even in cultures of many thousands of individuals. It is therefore possible that they no longer exist in my species in a latent con- dition. It is, of course, possible that my plants may not have descended from the same individual ancestors as those from which these two species arose. So that my obser- vations do not afford a definite proof that the latent char- acters of these species have been Jost. But, inasmuch as the whole lot of the Oenotheras in the wild locality has 468 On the Latent Capacity for Mutation. only sprung from a few individuals, the conclusion that they may have been lost seems to me very probable. It is hardly possible to discover whether single plants in my cultures may sometimes lose the power of giving rise to particular mutations. The negative results of the experiments do not enable us to decide. Far more extensive cultures would be necessary to answer this question definitely by experiment. Meanwhile I incline to the view that the separate latent characters, which become visible by mutation, may be lost sooner or later. § 30. THE LATENT INHERITANCE OF OTHER CHAR- ACTERS IN OENOTHERA LAMARCKIANA. The foregoing argument has led us to regard the capacity for producing mutations as a latent heritable property. The characters of the new species exist po- tentially in the parent species but remain invisible until they are called into active existence by definite external causes.^ That this hypothesis bears strongly on the theory of mutation and on our whole conception of the nature of heritable characters is evident.^ For this reason, I have been trying for many years to render the inheritance of latent characters accessible to experimental study, not only in Oenothera but else- where. The best material for this work seemed to be afforded by monstrosities or teratological phenomena, which used to be looked upon as something fortuitous 'Variabilite et Mutabilite, Rapport du Congres international de botanique, Oct. 1900, Paris, p.i. ^ See Intracellulare Pangenesis, p. 16, and the second volume of this work. Other Characters in Oenothera Lamarckiana. 469 but are now generally regarded as visiljle manifestations of a latent heritable potentiality. In the members of certain families, (which may be large or small) the deviations in question become visil)le so often, that the presumption in favor of a common internal cause becomes very strong. On the other hand the monstrous individuals are so frequently separated from one another in pedigrees by perfectly normal ones that the cause, if it is a continuous one, that is, if it is handed on from one generation to another, must be in- operative most of the time. Finally the appearance, or non-appearance, of the monstrosity in particular individ- uals is dependent on external influences and mainly on nutritional conditions. This latter fact alone seems to me sufficient to prove their presence, and consequently their inheritance, in a latent state. Monstrosities are much more favorable material for this purpose than mutations. For they are accessible to everybody, and dependent for their appearance and de- gree of development on their environment in ways which are easily investigated. Except for hybrids, they afford the best material for studying and elucidating the general principles of latent characters. Monstrosities differ from mutations in that their ap- pearance is partial : by this I mean that they do not affect all the homologous organs of the same plant but only some and usually very few of them; whereas the muta- tions described in this part are absolutely individual. Monstrosities need, by no means, be monstrous. The appellation monstrosity is a very unfortunate one; be- cause, in other species many of these teratological sports are quite normal characters.^ As an example I might ^ Monstmosites taxinomiques, as they are called by De Candolle. 470 On the Latent Capacity for Mutation. quote the pitchers or ascidia, which are analogous to the peltate leaves. It is true that the pitchers often have the form of a cornet or pocket and this restricts the assimila- tory capacity of the leaf; but that only depends on the form of the normal leaf in the species in question. If the latter is auriculate, the pitchers can be quite or nearly flat, and form per- fectly typical peltate leaves ; for example a pitcher-forming Pel- argonium zonale which I have had in cultiva- tion for years and have propagated by cuttings, gives rise, every year, to a num- ber of such peltate leaves especially on short shoots. Simi- larly the first leaves of the twigs of Tilia parvi flora, when changed into ascidia, are almost absolutely flat (Fig. 106 C). I ought to say that my conclusions on the mode of inheritance of monstrosities are chiefly based on species of plants other than Oenothera. They have partly been dealt with already,^ and partly will be de- scribed in the second volume of this work. Fig. io6. Tilia parviiiora. The forma- tion of pitchers from leaves. A, B, ordinary ascidia ; C, a pehate leaf ("flat pitcher") seen from below. ^ Ueber die Erblichkeit der Zwangsdrehungen, Ber. d. d. bot. Gesellsch., 1889, Vol. VII, p. 291 ; Eine Methode Zwangsdrehungen Other Characters in Oenothera Lamarckiana, 471 But these results agree so closely with the facts re- lated here, that there can be little doubt that my general conclusion applies to Oenothera as well. I have only carried out a few direct ex]:)eriments bearing on the inheritance of the abnormalities in Oeno- thera Lamarckiana itself. They relate to tricotyly and variegated leaves, and will be described in suljsequent sections. I have however collected a fair number of ob- servations which favor the argument for this inheritance of latency. A series of abnormalities have occurred in the plants in the field at Hilversum, as well as in my own cultures; some of them seldom, some of them often; but all of them in such a way as to leave no doubt as to their heritability.-^ The only exception to this Is afforded by the cases of virescence, which changes the different parts of the flow- ers into small green bracts. I never found it on the Oenotheras at Hilversum, and only on one example in my own cultures. This was a biennial dwarf which flowered in the summer of 1890, and came very near bearing no seed at all on account of this abnormality. I regard this malformation as due to the attack of some aiifzusuchen, ibid., 1894, Vol. XII, p. 25 ; Uehcr halbc Galton-Curvcn, ibid., 1894, Vol. XII, p. 197; Monographie dcr Zwangsdrchungen in Pringsh. Jahrb. f. wiss. Bot., Vol. XXIII, p. 14 and Over dc crfclyk- heid van fasciatien, Kruidkundig Jaarboek, Gent, 1894, IV Jaargang, p. 72. * For the inheritance of monstrosities see : Erf dyke Moustrosi- teiten, Kruidkundig Jaarboek Dodonaea, 1897, p. 62 ; Over de crfelyk- heid van Synfisen, ibid., 1895, p. 129; Sur la pcriodicite des anomalies dans les plantes monstrueuses, Archiv. Neerl. d. Sc. exactes et nat., Serie II, Tome III, p. 371 ; Ueber die Abhangigkeit dcr Faseiation vom Alter bei zweij'dhrigen PHanzen, Botan. Cenlralbl., Vol. yy, 1899; On Biastrepsis in its Relation to Cultivation, Annals of Botany, 1899; Vol. XIII, No. 51, p. 395; Sur la culture des nionstruosites, Comptes rendus de I'Acad. d. Sc, Paris, Janv. 1899; Sur la culture des fascia- tions des especes annuelles et bisannuelles; Revue gcnerale de Bo- tanique, T. XI, 1899, p. 136; Erndhrung und Zuchtivahl, Biol. Cen- tralblat, Bd. XX, No. 6, 1900, p. 193. 472 On the Latent Capacity for Mutation. disease, analogous to those cases in which parasites have been observed as the causes of virescence.^ Monstrosities often differ from ordinary individual variations in the fact that they are deviations on one side only of the type of the species, whilst the latter deviate from both sides of the mean. In this way we get, when we have a sufficient number of instances of the same monstrosity in a given species, the socalled half-Galton Fig. 107. Oenothera Lamarckiana. Fruit in the axil of a deeply split double leaf; the flower from which this fruit arose had the double number of sepals and petals and stamens as a normal flower, and was elongate in trans- verse section. curves.^ Polymerous flowers, 5-9 partite fruits, split stigmas and even fasciation, all illustrate the same law.^ But the majority of monstrosities are much too rare to "^ Een epidemie van vergroeningen, Kruidkundig Jaarboek, Gent, T. VIII, 1896, p. 66. ^ Berichte d. deutsch. Bot. Gesellsch., Bd. XII, 1894, p. 197-207, with Plate X. ^ See Sur les courhes Galtoniennes des monstruosiies, Bull. Scien- tif. de la France et de la Belgique public par A. Giard^ T. XXVII, p. 396, Avril 1896. Other Characters in Oenothera Lamarckiana. 473 afford material for statistics of this kind unless we breed them for the purpose. I shall give an account of some isolated observations which will, I hope, incite others to further investigation in this field. In Penzig's excellent work on Teratologv (Vol. I, p. 481) the whole genus Oenothera takes up only about half a page. Our O. Lamarckiana is not mentioned there and of course no monstrosities of it are described. The account of the abnormalities of O. biennis, however, is im- portant from our point of view. This plant exhibits an extraordinary tendency to fasciation and often gives rise to pentamerous flowers and 5-9 partite fruits. I can confirm these statements from numerous observations of my own ; I also found the number of stigmas varying in the same way as in Lamarckiana. Clos cites a pistil divided into seven in O. campylocalyx (ibid.), and syn- anthy in Oenothera has been mentioned by Masters (see also Fig. 107). Just as in O. biennis, the chief constituents of the monstrosities presented by the Oenotheras growing in the locality near Hilversum and by the families derived from them in my cultures, are fasciations, pentamerous and polymerous flowers, 5-9 partite fruits, and an in- creased number of stigmas. These, together with varie- gation of leaves and tricotyly in seedlings, which occur in other Oenotheras as well, are the common abnormal- ities; the rest, in my experience, are relatively rare. I shall therefore divide the various instances into two groups, the common and the rare ones. The rare monstrosities were tolerably well repre- sented in the locality at Hilversum, as compared with other wild plants. This was one of the causes of the lively impression which the great degree of variability 474 On the Latent Capacity for Mutation. of this plant made on me at first. I was at first inclined to regard this phenomenon as local, like the actual muta- tions, but had no opportunity to institute an investiga- tion of the matter. Perhaps other observers in other places will be able to fill up these gaps. The chief point for my purpose is the proof that a high degree of her- itable variability was actually in existence in the plants in the field at Hilversum. Tricotyly.^ Tricotylous seedlings are fairly abundant in my cultures; hemitricotyly, which simply consists in the splitting of one of the cotyledons, is somewhat rarer. I have onlv recorded these two abnormalities occasion- ml ally, as compared with the others, because I regarded them as of little importance at first. The following summary of the cases noted will however give data as to their occurrence and frequency in the different fam- ilies. I have started three experiments on the inheritance of these abnormalities by sowing seeds of tricotylous plants in three different families, of 0. nanella, 0. laevi- folia and 0. ruhrinervis : in the latter only, however, was the experiment continued through subsequent genera- tions. In the following summary the years refer to the seed- lings and not to the parent plants of the preceding har- vest. In 1887 I got tricotyls from seeds collected at Hilver- sum; one of them grew up as an 0. lata, but set no seed. In 1890 I found one tricotylous seedling in the chief strain of the Lamar ckiana-i^mWy (p. 224), and in a crop of one of its wa^^j/Za-subfamilies : the latter grew up as a dwarf. ^ See also the second volume. Other Characters in Oenothera Lamar ckiana. 475 In 1890 I got two tricotyls in the LaevifoUa-i^.m\\y (p. 273). In the spring of 1892 I sowed the seeds of the previous harvest on a large scale in the greenhouse of my laboratory. I searched for the tricotyls amongst the many thousands of seedlings, and planted them out in pots in May. There were 71 of them. Of these, 63 set seed in the same year; the seed of each plant was harvested separately and sown. In this crop I counted (March 1893) the tricotyls in 100-200 seedlings for every seed-parent. Altogether I looked through over 13,000 seedlings, and found amongst them about 1% tricotyls on the average. The proportions amongst the individual seed-parents varied between 0 and 2% ; only five contained more ; in these the proportions of tricotyls were 2.5 — 2.7 — ?>.2> — 3.4 and 3.8%. Occasional hemi- tricotyls were discovered in this extensive experiment, and a single syncotyl. Nothing was transplanted from this crop. I found a tricotylous plant of O. nanella in 1889, when 0. nanella first appeared ; in 1892 I also found three tricotyls in this race, which was well established by this time : they all remained dwarf, and set seed. In April 1893 these seeds gave four tricotyls in 800 seedlings, i. e., 0.5%, and one hemitricotyl besides. The tricotyls grew up as dwarfs ; the hemitricotyl was not planted out. In 1890 I found one tricotyl in the sowing of the /aia-family of that year (p. 288). In 1890 I also found one hemitricotyl in the rubrincr- T'W-family (p. 273), a single tricotyl in 1891, and numer- ous tricotyls in the larger crops that were raised in 1892. From these latter I have since formed a tricotylous sub- family which I still cultivate, without, however, being- able to increase the percentage of tricotyls to a better 476 On the Latent Capacity for Mutation. amount. In 1892 I had, besides 20 tricotyls, 6 henii- tricotyls which however I did not cultivate further. The seeds of each of the former was harvested separately and sown; the best of them gave 2.6 — 2.8% tricotyls, but the majority less than 1.5%. The proportion in 8000 seedlings was 0.7% ; there were also 7 hemitricotyls and 2 syncotyls. In 1893 I planted out 70 seedlings derived from five seed-parents which had given from 1.5 to 2.8% tricotyls. In 1894, however, they yielded a harvest in which the percentage was very low, having sunk, in the case of the best seed-parents, to 1%. I planted out about 90 of the best seedlings, with a view to obtaining seed from them. Besides the above mentioned tricotyls in the crop of 1894, I found several hemitricotyls and a single tetracotyl ; also a considerable number of syncotyls and one amphi- syncotyl or seedling with the cotyledons fused together on both sides. This brief resume suffices to show that tricotyly is heritable and that, in my families, it is transmitted from generation to generation even through plants with normal cotyledons, i. e., in a latent state. Fasciation. Split and fasciated stems occur almost every year in my Oenothera Lamarckiana.^ The ab- normality usually occurs in the axis of the inflorescence ; very rarely lower down in the stem or in the rosette. Fasciated plants occurred in all my families with a few inconsiderable exceptions ; but, as far as possible I never chose them as seed-parents. The "split stem" is, so to speak, the lowest stage in the development of this abnormality, and, consequently, ^ Over de crfelykheid van fasciatien, in Botanisch Jaarboek Do- donaea VI, 1894, pp. 92 and 95. Other Characters hi Oenothera Lamarckiana. A77 the commonest. In the first years of my observations in the field I made careful notes on the mode of fascia- tion. There were 20 cases. Of these 14 had split stems (of which one was split twdce) ; 5 formed narrow ''bands," and in only one of them was the top of the stem really broad. These figures are sufficient to show that the distribution of the frequencies of the various degrees of development of this abnormality will form a half Galton-curve. I first found fasciations in the field at Hilversum 1886, in a flowering plant and in a dead one of the ])re- vious year (1885). I found them again in 1887, 1888, 1889, 1892 and 1893 — altogether 15 cases, which were all found in one and the same corner of the field. In 1894 the fasciations were much more numerous and scat- tered over the whole field ; I myself observed six cases ; further ones were observed by others. I observed two cases of fasciation in 1888 in a garden which I had then at Hilversum : one was a plant, which I had raised from a seed which had given rise to a tricotyl in 1887, and had a stem which split twice successively ; the other was a case of fasciation of a three-year-old plant which was planted as a rosette in the garden in 1887. In 1894 I found an example of O. brevisfylis with a narrow fasciation and a case in O. laevifolia was also brought to me. In my cultures the following cases occurred. I had three cases in the Laniarckiana-i?iVL\\\y (p. 224) in two annual dwarfs in 1888 and 1890 respectively; neither of them were grown to maturity. In 1889 there occurred in this family a biennial plant of 0. lata which bore two split lateral twigs. Fasciation also occurred in the lata- family itself (p. 285), but not until the third generation 478 On the Latent Capacity for Mutation. in the year 1894 in which three of the sixteen individuals that were grown showed signs of spHtting in the quite young rosettes; two of these developed strong and tall flowering stems. The fasciation repeated itself, here and there, on these plants. In my later cultures (1895-1900) fasciation grad- ually came to show a predilection, so to speak, for two distinct periods on the life of the plant. First for the seedling stage. In this case the axis divides either above the cotyledons, or above the first two leaves. There arise Fig. io8. Oenothera Lamarckiana. A double rosette of radical leaves at the beginning of July. The cotyledons are still on. in this way two rosettes, whose leaves intertwine because of the closeness of the two axes to one another. In the plant figured in Fig. 108 I have bent the two axes apart and separated the two groups of leaves as much as pos- sible before photographing it, in order to make the figure clearer. When a plant like this grows up it usually has two equall}^ strong stems which attain the same height and begin to flower at the same time. I have only arti- Other Characters in Oenothera Lamarckiana. 479 ficially fertilized such plants when it happened to be ne- cessary to record the progeny of all the plants on the bed on which it stood. Otherwise I have suppressed them, so as not to load the cultures with plants of this incon- venient form. Double rosettes of the sort figured have appeared al- most every year since the beginning of my experiment and very often in large numbers. I found most of them in 0. Lamarckiana, but also in O. lata, O. nanella, 0. Jiir- tella, etc. The second period of the life of the plant in which fasciations are commonest occurs in autumn. If we allow the main stem to go on flowering till autumn its top often broadens out. But most of the plants in my cultures have stopped flowering by that time. Those which have been recorded and are not wanted for other reasons are weeded out, seed plants are decapitated, and plants fertilized by in- sects are so heavily laden with fruits that flowering ceases of its own accord. But O. hrcvistylis is very suitable in this respect, because it practically bears no fruit and sets no seed; a character by which it is easily recognizable even when flowering is over. I have often allowed a whole bed of this species to go on flowering into Novem- ber; with the result that the top of many of the plants began to broaden out either in September or October, and so quickly that, in a very few weeks, it attains a breadth of 1-2 cm. The fan-shaped tops of the stems were often as broad as they were long. As to their fre- quency, I had, for example, in 1898, 20 fasciated indi- viduals in a bed of 49 flowering plants of O. brez'isfylis: that is about 40% ; and in another culture of the same species 63 fasciated and 11 not; that is about 70%. 480 On the Latent Capacity for Mutation. Other new and old species were also much subject to fasciatlon. For example in October 1899 they were par- ticularly numerous in 0. hirtclla and some of its hybrids ; many occurred amongst 0. lata and 0. albida in 1897; and amongst 0. nanella in 1895. In one culture of O. muricata in 1896 there were as many as 80%. fasciated individuals; and in the cross O. muricata X 0. biennis 30% in 1896 and 25% in 1898; and so forth. These and other observations, not worth printing, made in the garden and the field, seem to me to warrant the conclusion that the capacity to produce fasciations under suitable circumstances is heritable in a latent con- dition in the genus Oenothera or at least in the group of the bicnnis-speciQS (subgenus Onagra). Variegation of leaves. I only very seldom found plants with yellow edged leaves — the first time was in 1887; otherwise the variegated leaves were streaked in the ordinary way. I found two of these at Hilversum in 1887 and two again in 1893; I sowed the seeds of the former and got a single variegated plant amongst m.any green ones in 1888. Some seeds collected at Hilver- sum in 1888 gave one annual variegated plant. This abnormality also appeared in my cultures from time to time. For example in the /a/a-family in 1888, 1890 and 1899; in the laez'ifolia-immly in 1889 (6 ex- amples), 1891, 1894 and 1899. In the riibrinej'Z'is-iRmUy in 1893 and 1894, in O. nanella in 1899 and amongst the scintillans of 1890. The Lamarckiana-i3.mi\y gave two in 1888 and two in 1890; the first two were annual and set seed, from which I got a fair number of beautifully variegated rosettes in the following year 1889. In the rubrinervis-idiVcnly there were occasional cases Other Characters in Oenothera Lamarckiana. 481 of an absolutely yellow seedling. These seedlings appar- ently contain no chlorophyl and, therefore, die after tlie unfolding of the cotyledonary leaves. It is worth while going into this case a little more closely. Of the tricot- ylous riibri}ie7'vis-p\2ints whose seeds had been collected separately in 1892 there were several which gave rise to occasional yellow seedlings. One parent plant was par- ticularly fertile in this respect. It gave rise to 498 seed- lings of which 95 were absolutely yellow and 3 had varie- gated cotyledons. The rest were pure green ; these grew well, whilst the yellow ones died young. The proportion of yellow and variegated individuals was therefore 20% : and these abnormal seedlings soon perished. Of the green ones I kept 64, some of them till they ripened their fruits, but. none of them showed any signs of variegation. Inasmuch as variegated plants were never chosen as seed-parents (except in special experiments devoted to that character) and were usually destroyed before they flowered, it follows from these observations that this ab- normality is not only heritable but is maintained in the various families in a latent state from generation to gen- eration. Variegated plants occurred from time to time in other cultures than those of O. Lamarckiana itself, as I have already stated. They also occurred amongst the result of crosses between O. Lamarckiana and its subspecies, and between this and the older species. But the details of these observations are not worth printing. Polymery in the flowers has not been a rare phenom- enon at Hilversum during my acquaintance with the spot. Whenever I examined a large number of flowers I usually found at least one polymerous one. This was also the case in my cultures. In the first years of my 482 On the Latent Capacity for Mutation. experiments I recorded about 30 polymerous flowers partly in the field and partly in my laevifolia-isimily. Be- low is a summary of these cases, giving also the date and place or family in which they occurred. The numbers of stigmas is noted separately (N), but the number of divisions of the fruit (O) have been omitted in some cases. MBE R FORMULA DATE LOCALITY 1 K4C5S8N5O4 1887 Hilversum. 1 K4C5S8N6O4 1887 ( ( 1 K4C5S9 1894 laevifolia. 1 K4C5S10N6O4 1888 Hilversum. 1 K4C6S10N8 1887 laevifolia. 1 K4C4iS8N505 With regard to Papavcr somnifernm polyccphaliim we saw in the first part that it was not possible to separate selection from nutrition. I mean, if we choose our seed- parent, paying attention to the greater or less beautiful development of the circlet of secondary fruits, we in- evitably chose either the strongest or the weakest ]jlants. There seems therefore no escape from the conclusion tbat the variability of this circlet is simply a plienomenon of nutrition and that selection in one direction merely chooses the most highly nourished individuals ; and in the other, the most poorly nourished. In an investigation of this kind one must take into account the susceptible period. One organ will pass through this period earlier; another later, as I have pointed out in the case of the poppy referred to. The same is true of oats and wheat in relation to the amount of water in the soil. In the first vegetation-period this condition influences the number of internodes in the haulm as well as in the panicles, or ears. At the time of shooting, the amount of water in the ground afifects the length of the internodes, and the size of the parts of the inflorescence (the foundations of which have already been laid down by this time) as well as the greater or less fertility of the ears. Much water at the time of shooting increases the amount of straw as well as the yield in grain. ^ Jahrb. f. w. Bot., Bd. 30, 1897, p. 453 and W. Haacke. Ilnhciikcluui:s- mechanische Untersuchungcn, Biol. Cenlralbl., 1900. ^VoN Seelhorst, Journal fiir Landwirthschaft, Bd. 4S. p. 163: Reference in Botan. Ccntralbl, 1900, No. 41, Bd. 84, p. 54. 520 Influence of Nutrition and Selection. The truth of the theory put forward by Schindler and Von Proskowetz that it is impossible to unite many good quahties in one individual, depends partly on the absolute productive capacity and partly on the correct nourishment of the individual qualities at the sensitive period of their development. Johannsen's exhaustive and epochmaking researches into the correlation between seed-weight and nitrogenous contents of barley point in the same direction. The heavier the grain the greater is the amount of nitrogen Vv^hich depreciates the value of the grain. ^ Evidently both vary in the same direc- tion under the influence of high nutrition. But if the sensitive periods for the two should not coincide, the supply of nutriment might be so managed that the weight of the seed is increased without effecting a corresponding increase in those constituents of the seed which are rich in nitrogen. At present it is not possible to do this directly, but Johannsen succeeded in getting a much better harvest without having increased its proportion of nitrogen, by selecting the one value in a positive direc- tion and the other in a negative one. A further series of experiments is necessary before the conclusions (important alike to the pure and applied biologist) based on these remarkable results can be re- garded as thoroughly established. I am simply using them here as a proof of the relation between nutrition and selection in general. For there is yet another method of studying the re- lation between manuring and selection. We can alter both factors; and allow them to operate either in the *W. Johannsen, Ueher die VariahiUtat der Gerste mit heson- dercr Ri'icksicht auf das Verhaltyiiss zwisclien Korncrgewiclit und Stickstoffprocent. Meddelelser fra Carlsberg Laboratoriet, Bd. 4, Heft 4, 1899. Variability as a Nutritional Phenomenon. 521 same or in opposite directions. We can, so to speak, add their effects or subtract the one from the other. If this experiment succeeds it proves that the two plienom- ena are of the same order, and suggests a method of determining their relative importance. I shall therefore describe in this chapter a series of experiments carried out on this principle. They deal with measurable or countable characters which are ca- pable of experimental as well as of statistical treatment. I chose for this purpose the length of the fruits of the ordinary Oenothera Lamarckiana (Figs. 114 and 115, pp. 529 and 530), and also the material employed by LuDWiG which is afforded by the ray florets of Compo- sites and the rays in the umbels of Umbelliferae (Figs. 117-119, pp. 561-565). In the case of the fruits I tried both the addition and subtraction of the factors ; but in that of the ray-florets and the rays of the umbels only the simultaneous operation in opposite directions of heavy manuring and negative selection. The result of the ex- periment was that sometimes the one factor and, at other times, the other predominated. The inquiry into the effect of nutrition (manuring, plenty of room, light and water, etc.) has led to the dis- covery of two principles (foreshadowed in the discus- sions in the first section p. 137) which I think ought to be enunciated here in the interest of a clear under- standing of the whole range of phenomena. These two principles are the following: 1. The younger a plant is the greater is the influence of external conditions on its variability, that is, on the place which its various characters will occupy in the curves of variability of the wdiole culture or race. 2. In connection with this principle the nutrition (^f the 522 Influence of Nutrition and SetectioH. seed on the motherplant has, in many cases at any rate,^ a greater effect on variabihty than nutrition during ger- mination and vegetative Hfe itself. It seems to me that these principles which I only appreciated after many years of experimenting, are now perfectly clear and evident. From these principles there follows the experimenta' method which I call the Principle of the manuring of ihc parent-plant. That is to say, the effect of manuring on variability must be studied not only on the plants which have been heavily manured, but mainly on the generation produced by their seeds. These principles lead to a further problem, the solu- tion of which will perhaps be of great importance from the point of view of the theory of selection. For it is clear that the principle of the manuring of the parent- plants is not necessarily confined to one generation. We shall obviously not get the best nourished seeds from ill favored parents; that is from parents which have them- selves arisen from poor seeds. On the contrary the operation of high nutrition of the seeds must be capable of accumulation through two or more generations. The same is true of low or defective nutrition. But inasmuch (as a general rule) those individuals which exhibit the character dealt with in a high degree are the best nour- ished we naturally choose the most highly nourished individuals as seed-parents when we are selecting for any particular character. In the course of generations the effect of nutrition accumulates, and in this way the devia- tion of the particular character from the original type is continuously increased. The question arises therefore : ^ Sometimes, however, a greater effect can be produced on varia- tion by a good or bad treatment of the seedhngs than by the choice of seeds ; for example in Papaver somniferum polycephalum. Methods of Investigation. 523 what part 6i the result of selection is due to this accumu- lation of nutrition during the succeeding generations? These considerations tend to draw selection and nutri- tion closer and closer together. The exact morle of nutrition seems to me a matter of secondary importance; what is of the first importance is to discover tlie effects of nutrition on the susceptible periods in development, and to study the accumulation of this effect in the course of some generations. Now, just as nutrition reaches its maximum effect, in practice, in the course of a few gen- erations, so the limit reachable by selection is very soon attained.^ The significance of the parallel between these two limits seems to me to be obvious. The closer variability is drawn towards nutrition the wider becomes the gulf between variability and muta- bility. § 2. METHODS OF INVESTIGATION. The effect of nutrition and selection can either be exerted in similar or in opposite directions; the sum of, or the difference between, their effects can thus be de- termined. The general effect of both factors is well known. We are not concerned to prove that the, effect of high nutri- tion is to produce large fruits, and that that of insuffi- cient manure is to produce small ones, and so forth. It seems more important to show that the number of ray- florets can either be increased or diminished by selection : but even on this point there is no doubt whatever. 11ic only question is which of these two factors will pre- ponderate in given instances? ' Part I, § 9, p. 85. 524 Influence of Nutrition and Selection. The experimental part of the work is to provide the nutrition, i. e., generally favorable conditions of culti- vation. The results, however, have to be dealt with by statistical methods which were originated by Quetei.et and Galton^ and have been developed in recent years amongst others by Pearson, Ludwig, Duncker, Daven- port and Amann.^ Let us begin with the latter point and let us seek to delineate the main features of this method in a few short paragraphs in order that we may have a clear idea of the manner in which they are employed. I have chosen Gal- TON^s method as the simplest and most convenient for the latter purpose. Ouetelet and Galton have shown that the indi- vidual variations of men and other animals follow the laws of probability. The deviations from the type of any fluctuating character can be expressed by a curve since they are grouped symmetrically round the type as a center of greatest density. The more numerous the observa- tions the more exactly does the curve of variability coin- cide with the curve of probability. The cause of this parallel is, pretty obviously, that the various deviations from the normal are determined by a vast number of ex- ternal and internal influences. Quetelet asserted that the above law applied to plants and Galton demonstrated it by a few experi- ments. My cultures of races and varieties extending, as they have done, over many years, have given me plenty ^ Galton's Natural Inheritance is indispensable for a proper understanding of the foundations of this method and the reader is advised to refer to it in conjunction with this chapter. ^My experiments were made in 1892- 1894, i- e., before the pub- lications of these authors had appeared. Methods of Investigation. 525 of opportunity of convincing myself of its general ap- plicability in the vegetable kingdom.^ When it is once proved that the form of the emi)irical curve of fluctuations in plants coincides with that (jf the theoretical curve of probability, so far as unavoidable errors in observation permit, the properties of the latter may evidently be ascribed to the former. The most important property of the curve for our purposes is that it may be definitely descril^ccl by two magnitudes, (I) the mean value of the character in ques- tion and (II) the amplitude or extent of variation. The mean value used by Galton is that magnitude which half of the individuals exceed, but which the other half do not attain. This he calls the median. It need not be a mag- nitude which actually exists, but is found by interpola- tion on the assumption that variation is unbroken and continuous. Galton 's median can be determined more easily than the ordinary mean, which is obtained by dividing the sum of all values by the number of observations. It has exactly the same justification and in symmetrical curves the two necessarily coincide. The second factor is the amplitude of variation which finds its simplest expression in the remoteness of the ex- treme variants, provided that the number of individuals is not too small. But the raritv of these extremes makes the determination of these limits by their simple observa- tion largely a matter of chance. Galton therefore uses another value borrowed from the theory of prol)ability, as a measure of the amplitude. This is the magnitude of the deviation from the mean which is exceeded by a ^ See Ber. d. d. Bot. Gcsellsch., Bd. XTT. 1894. p. 197. wlicrc lie previous literature is cited. 526 Influence of Nutrition and Selection. quarter of the individuals and therefore analogous to the so-called ''probable error." He calls it the Ouartile (Q). There is obviously one quartile on either side of the Median (M) ; these are called Qi and Q2. If the curve is symmetrical, the two quartiles have the same value; otherv^ise the dissimilarity of the empirically determined Qi and Qo is a measure of the degree of symmetry of the curve. If the difference between the two is within the range of the error of observation, their mean value Q=(Qi + 02)/2 is the measure of the amplitude of variation of the material under consideration. If we wish to compare the amplitude for different characters together we must reduce them to a common measure. This is done by dividing Q by M /^ We see therefore that Qi, M and Q2 are the numbers which have to be determined by observation. The form of the curve is determined by them and any differences between the curves so determined and the actual figures themselves must be ascribed to errors in observation, at any rate in symmetrical curves. The greater the number of observations which go to make a curve the smaller will these differences be. In the following sections I shall deduce these val- ues from the data ; and use them as a basis for discussion. One advantage of this will be that it will render drawings of the curves superfluous, or at any rate only useful for the purpose of demonstration; and that it will compress the numerical material into a few figures. A few remarks on the subject of construction of these curves (Figs. 115-118) are called for. The number of ordinates is by no means necessarily the same as the * Ed. Verschaffelt, Ueher graduelle VariahiUt'dt von pUans-, lichen Eigenschaften, Ber. d. d. bot. Gesellsch., Vol. XII, 1894, p. 350. Methods of Investigation. 527 number of groups in the tables. This is sufficiently evi- dent where we are dealing with continuous variations such as length. For here the unit chosen is quite an arbi- trary one. For example, if I had measured the fruits of Oenothera accurately to tzvo millimeters only {or if I had measured them in English inches), I should have had fewer ordinates; but if I had measured them to half a millimeter, I should have had twice as many. And in dealing with ray-florets we may consider units or pairs or larger groups. In fact the data may be grouped in any desired way, to suit our purposes. The number of units to be used in the construction of a curve depends in principle on the number of individuals. If this is small, they must be made correspondingly few. In order to do this the two or three groups of figures, in the midst of which the interpolated value of M lies, are united to form a single ordinate; this forms the apex of the curve. We then deal with the groups to the right and to the left of it in the same way. This is the only way in which the peaks and valleys, in the curve, resulting from an insufficient number of observations can be smooth.ed away. Finally, if the various curves are to be compared with one another, the empirical data must of course be reduced to percentages. 11. THE LENGTH OF THE FRUIT IX OENO- THERA LAMARCKIANA. § 3- CORRELATION BETWEEN INDIVIDUAL STRENGTH AND LENGTH OF FRUIT. Let us now consider the relation between the indi- vidual strength of the plant and a character which can be conveniently studied by statistical methods ; partly as an example of the method of dealing with measurements, described in the preceding chapter ; and partly on account of the importance of the question itself. For this purpose I have chosen, as I indicated in the previous section, the length of the ripe fruits of our Evening Primrose (Fig. 114). These fruits are highly variable, not only in plants treated differently but also in the various individuals of the same culture. It is not difficult, as a rule, to find among the longest ones individuals which have twice the length of the shortest ones (Fig. 114 A and C). Such fruits are however very rare; the intermediate ones (Fig. 114 B) are always by far the commonest. In sorting such material we easily find that the frequency of the various values is describable in terms of the Quetelet- Galton law, especially when the number of plants meas- ured is large. Individual Strength and Length of Fruit. 529 Fig. 1151 exhibits these values grapliically. The meas- urements were made on 568 plants; and, in each case, the Fig. 114. OetiotJicra iMmarckiana. Lower sections of three fruit bearing stalks taken from the main stems of three plants, natural size. A, small; B, median; C, long fruits. Culture of 1899. lowest ripe fruit was measured. Hie lengths of these fruits ranged between 15 and 34 millimeters, and their ^ Ueher halhe Galtonctirvcn als Zcichcn discontinuirlichcr I'ana- tion, Ber. d. d. Bot. Gesellsch., Bd. XII, 1894, Tabic X, Fig. i. I he 530 Length of the Fruit in Oenothera Lamarckiana. mean length was 24 millimeters. They agree, as a com- parison with the dotted line shows, fairly exactly with the probability curve. '^^^ ^^> :5H ^ IS J^ '^\ ^ ^ P H t; a < rjcoO t M-l J SOW f-H ^ £a 2"^ a; V be r- ^ c s (U Oi) HH o aj P bh rt UG S If we calculate the values, which we described in the preceding paragraph, from these data we get the follow- ing, in millimeters : Minimum Qo M Qp Maximum 34 15 22.2 24.1 26.1 data for this curve appear on page 200. This curve v^^as the first to demonstrate the appHcability of the Quetelet-Galton law^ to the vegetable kingdom. Individual Strength and Length of Fruit. 531 Maximum and minimum simply refer to the len^^h of the longest and shortest fruits. M is Galton's methan or the mean — the value of which half tlie individuals do not attain, but is exceeded by the other half. This median is found by interpolation, on the assump- tion that the fruits measured as 24 mm. varied contin- uously between 23.5 and 24.5 mm. Qo and Qp are the ordinates which are separated from M by a quarter of all the individuals in each case. They are also found by interpolation. Galton^s quartiles are therefore: Qi = M — Qo and Q^ = Qp—M. Further (Oi + Q2)/2 = Q is the measure of the amplitude of the curve. Lastly Q/M is a measure of this value independent of the size M and of the nature of the variable character; a number therefore by means of which the variation of the fruit length in Oenothera may be compared with the variation of other characters in other plants. These values calculated from the above data are as follows : Qt Q2 Q % 1.9 2.0 1.95 0.08 In the description of the experiment Qo and Qp can be omitted, now that the values 0\, M and O2 have been calculated from the empirical data. A greater degree of accuracy can be attained in these investigations by determining the mean length of the fruits on a given plant instead of determining that of a single fruit only. The question then arises : from how many fruits should this mean value be calculated. I ha\e measured five; and in this section and in the following one I have used the mean of the lengths of the lowest 532 Length of the Fruit in Oenothera Lamarckiana. five good fruits as a measure of the length of fruit in each individual. The reasons which led me to this choice are the fol- lowing. My selection was always an individual one ; that is to say I did not search for the longest and shortest fruits in the harvest, but for the individuals the mean of whose fruits was the longest, and for those the mean of whose fruits was the shortest. But on each inflorescence the size of the fruits gradually decreases from below up- wards with the gradual exhaustion of the plant. Lateral branches often have small capsules ; but as a rule I did not allow these to develop ; I simply broke them off quite young. For that was the only way in which it was pos- sible to grow a large number of healthy plants on the relatively small space at my disposal. The mean length of the five lowest fruits is obviously more or less an arbitrary measure of the mean length of the fruit of a plant. It would be more accurate to measure ten or twenty fruits. We cannot count, with sufficient certainty, on more than twenty ripe fruits per plant; many individuals do not bear so many; for the flowers which open after the first of September usually do not ripen their fruits with us. To measure the mean length of all the fruits on a plant, all the lateral branches would have to flower, and measurements would have to be made of the ripe fruits of all the flowers. But this is absolutely impossible ; at least in our climate, and when the plants are cultivated as annuals. Fortunately the measurements of the five lower fruits gives figures the accuracy of which is sufficient for our experiment. In order to prove this statement by a direct experiment I took 2>d> plants in November 1 893 and meas- Individual Strength and Loujth of Fruit. 533 iired the mean length of the lower five and of the luwer twenty fruits on them. The mean length of the fruits is found by dividing the sum of their lengths by the number of fruits meas- ured. For this purpose the fruits were cut tlirough just at their base (which is marked at its point of junction with the bract by a constriction, so that the measure- ments could always be taken from a fixed point), laid one after another, end on end, in a row, great care being taken in arranging them; and the length of the wliole series was read off. In this way a greater exactitude of the measurements is attained, whilst only one measure- ment is necessary for each plant. Let us choose an example. On one plant the total of the lengths of the five lower fruits was 167 mm., that of the twenty lower fruits 688. The mean numbers were therefore 33.4 and 34.4. The difference is 1.0 mm. In this way the differences for the 38 plants were determined; some were positive, others negative. Neg- lecting the sign the differences were now written in a series in order of magnitude. The result was that in half the individuals the difference was less than 1.25, but in the other half greater. In one case only (hd it reach as much as 4 mm. The probable error is therefore 1.25. In other words: In the highly improbable case of all the differences being positive, or all negative, the figures in our table would have been 1.25 mm. more accurate if I had alwavs measured 20 instead of 5 fruits. Differ- ences of 1.25 and less must therefore be regarded as within the limits of errors of observation. The difi'cr- ences occurring in the experiments to be described are, as 534 Length of the Fruit in Oenothera Lamarckiaiia. a matter of fact, almost without exception considerably larger. With a view to studying the correlation between indi- vidual strength and length of fruit^ I also measured, on the same 38 plants, the length and thickness of the stem and the thickness of the fruits. The thickness of the stem was measured just above the root and round the lowest fruit-bearing internode. The length of the stem was complicated by other factors ; the plants which came up late were abnormally elongated because the light was kept off them by their taller neighbors. I shall therefore not deal further with this character. The data are summarized in the table below. This has CORRELATION BETWEEN THICKNESS OF STEM AND LENGTH OF FRUIT IN OENOTHERA LAMARCKIANA. THICKNESS OF STEM MEAN THICKNESS MEAN LENGTH OF NUMBER OF BELOW ABOVE OF FRUIT FRUIT INDIVIDUALS 16 12 3.85 38.6 2 15 9 3.5 35.0 2 14 9 3.7 31.8 2 12 8 3.5 34.1 3 11 9 3.2 30.2 2 11 8 3.3 32.7 2 11 7 3.4 31.6 3 10 8 3.1 31.9 2 10 7 3.0 30.6 9 9 7 2.9 29.2 2 8 7 3.1 29.7 3 8 6 3.0 29.9 3 7 5 3.1 30.1 3 ^ The method of measuring and estimating correlations between variable organs was originated by Galton. See Correlations and their Measurements. Proc. Royal Soc., Vol. 45, (1888), p. 135. See also Galton, ibid., Vol. 40, p. 42 and Weldon, ibid., Vol. 51, (1892), page 3. Individual Strength and Length of Fruit. 535 been condensed by uniting the Individuals with simihir thickness of stem, and by giving the mean length of theii fruits. The number of individuals per group is given in the last column. The length and thickness of the fruil was measured on the lowest 20 fully developed fruits in the case of each individual in the manner described above. All the values are expressed in millimeters. The table brings out the strong correlation existing between thickness of stem and thickness and length of fruits. For, apart from negligible individual differences, the fruits are longer and thicker, the thicker the stem is. These figures are not sufficiently extensive for the determination of Galton's value r;^ but they serve their immediate purpose well enough. Taken in conjunction with the rest of what we know about nutrition and growth in our plant they tell us that as a rule the fruits are longer, the more vigorous the plant is, and especially that the longest fruits are only found on the strongest plants. Selection in the direction of long fruits therefore chooses the strongest plants whilst selection in the opposite direction must choose the weakest.^ It should be mentioned here that manuring and the choice of good seed are not the only methods of ensuring the vigor of a plant. The distance of the plants from one another, especially in youth, plays a very prominent part in determining this. Plants standing alone usually grow up very luxuriantly; the more plants one grows per square meter the less vigorous are they. Another method of effectively increasing the individual strength of the ^ r = ratio = measure of correlation. ^ It will be seen that this generalization agrees perfectly with the considerations set forth in the critical part of this work. Cf. in this respect, Papaver somnifcrum polyccphalum, pp. 137-140. 536 Length of the Fruit in Oenothera Lamarckiana. plant — the culture of seedlings in pots — will be described in the next section. § 4. THE SIMULTANEOUS OPERATION OF NUTRITION AND SELECTION. The length of the fruits of the large-flowered Even- ing Primrose (Figs. 114 and 115) will afford suitable / / r ^ / ws ,•' / / -3- Q_ ^^ ■^ y / ITS oi •■^^^ -^ y y «:> ^^ T^'^ ^ ^ ^ \, >• -^ 0 + ^'' ^- ---. >--5 1 y ,/' V /- SO CO G r<- Qol y ;- SO *^ ^ ^•^..^ / k «-1 / >, \ \ 10 \ "^^^ \.^ --^ ^__ '"v.. - SJ ^ ~"^ -^ >.o X \ ^ ^ I ID (U fe^ 1/2 o ^ o . u O •;r TO " ^ s material with which to gain an insight into the interaction between nutrition and selection. With regard to nutrition *The curve is constructed from the tables given in the text by reducing the number of ordinates to one-half. All figures are con- The Operation of Nutrition and Selection. 537 I have confined myself to positive changes, but with re- gard to selection to both positive and negative ones; and I have also studied the effect of high nutrition continued through a number of generations without selecti(jn. High nutrition has proved itself superior to the most stringent selection (Fig. 116). Even when combined with negative selection it has improved the mean H^ig. 116B) and positive selection has, in combination with it. only been able to achieve very little more (Fig. 116Cj. And without any selection at all an exceptionally high nutrition has had a far better result than the first two combinations (Fig. 116 D). Fig. 116 shows the main result of this whole series of experiments. The curves B (negative selection) and C (positive selection) are taken from the first year; tliis was done because the two following years brought no further progress in the same direction in spite of con- tinued selection. The experiments extend over three generations ; but all nine curves have not been included since this would have rendered the figure practically un- intelligible. For this experiment, seeds of the laez'ifoIia-\mm\y (p. 273) were sown in 1891 and some of them highly manured with horn-flour (steamed and crushed horns verted into percentages. The distance of the ordinates apart is 7.5 mm. The height of the ordinates is 1% ^2 mm. A. (123 plants) The original curve of the mean fruit-length after the first apphcation of nitrogenous manure in 1891. B and C. The result of manuring the mother-plant in 1891. B (78 plants) The next generation after selection of short-fruited seed-parents. The curve has nevertheless shifted distinctly to the right. 1892. C. (147 plants) The same generation as B. but after selection c^f long-fruited secdparents ; the curve has only shifted a little more than B. 1892. D. After three years of cultivating the seedlings in pots. i. e.. by the most effective form of nutrition, but without selection. The curve has shifted to the right far more than C". 1894 (88 plants). 538 Length of the Fruit in Oenothera Larnarckiana. and hoofs). One plant with very long fruits and two with very small fruits were chosen from the harvest, after its curve (Fig. 116 A) had been determined. Their seeds were sown separately in 1892 and moderately ma- nured; the curves of their offspring were determined. The result was, as Fig. 116 shows, that the length of the fruit had increased in both cases. We see that this occurred in spite of the choice of short-fruited seed- parents in one-half of the experiment. The effect of manure therefore exceeded that of selection (Fig. 116 A, B, and C). HARVEST OF 1891. MEAN LENGTH IN MM OF FRUIT • NUMBER OF PLANTS WITH WITHOUT HORN-MANURE HORN-MANURE 20 1 0 21 4 0 22 7 0 23 11 5 24 21 5 25 25 10 26 20 8 27 14 10 28 10 11 29 3 4 30 4 5 31 1 4 32 1 3 33 1 4 34 0 2 Totals 123 71 To go into the details of the experiment : it must be related that the family in question (1887-1890) was grown in very rich soil but without manure. I selected The Operation of Nutrition and Selection. 539 at random two samples of seed, each of wliich liad been well mixed, from those harvested in 1890. One was sown on two beds of about 2 square meters each ; which had received a dressing composed of 5 kilos of horn- flour, and 10 kilos of this manure per square meter respectively. This culture gave 123 healthy plants with ripe fruits. The greater part of the seed was sown on five beds, of 2 square meters each, some of which re- ceived no manure at all, whilst others were given % to 2% kilos of ordinary guano per square meter. Of this crop 14-15 plants were chosen at random from each bed and employed for determining the second curve. The total length of the five lower fruits was meas- ured in millimeters (see p. 532) and the mean length of fruit calculated from this. In this way I obtained the data in the table on p. 538. The following figures can be calculated from the first generation of this experiment, raised from the same lot of seeds : Qt M Q, With horn meal 1.3 25.2 1.5 Without horn meal 1.9 27.2 2.4 Hilversum 1.9 24.1 2.0 In the third row I have written the corresponding values for O. Lamarckiana (1893), for the sake of com- parison (see pp. 530, 531 and Fig. 115). The cultivated plants have, it will be seen, a slight advantage over the wild ones, which is smaller in the case of the plants which had horn-flour than in those in the control experiment. The horn-flour culture shows a slight decrease in the amplitude of variation; the control experiment a trifling increase. The horn-flour culture was the only one which was 540 Length of the Fruit m Oenothera Lamarckiana. continued. Three individuals from it were chosen as seed-parents; the mean lengths of their five lowest fruits measured as above were 20.6, 20.6 and 32.6. For the sake of greater certainty I also measured all the ripe fruits on these stems (27-33 fruits each) and found the mean values for the plants to be 19.0, 19.2 and 31.3. The seed of the single long-fruited stem was sufficient for the culture of 1892; but I had to take two of the short-fruited ones to get a sufficient quantity of seed. HARVEST OF 1892. MEAN FRUIT LENGTH NUMBER OF PLANTS IN MM. K L 23 2 0 24 2 0 25 4 0 26 5 5 27 7 5 28 12 4 29 5 8 30 5 10 31 5 17 32 12 13 33 6 13 34 6 16 35 5 16 36 2 13 37 0 5 38 0 10 39 0 6 40 0 1 41 0 2 42 0 0 43 0 3 Totals 78 147 In 1892 I set apart a bed of 4 square meters for each The Operation of Nutrition and Selection. 541 of these two cultures: it was dressed with Y:\ of a kilo of dried cow manure and % of a kilo of steamed horn- flour per square meter. This has proved the most satis- factory manure I have tried: plants do not react, in the long run, to larger quantities. Otherwise the treatment of the plants was the same as in the former year; they grew healthily; tliere were 147 individuals from the long-fruited parentage, and 78 from the short-fruited. The fruit lengths were deter- mined in the usual way ; and the numher of plants which exhibited the various fruit-lengths are given in the table on page 540, in which K signifies the offspring of short- fruited and L that of long-fruited parents. From the table on page 540 together with result of the sowing of 1891 the following values can be calculated : under S are given the fruit lengths of the seed parents. S Q, M Q^ Harvest of 1891 — 1.3 25.2 1.5 (Fig. 116^) 1892. Short fruited culture 20.6 2.5 29.9 2.6 (Fig. 116 ^) 1892. Long fruited culture 32.6 2.6 33.4 2.4 (Fig. 116 C) and further: MINIMUM MAXIMUM Harvest of 1891 20 mm. 33 mm. 1892. Short fruited culture 23 mm. 36 mm. 1892. Long fruited culture 26 mm. 43 mm. We find therefore that the mean fruit-length has in- creased considerably in both cultures of 1892, and that this increase has been more considerable when long- fruited seed-parents have been chosen than when short- fruited ones have. The same is true of the extremes of the crops : fruits as small as those which occurred in 1891 did not occur in the cultures of 1892: on the otlicr hand the size of the longest fruits increased consider- 542 Length of the Fruit in Oenothera Lainarckiana. ably (the maximum increase being almost a third of the original length). For the sake of further discussion, we may sum up this result briefly in two theses : 1. In both cases the length of the fruit increased. 2. With the selection of a long-fruited seed-parent this increase was considerably greater than when a short-fruited one was selected. It is evident that the latter fact is simply the result of selection. All the other conditions of the experiment- were exactly the same, and the difference in the results is exactly what one would expect as the result of selec- tion. We need not therefore enter further into it. But it is a very different matter that the length of the fruit increased in both cultures and especially that this happened in the case of the choice of short-fruited seed-parents. This cannot have been the result of selec- tion, and the only other possible cause can have been the heavy manuring of the parent-plant with horn-flour. In the long-fruited culture the mean fruit-length (33.4 mm.) was larger than the corresponding value in the seed-parent (32.6 mm.). The known principles of selection, and particularly Galton's researches on re- gression, make the interpretation of this result as the effect of selection, impossible. Selection would, of course, effect an increase in the length of the fruit, but the new value would have to lie between the original mean and the fruit-length of the seed-parent. Here however it was greater than that of the seed-parent, and this can only be ascribed to the heavy manuring of the parent- plants.^ ^ I have often observed this effect of the manuring of the parent- plants in cuhures with other species, for example in Ranunculus bulbosits in pleiopetaly. See the second volume. The Operation of Nutrition and Selection. 543 Let us now turn to the amplitude of variation (Q). This was, as the above table shows, the same in both cultures in 1892 and double as large as in the horn-flour culture of 1891. In this latter it was, in fact, smaller than in the ordinary cultures (p. 539). The amplitude of fluctuation is well known to be brought about by the multiformity of internal and external conditions which afifect development, and it is obvious that heavy manuring will tend to level these differences down. We shall refer to a parallel result when we come to describe the contin- uation of the short-fruited race. Finally, it will be seen that Q\ and Qo have remained equal to one another and therefore that the curve, in spite of the shifting of its apex, has remained symmetrical. Regarded from the methodological point of view, this experiment contains a warning to keep the external conditions, particularly those of manuring, as constant as possible; and not to be too ready to interpret any changes that may occur as the effects of selection. As already stated I have cultivated the two races for two more years under exactly the same treatment ( 1S^\>. 1894). The long-fruited race underwent no further im- provement; in fact they deteriorated a little. This result is an illustration of Hallett's principle (see ]). ll(V), which enabled him to evolve his new varieties of cereals. During the first year of his experiments notable progress was made; but after that, further selection either made ver}^ slight further progress, or only served to fix what had already been attained. In 1891-1892 I left pollination to the agency of in- sects, but in the summer of 1893 I pollinated the fl 1 ^ '\ y \ \ ,iS< }S y y /,^' \ / \ \ _^ y^-^ \ ^ ^^^ ..J-- ^ __.- '^-^''^ir^ ^y—^:. J,i 66 3S !iO U2 Fig. 117. Anethnm gravcolcns. Curves of the rays of the temiinal umbel. The numbers under the abscissae refer to the rays of the primary umbel. In accordance with the rule discussed on page 527 the number of ordinates is half the number of groups in the table. The figure 8 therefore means eight and nine rays and so forth. A. (56 plants) Curve of 1892, irregular on account of the small number of individuals. It is also asymmet- rical being drawn out more to the right. B. (518 plants) Curve of the following generation 1893. As a result of nutrition and selection it has be- come nearly symmetrical. Lamarckiana, where it was 0.08 (p. 531). The number of umbel-rays is therefore, when measured in this way. twice as variable as the fruit-length in Oenothera. The asymmetrical curve of the year 1892 ((?2>(?i) 562 Curves of Compositae and Umhelliferae. became symmetrical in the two following years as can be clearly seen by comparing curve A (1892) and B (1893) in Fig. 117. § 7. EQUILIBRIUM BETWEEN THE EFFECTS OF SELEC- TION AND NUTRITION. These experiments were conducted with Chrysan- themum segetum (Fig. 118), Coreopsis tinctoria and Bideiis grandiflora (Fig. 119). Of the former I re- ceived, in exchange from botanical gardens, a certain number of packets of seeds from various sources. The contents of the various packets were mixed before sow- ing. This multiple origin showed itself clearly in the number of ray-florets; for the curve expressing their variation was not homogeneous as usual but had two apices (Fig. 118 A). One peak was at 13-14 florets, the other at 21. This must mean that there were two races present, mixed together.^ This interpretation was proved to be correct in the following year (1893) when, as a result of choosing seed-parents from one of the supposed races (the 13- rayed one) every trace of the second peak disappeared (Fig. 118 B). It did not appear again in 1894. Two-peaked curves occur also in man, and here again they are regarded as the expression of the incomplete fusion of types which have interbred for many cen- turies.^ Such curves have also been observed by Bate- son^ and Weldon* in their important investigations * Eine sweigipfelige Variatwnscurve, Roux' Archiv fiir Ent- wickelungsmechanik, II. Band, 1895, P- 52. See also the second volume. ^Otto Ammon^ Die natilrliche Auslese beim Menschen, 1893. ^Bateson^ Proc. Zool. Soc, London, 1892, p. 585. *Weldon, loc. cit. Equilibrium Between Selection and Nutrition. 563 into the variability of various animals {Forficula, Car- cinus, Xylotropus, etc.), and repeatedly by Ludwig in plants. Bateson^ in his work on discontinuous varia- tion, has emphasized the great importance of such cases to the student of variability, and given examples of them.^ Tbe two-peaked curves are separated by him as cases of dimorphism from the ordinary or mono- morphic curves. The duplicate character of curves can be brought about by the most various causes. Giard, for example, has made the remarkable discovery that a dimorphism of this kind may be brought about when some of the individuals in a locality are infested by a parasite. Thus Carcinus mocnas which were infected by Sacciilina car- cini or Portunion moenadis differed widely from the normal ones.^ But the double curves in plants can be dealt with ex- perimentally much better than those in animals or men. Let us now proceed to the description of the experi- ment with Chrysanthemum segetum. In 1892 this ex- tended over an area of 2 square meters. The number of individuals, when I came to select them was 97. For making the curve only one head was taken from each individual, the so-called primary one at the top of the main stem. All plants whose terminal inflorescence had 14 or more ray florets were pulled up immediately; fourteen plants with 13 such florets and one with 12 were saved. ^W. Bateson, Materials for the Study of J^an'ation, London, 1894, PP- 39-41- Comptes rendus, T. CXVITT, 1894, No. 16 (April 16). p. 870. This case has now been thoroughly investigated by Geoffrey Smith, Fauna and Flora of the Gulf of Naples. Volume on Rhizocephala (Note of 1908). 564 Curves of Coinpositae and Umbelliferae. In 1893 the seeds of these 15 plants were sown on 8 square meters of ground; 162 plants were raised. All of these were weeded out, with the exception of 12 plants whose terminal heads had 11-12 ray florets. That is to say, the seed-parents exhibited an advance in the negative direction as compared with the previous year. CHRYSANTHEMUM SEGETUM. NUMBER OF RAY- NUMBER OF PLANTS FLORETS 1892 1893 1894 8 0 2 0 9 0 1 1 10 0 0 3 11 0 7 8 12 1 13 31 13 14 94 221 14 13 25 50 15 4 7 8 16 6 7 5 17 9 1 4 18 7 2 3 19 10 0 1 20 12 3 2 21 20 0 1 22 1 0 0 The curve of 1892 was therefore dimorphic ; those of 1893 and 1894 monomorphic. From the two latter the following data have been calculated. ^ears Seed parents Qi ' M Qz Q M 1893 12—13 0.4 13.1 0.6 0.04 1894 12 0.4 13.1 0.4 0.03 Increase 0.0 0.0 —0.2 In the third year, 1894, the culture occupied 6 square meters : it was raised from the seed of three plants of 1893, each of which had 12 rays in the terminal head, and only 13 in the later ones. The number of plants at the time of selection was 338. Equilibrium Between Selection and Nutrition. 565 1 B J8 93 1 ,' \ 1 / 1 / \ 18 / / 92_ ' / 1 A / / / / i \ / / / 1 1 / \ / / \ 8 10 n u 16 18 no Fig. Ii8. Chrysanthemum segetum. Curves of the ray- florets of the terminal in- florescences. Under the ab- scissa are the numbers of these florets. The number of ordinates is reduced to the half; 8 therefore means 7-8 ray florets etc. (height : I mm=i%). A (97 plants) Dimorphic curve from a mixed sowing 1892. B (i62plants) Bythe selec- tion of plants belonging to the group with 13-14 florets as seed-parents the curve has become monomorphic in the next generation, 1893. — The curve for 1894 was almost exactly the same as that for 1893. 10 11 12 Fig. 119. A, Coriandrum sativum (334 plants). Curve of 1894. The numbers under the abscissa refer to the number of rays of the primary umbel. Height of the ordinates : I mm=i% of the individuals. B, Bidens grandiiiora (152 plants). Curve of 1894. The numbers under the abscissa signify, in curves B and C, the numbers of ray-florets in the primary inflorescences. Height as in A. C, Coreopsis tinctoria (495 plants). Curve of 1893. 566 Curves of C o in po sitae and Unihelli ferae. During these three years the germinating power of the seeds, and the individual strength of the whole cul- ture increased considerably. In the first year I only got the proper number of plants per square meter by sowing a large quantity of seed; in the following year less seed was sown and the crop was correspondingly scanty: in 1894 more seed again was sown and many seedlings had to be weeded out. The result of these observations is summarized in the table on page 564. (See also Fig. 118.) Selection, we see, in this case has been unable to effect any further alteration in the mean or in the am- plitude of variation. It has simply maintained the mean at the same point. We come now to Coreopsis tinctoria (Fig. 119 C). The inflorescences of this beautiful composite have, as a rule, 8 ray-florets. Yet this number varies on the same individual as well as from plant to plant. I obtained my seeds in the winter of 1891/92 from MM. Vilmorin- Andrieux & CiE. of Paris, and tried simultaneously to increase the mean number of ray-florets by manuring, and to diminish it by selection. The result was that the mean number maintained it- self almost unaltered at 8, that is to say that the effects of the two opposing factors neutralized one another. My cultures in the years 1892, 1893, 1894 extended over 1, 8 and 6 square meters respectively. I determined no curve for the first year; the vast majority of the plants had 8 rays; occasional ones 9 and 10; and fewer still 11, 12 or 13. These were all pulled up: I only saved a few, most of which had 7 ray-florets. In 1893 I had 495 plants; all those which had 8 or more ray florets were pulled up as soon as the rays could Equilibriuni Between Selection and Nutrition. 567 be counted, and recorded. About 60 plants with 5, 6 and 7 florets were left over. Amongst these a further selection was made of those whose branches were richest in 5-7 rayed inflorescences. Immediately after the weed- ing out had taken place these plants were deprived of all inflorescences which were either in flower or over, in order that all their seed might result from pure fertili- zation. Of the twelve plants thus treated I chose the four strongest and most fertile as seed-parents for next year's crop: their terminal inflorescences had 5, 5, 6 and 7 ray-florets respectively, and their lateral branches bore heads with few rays. In 1894 I obtained from their seeds 256 flowering plants and determined the curve from them in the usual way. The figures I obtained are summarized in the table on page 568. (See also Fig. 119 C.) The third experiment was carried out with Bidcns grandi flora (Fig. 119 B). In this species the inflor- escences are usually five-rayed, but the number, here also, is subject to variation and wuthin limits similar to those in Coreopsis. In the flowers of Dicotyledons the number 5 is as a general rule remarkably constant, and probably in a great many cases hardly subject to any fluctuations. The question naturally presents itself : why is this number inconstant in this case? This problem has however not yet been investigated; a solution of it would of course be of fundamental importance to the student of varia- bility. ^ I obtained my seeds in the winter of 1891/92 from Messrs. Haage & Schmidt in Erfurt, sowed a square ^The question Is whether a cyclic arrangement diminishes the variability of the number of the parts involved and if so: why? 568 Curves of Compositae and Umbelliferae* COREOPSIS TINCTORIA. NUMBER OF FLORETS NUMBER OF PLANTS 1893 1894 1 2 3 4 5 6 7 8 9 10 11 12 13 0 0 1 0 2 13 49 311 76 28 12 3 0 2 0 1 3 5 10 53 191 14 5 2 0 0 From which I have calculated the following values Year Seed parents Qi M Q2 Q M 1893 7 0.4 8.1 0.3 0.04 1894 5, 5, 6 and 7 0.5 7.9 0.3 0.05 Increase 0.1 -0.2 0.0 meter with them in 1892 and chose as seed-parents a few examples on which I had seen 3 and 4 rayed in- florescences. In 1893 I sowed 8 square meters with their seeds and got 557 flowering plants; and made a curve of the num- bers of rays of their primary heads. I chose a series of plants with four-rayed inflorescences and when their seeds were ripe made a further selection of three of them which had exhibited the lowest numbers of rays in their other inflorescences. From their seeds I raised, on 6 square meters, 152 flowering individuals. I again made a curve from their Obliteration of Effect of Nutrition by Selection. 569 terminal heads; and found the figures given in the table on this page. Here again as in Chrysanthemum and Coreopsis there was no marked effect on the curve while nutrition and selection were operating in opposite directions. BIDENS GRANDIFLORA. DUMBER OF RAYS NUMBER OF PLANTS 1893 1894 2 3 4 5 6 7 8 9 10 1 10 31 355 113 40 6 1 0 2 8 16 117 6 2 1 0 0 Totals 557 152 From which I have calculated the following values ^ear Seed parent Q^ M Qz Q M 1893 4 0.4 5.2 0.5 0.09 1894 4 0.3 4.9 0.4 0.07 Increase —0.1 -0.3 -0.1 § 8. OBLITERATION OF THE EFFECT OF NUTRITION BY SELECTION. The experiment was carried out partly with Corxan- drum sativum, the common Coriander, and partly with Madia elegans, as species related to the oil Madia (Madia sativa). The seeds of the former were obtained from MM. 570 Curves of Compositae and Umhelliferae, Vilmorin-Andrieux & CiE. in Paris and sown on a bed of one square meter. When I came to select them the number of plants was 45; the vast majority had five rays in the primary umbel, some 6, very few 7 and 8, and none any more. Two plants had four-rayed ter- minal umbels, and on one of them most of the secondary umbels were also four-rayed. I only harvested seed from the two latter plants. A curve was not determined. Next year the culture extended over two square meters, and the number of adult plants was 52. Of these the great majority had 5-rayed terminal umbels. Of the three plants which were chosen as seed-parents one had a 3-rayed terminal umbel, the two others 4-rayed ones. The seed of these three plants was sown separately in 1894; each lot on two square meters. The number of individuals at the time of selection was 334, amongst which there occurred two with a two-rayed terminal umbel, a result which means an advance in a negative direction on the stage attained in the previous year; but may, perhaps, be partly attributed to the larger number of individuals in the culture. The plants were harvested separately on the three beds, but the results are all given together in the table on page 571. It is very curious that the offspring of the three-rayed parent exhibited on the average a greater number of rays than those of one of the two four-rayed parents. The character of the parent is therefore only an imperfect guide of the average character of its progeny. As shown in the following table, selection has suc- ceeded, in spite of the heavy manuring, in reducing the number of rays in the umbel by almost a whole unit. Obliteration of Effect of Nutrition by Selection. 571 CORIANDRUM SATIVUM (Fig. 119 A). RAYS IN TERMINAL UMBEL NUMBER OF INDIVIDUALS 1893 1894 2 0 2 3 1 43 4 8 146 5 30 133 6 12 10 7 1 0 From which I have calculated Year Seed parent Qi M Qz Q M 1883 4 0.5 5.1 0.4 0.09 1894 3, 4 and 4 0.5 4.3 0.6 0.13 Increase 0.0 -0.8 -f-0.2 The amplitude of variation Q/M is intermediate be- tween those of Oenothera (0.08) and Anethnm (0.16). We come now to the second series of experiments, carried out with Madia elegans. This species is more suitable for experiments with selection than either Bidcns or Coreopsis. In the first place the growth is much more uniform especially in youth ; in the second, the number of ray-florets is consid- erably larger ; and — last and most important point of all — there is m^uch less partial variability in this case. This means that the various inflorescences on the same ])Iant differ from one another only slightly in the number of their rays (at least in my race) so that the numl)cr on the terminal inflorescence can be more justly regarded as characteristic of the whole plant. I obtained the seeds from Messrs. Ha age & Schmidt in Erfurt. I sowed them in 1892 over a square meter of soil. Most of the individuals had 21 ligulate florets, 572 Curves of Compositae and Umhelliferae. many had 20 or 22, a few had 23 or 25. These were all pulled up. There remained 6 plants with 16-19 rays; their seed was harvested in autumn. In 1893 this experiment occupied 8 square meters. I made a curve of the ray florets of 411 plants that were raised on it. Eight plants with 13-15 florets were chosen as seed-parents. Of these I chose the best three with 13-rayed terminal inflorescences, sowed their seeds on 6 square meters and obtained no more than 213 adult plants as a result of an accident by which a number were lost. The variability in the number of ray-florets of these plants is given in the following table, together with those of the 1893 crop. MADIA ELEGANS. NUMBER OF RAY -FLORETS NUMBER OF PLANTS 1893 1894 12 13 14 15 16 17 18 19 20 21 22 1 IS 11 18 18 43 63 101 82 54 5 0 12 16 18 20 29 32 50 23 12 1 Totals 411 213 From which I have calculated the following values Year Seed parent Q. M Q^ Q M 1893 16—19 1.5 18.9 1.1 0.07 1894 13 2.1 17.9 1.3 0.09 Increase 0.6 —1.0 0.2 Summary. 573 That is to say a definite though shght decrease in the mean number as the resuh of fairly rigid selection. § 9. SUMMARY. In conclusion, I will give the results described in the last three sections in a short summary. The general result is that they are in complete har- mony with those obtained with Oenothera Lamarckiana and O. rubrinervis (§§ 4-5) and can therefore be re- garded as a confirmation of these. They show that when nutrition and selection are brought into conflict, in some cases one of them triumphs, and in others the other. In Anethum it was nutrition, in Coriandrum and Madia selection, in Chrysantheumm, Coreopsis and Bidens it was a drawn battle. The differ- ences between the results of the individual experiments has evidently more to do with the relative power of these two factors than with any putative differences between the species investigated. For obviously the same amount of manure per square meter means a very different amoujit of nutriment for different plants; and, on the contrary, selection, however stringent it may be, is effec- tive in analogously different degrees. We conclude, therefore, that selection and nutrition influence the plant in the same direction and that it de- pends on circumstances whether the one or the other of the two preponderates. Perhaps the simplest and clearest way of proving this generalization is to exhibit the means of the num- bers of rays and ray-florets of the primary inflorescences of all the species investigated. 574 Curves of Coinpositae and UmhclUfcrae. Increase 1892 1893 1894 1893—1894 Aiiethum graveolens 18.3 21.2 25.2 +4.0 Chrysanthemum segetum 13 — 14 13.1 13.1 0.0 Coreopsis tincioria d=8 8.1 7.9 —0.2 Bide7is gi^andiflora ± 5 5.2 4.9 —0.3 Coriandrum sativum ±5 5.1 4.3 — 0.8 Madia elegants ±21 18.9 17.9 -1.0 The varying result of the conflict between heavy manuring for three years and the selection of individuals with a small number of rays is shown by the figures in the last column. I shall now exhibit the values for Q/M in a single table. Q, as we have already said, may be made inde- pendent of the nature of the varying character by di- viding it by M ; and in this way the amplitudes of varia- tion of the different characters may be compared with one another. The subjoined values for 0/M are the means calculated from two or more generations in all the above cases. I have added Oenothera Lamarckiana to the list. M Anethwm graveole^is 0.16 Coriandrum sativmn 0.11 Oenothera Lamaixkiafia .... 0.08 Bidejis grajidiflora 0.08 Madia elegans 0.08 Coreopsis tincioria 0.04 Chrysanthemum segetum . . . . 0.03 The observed amplitudes of variation, estimated by this measure, differ considerably from one another. But they are, of course, also affected by selection and nutri- tion. The phenomena of fluctuating variability are, there- Summary. 575 fore, caused by these two factors. The amount of hte devi- ation of any given character from its mean is determined partly by selection, i. e., by the characters of its parents and grandparents and partly by nutrition, i. e., by the ojjcration of external influences on the individual itself. But the characters of the ancestors were also determined by the conditions of life; so we arrive at the conclusion that the phenomena of variability in the strict sense of the term, that is, the individual deviations from the mean of the species are solely caused by external conditions. Only it must be remembered that nutritional influences may be cumulative over several generations, inasmuch as only the best individuals will bear the best seed. Fluctuating variability therefore falls within the province of the physiology of nutrition. The external causes of mutation are, on the other hand, as yet wholly unknown. INDEX. Abanderungsspielraum, 149. Acacia, 363. Acclimatization, 86, 92. Aconitum Napellus, 59. Acquired characters, 130, 135, 213 Adaptation, 149. Aesculus Hippocastanum, 62. Affoler, 492. Ageratum mexicanum, 194. A gratis segetum, 320. Alnus laciniata, 193. Alpine plants, 145. Ammon, 149, 154. Amphisyncotyly, 476. Analytical tables, 448. Anethum graveolens, 559. Animal kingdom, 3. Apples, 179. Arctic plants, 145. Ascidia, 470. Associated characters, 329, 375. Association of units, 3. Atavism, 20, 136, 196, 364. Atavistic characters, 311, 362. Atavists, 79. Aulax Hieracii, 407. Auslese, Natiirliche, 156. Avena fatua, 98. Bailey, 181. Barley, brewer's, 116. Barley, Chevalier, ill. Bateson, 6. Beans, 47. Beans, Curve of, 48. Beech, copper, 192. Beets, Sugar, 99; Sugar in 40- 000, 103. Beissner, 364. BcUcvue de Talavera, 178. Beseler, 81. Betula alba laciniata, 193. Beyerinck, 406. Bidens grandiiiora, 562, 567. Biscutella laevigata, 62. Blankinship, 56. Bleu^ Alfred^ 47. Bonnier, 144. Bourgeons multiples, 487. Brassica oleracca, 192. BucKMANN, 90, 124. Bud variations, 53. Buds on cotyledons, 488. BuRBANK, 115. BURKILL^ 160. Cabbage, Scottish, 124. Caladium, 47. Calliopsis tinctoria, 197. Carcinus mocnas, 563. Carriere, 90. Carrot, wild, 89. Carter & Sons, 496. Cclosia cristata, 125. Ccntaurca Cyanus, 134. Cereals, 106. 578 Index. Characters, innate, 136; specific, 186 Chelidonium laciniatum, 189. Chrysanthemum indicum, 99; sc- getum, 151, 562, 565. Clos, 360. Cockscomb, 125. CoefBcients of mutation, 337, 351. Compositae, 556. Conditions, external, 2)7- Constancy, 507. Convariants, 51. Cope, 62,. Coreopsis tinctoria, 566. Correlation, 115, 160, 328. Coriandrum sativum, 569, 571. Cornflower, 134. Corn, Tarascora, 95. CosTANTiN, 28, 86, 99. Cotyledons, split, 488. Crab apple, 120. Crosses, 298, 299, 300. Crossfertilization, 76. Crossing, free, 78. Curves, transgressive, 56, 358. Cuttings, 386. Crowding of plants, 139. Cyclamen latifolium, 191. Dahlia, double, 185; striped, 54. Daniel, 161. Darwin, 153. Darwinism, 39. Datura Tatula, 21, 170. Daucus Carota, 89. Davenpobt, 56, 430. De Candolle, Alphonse, 30, 86. Delboeuf, 208, 254. Delpino, 186, 364. Devariants, 51. Dollo, 63. Donkelaar, 185. Drab a verna, 22, 494. Duncker, 49, 131, 159. Duration of selection, 85. Elements of species, 4. Elite, 108. Ericksson, 175. Eucalyptus Globulus, 364. Eupatorium, 407. Fagus, 144. Fagus asplenifolia, 193. Fan Type of Plotting, 52. Fasciation, 476. Fertilization, 153, Filaments, fusion of, 488. Flax seed, 128. ■ Fleeming Jenkin, 2>7- Fleshiness in fruits, 120. Flowers, bimerous, 483 ; trimer- ous, 483 ; with bracts, 486. ForHcula, 563. Forms, intermediate, 504. Fragaria alpina, 32, 192 ; mono- phylla, 193. Fruit trees, 179. Fruits, pentamerous, 483. Fruwirth, yy. Galls, 406. Galton, 49, 136; curves, half, 51 ; median, 525 ; polyhedron, 53. Gandoger, 28, 174. Gauchery, 360. Gemmules, 38. Genealogical tree, 6. Generations, intermediate, 128. Germination, belated, 261. Giard, 153. *% Gideon, 181. Godron, 24. Groups, nebulous, 495, 511. Groups of units, 3. GuLiCK, 208. Index. 579 Haacke, 519. Habit, 507. Hallett, 1 10. Haycraft, 158. Hedera Helix arhorea, 44. Heinsius, 151. Helianthemiim vulgare, 146. Herbert, 36. Hertwig, O., 57, 62, 154, Heterogenesis, 68, Hieracium, 495 ; umbellatum,4oy. Hitchcock, 437. Hoek, 431. Hoffmann, z7- Homegrowing, 127. Hooker, 58. Horn meal, 537. Hybridization, 4, 47, y6, 96; laws of, 301. Ilex Aquifolium, 196. Index Kewensis, 21. Inheritance, latent, 468. Jaggi, 192. Jagtlust, 266. Janet, 211. Jencic on pollen of hybrids, 418. Jenkin, Fleeming, 37. johannsen, 520. Jordan, 22, 127. KiDD, 157. Kleebahn, 167. Knight, 133. KoBUs, 148. Kolliker, 68. kollmann, 49, 155, Korschinsky, 46, 67. KUHN & Co., 102 Lactuca, 144. La Gasca, 177. Lamarck, 17. Layering, 116. Leaves, concrescence of, 485 ; pel- tate, 470; split, 484. Le Couteur, 177. Leveille, 437. Lignier, 153. LiNDLEY, 127. Linnaeus, 20. Linum crepitans, 194. Lomaria proccra, 59. LuDWiG, 49, 521. LuTZ, Anne M., 325. Lysimachia vulgaris, 407. Mac Leod, 134, 516, 518. Madia elegans, 84, 571. Maize, number of rows, 125 ; of Baden, 94; selection with, y2>- Malthus, Essay on Population, 34. Mansholt, van, 127. Maple, cutleaved, 185. Matricaria Chamomilla discoidea 196. Matthiola incana, 204. Median, Galton's, 525. Mercurialis laciniata, 192. Metzger, 93. Migration, 206. Minus-variations, 51. Modifications, nutritional, 146. Monde ambiant, 63. Monocotyledons, 186. MoNS, VAN, 179. Monstrosities, 469; taxinomiqucs 469. Mueller, on Maize. 71. MuNTiNG, 127. 182. Mutability, indiscriminate. 205. 419; periodic, 205. Mutants arising from new spe- cies, 297; percentage of 40 ^c . 264. 580 Index. Mutation, 24; coefficients, 415, 504; in nature, 300; in pre- Darwinian days, 1 1 ; laws of, 247; progressive, 6, 68; retro- gressive, 6, 68; smaller than variations, 55 ; theory, 3. Nanisme, 360. Nature abhors perpetual self- fertilization, 153. Naudin, 36, 169 NlLSSON^ 114. Nomen specificum, 19. Nomenclature, 172. Nutrition, and selection, 515, 536 ; and variability, 516; Effects of, 131; high, 551. Nutritional modifications, 132, Oats, Anderbecker, 81. Oenothera alhida, 229, 349; bi- ennis, mutation period, 440, 495; biennis, type, 431; brevi- stylis, 315; cruciata, 455; ellip- tica,2,92>', fatua, 419; gigas,226, 318; gigas, appearing thrice, 327 ; gigas, chromosomes, 325 ; gigas-nanella, 375 ; hirsutis- sima, 455 ; laevifolia, 308 ; lae- vifolia, origin of, 266; lepto- carpa, 353 ; Lamarckiancu, length of fruit, 528; Lam., long-fruited race, 545 ; La- marckiana nana, 2)^2 ; Lam. X O. biennis, 299; Lamarckiana X O. nanella, 298; Lamarck- iana, pitcher, 484; Lamarck- iana, premutational period, 496 ; Lam., short-fruited race, 546; lata, 239, 310, 402; lata na- nella, 239, 304; lataXO. bi- ennis, 299 ; lata X O. nanella, 299; muricata, type, 431; na- nella, 235, 360; nanella ellip- tica, 239 ; nanella lata, 374 ; na- nella oblonga, 239, 375 ; na- nella scintillans, 375 ; oblonga, 234, 284, 2>2>7; rubrinervis, 230, 282, z^7 ; scintillans, 243, yjy ; scintillans-nanclla, 375; semi- lata, 358; spathulata, 419; sub- linearis, 399; subovata, 419. Oenotheras, Anatomy of the stem, 335; seeds of, 437. Ogive, 49. Oleaster, 124. Olive, 124. Othonna crassifolia, 145. Oxalis corniculata, 58. Pangenesis, 38, 39; Intracellu- lare, 57, 61. Papaver bracteatum monopeta- lum, 16; dubium, 175; somni- feriim polycephalum, 138. Parsnip, 90, 124. Partial fluctuation, 6; variability, 143- Pastinaca sativa, 124. Pears, 179. Pearson^ 157. Peas, 123. Pedigrees, 503. Pelargonium zonale, 195. Period, susceptible, 519. Periods, mutation, 207. Periodicity, Darwin's belief in, 36. Petalomany, 195. Phaseolus vulgaris, 47. Phyt Optus, 407. Pitchers, 61, 470, 485. Pliny, 176. Ploetz, 50, 131. Plusia gamma, 320. Plus-variations, 51. Index. 581 POHL, 219, 240, 315, 408. Polymery, 481. Polymorphism, 173; by hybridi- zation, 46; systematic, 45. Portunion mocnadis, 563. Pofentilla Tormcntilla, 174. Premutation, 490, 510. Primula acaiilis, 20; veris, 20. Probstei, 108. Propagation, vegetative, 82. Prophylls, 487. Proskowetz, von, 520. Prunus Lauro-Cerasus, 49. Pterophorus, 407. Quartile, 526. Quetelet's law, 47. Races, Instability of, 120. Radish, wild, 90. Ramsay, 335. Ranunculus acris pctalomana, 194 ; arvensis inermis, 196. Ray-florets, 556. Regression, ^z, 88, 120. Reinke, 364. Retrogression, 123. Reversion to mediocrity, 95. RiMPAU, 96. RiSLER, 96. Rivett's bearded wheat, 98. Rogues, 78. Rosa, 495. RozE, E., 321. Rub us, 495. RiJMKER, Kurt von, 91. Running out, 96. Russell, 487. Rye, Schlanstedt, 113. Sacciilina carcini, 563. Saint-Hilaire, 17. Salter, 147. Saplings, 493. Saxifraga crassifolia, 61. Scahiosa atrupiirpurca, 197. Schaafhausen, 36. Schindler, 520. Schubeler, 98. Scott, W. B., 66, 201, Sedum crispum, 182. Seeds, 153; Change of, 126; orig- inal, 126. Seelhorst, vox, 519. Selection, agricultural and horti- cultural, 77; and nutrition, 562; cessation of, 122; empirical, 107; in a minus direction, 140; in agriculture, 79; in the field, 122; limits of, 119; method of, 121; methodical, 108; theory, 28. Semper, 63. , Scnccio Jacohaca, 172. Sensitive period, 161. Serres, Olivier de, 176, Shirreff, 178. Slum latifolium, 364. Six, 266. Social questions, 154. Spach, 437. Species, elementary, 57, 167, 171; incipient, 416; inconstant, zyj; Linnean, 20; sterile, 402. Specific characters, 186. Spencer, 59, i35» I54. 212. Spinacia, 204. Spontaneous changes, 53. Sports, 5. Stahl, 144. Sterility, 417. Strawberries, Gaillox, 32. 192. Struggle for existence, 212. Subspecies, 45, 165, 171. Substitution. 4. I Subvariations. Delpixo. 312. 582 Index. Sugar beets, 50, 99; cane, 148. Survival of the fittest, 59, 212. Svalof, 114. Synanthy, 485. Syncotyly, 476. Syringa vulgaris azurea plena, 184. Tilia parviHora, 470. TOURNEFORT, 1 8. Transitions, 504. Transmutability, 205. Transmutation theory, 16. Treub^ 407. Tricotyly, 474. Triticum turgidum compositum, 125- Typha latifolia, 56. Umhelliferae, 556. Units, 3 ; systematic, 18. Van den Berg^ 184. Variability, 43 ; fluctuating, 49 ; individual, 47; in man, 154; linear, 51; partial, 49, 143; transgressive, 426, 430. Variation, amplitude, 525 ; chance 35 ; continuous, 6 ; correlative, 160; discontinuous, 51; grad- ual, 5; linear, 118; meristic, 51; single, 53; slight, 34; spon- taneous, 165. Variegation of leaves, 408, 480. Variegated plants, 147. Varietates minores, 21. Varieties, 169, 360, 506; are only small species, 171 ; character- istics of, 252; sterile, 195. Verschaffelt, 242, 527. ViLMORIN, 83, 89, 100, 492. Viola arvensis, 24; tricolor, 23. Virescence, 407, 423, 472. Waagen^ 50, 66. Wagner, 206. Wallace, 12, 39. Watson, 124, 437. Weisse, 518. Wettstein, R. von, 91. Wheat, 177; Galland, 96; Pedi- gree, III; Rivett's bearded, 89 ; Smyrna, 125 ; talavera, 109 ; Zeeland, 128. WiER, 185. Yellow seedlings, 481. Zca Mays, 72 ; sterilis, 195 ; tuni- cat a, 60. Second Edition, thoroughly Corrected and Revised, with Portrait. SPECIES AND VARIETIES Their Origin by Mutation By Hugo de Vries. Professor of Botany in the University of Amsterdam. Price, postpaid $5.00 (21s.) net. xxiii+830 pages, 8vo., cloth, gilt top. The belief has prevailed for more than half a century that species are changed into new types very slowly and that thousands of years are necessary for the development of a new type of animal or plant. After twenty years of arduous investigation Professor de Vries has announced that he has found that new species originated suddenly by jumps, or by "mutations," and in conjunction with this dis- covery he offers an explanation of the qualities of living organisms on the basis of the conception of unit-characters. Important modifications are also proposed as to the con- ceptions of species and varieties as well as of variability, inheritance, atavism, selection and descent in general. The announcement of the results in question has excited more interest among naturalists than any publication since the appearance of Darwin's Origin of Species, and marks the beginning of a new epoch in the history of evolution. Professor de Vries was invited to deliver a series of lectures upon the subject at the University of California during the summer of 1904, and these lectures are offered to a public now thoroughly interested in modern ideas of evolution. The contents of the book include a readable and orderly recital of the facts and details which furnish the basis for the mutation-theory of the origin of species. All of the more important phases of heredity and descent come in for a clarifying treatment that renders the volume extremely readable to the amateur as well as to the trained biologist. The more reliable historical data are cited and the results obtained by Professor de Vries in the Botaniwil Garden at Amsterdam during twenty years of observation are de- scribed. Not the least important service rendered by Professor de Vries in the preparation of these lectures consists in the indication of definite specific problems that need investiga- tion, many of which may be profitably taken up by anyone in a small garden. He has rescued the subject of evolution from the thrall of polemics and brought it once more within reach of the great mass of naturalists, any one of whom may reasonably hope to contribute something to its advance- ment by orderly observations. The text of the lectures has been revised and rendered into a form suitable for permanent record by Dr. D. T. Mac- Dougal who has been engaged in researches upon the sub- ject for several years, and who" has furnished substantial proof of the mutation theory of the origin of species by his experimental investigations carried on in the New York Botanical Gardens. A New Book by Prof. De Vries PLANT BREEDING Comments on the Experiments of NILSSON AND BURBANK By Hugo De Vries, Professor of Botany in the University of Amsterdam. A scientific book in simple language. Intensely interest- ing as well as instructive. Of special value to every botanist, horticulturist and farmer. Pp. XV+360. Illustrated with 114 beautiful half-tone plates from nature. Printed on fine paper, in large type. Cloth, gilt top. Price, $1.50 net. Mailed, $1.70. MAN A CREATOR. If man can truly be said to have been created in the image of God, he ought to evince his divinity by imitating the creator in deeds of creation, and this, indeed, has long been recognized as the worthiest occupation of man. The poet, the artist, the inventor, in fact all original thinkers and leaders, produce new forms, new devices and contri- vances, new thoughts and higher ideals. Indeed it seems as if the world were the mere raw material purposely left unfinished so as to enable man to exercise the divinest of his qualities, his creativeness. The imperfections of nature appear from this point of view as if made on purpose so as to offer man the opportunity of accomplishing this ambitious task and building up a human world above the natural. A late Latin proverb characterizes the pride of the inhabitants of the Netherlands in this line: Deus creavit mare sed Batavus litora fecit. "God created the ocean, but the shores have been made by Batavians." The creativeness of man appears to acquire a special re- semblance of God's own work, when it extends to the pro- creation of new species, and this has actually been accom- plished of late by Dr. Nilsson, the Director of the Swedish Agricultural Station at Svalof, and our reputed countryman Luther Burbank, of Santa Rosa, California. However meri- torious these undertakings are, they remain exposed to the criticism of the narrow-minded, so we need not be sur- prised to find that Mr. Burbank was once called to account for arrogance by some ignorant clergyman who for the pur- pose of censuring him in the name of God, invited him one Sunday to his church, gave him a prominent seat in a pew exposed to the view of the congregation and denounced the supercilious ways of men who meddled with the plans of God by attempting to create new species. The incident is referred to by Mr. Harwood in his New Creations in Plant Life, (pp. 20-21) when he speaks of the troubles which Mr. Burbank encountered at the start of his career. He says : "Opposition now came from many quarters. Not only did his friends see the fulfillment of their predictions, — some of them very kindly telling him so, — but people who had heard of some of the strange things he had done and who had not the breadth of vision to see what manner of man this was, pronounced him a charlatan. — a man who was creating all manner of unnatural forms of life, mon- strosities, indeed a distinct foe to the race. A minister in- vited Mr. Burbank to listen to a sermon on his work, and when the guest was in the pew denounced him in bitter fashion as a man who was working in direct opposition to the will of God, in thus creating new forms of life which never should have been created, or if created, only by God himself." The incident is comical enough, but it was not so humor- ous to Mr. Burbank at the time when his only consolation was the hope of proving to the world that his hopes were not the useless dreams of a visionary, but definite ideals the realization of which would raise mankind a step higher in civilization and actualize its divinity in a more complete sense. Burbank's work stands now before the world and needs no further recommendation. He found out by experience, that to be a business man is one thing and to work for an ideal is another. He found that the business part had to be neglected for the sake of accomplishing the great task so near to his heart, and for this purpose Mr. Carnegie has come to his assistance by keeping a scientific station in Santa Rosa and aiding his work in general. Much has been writ- ten on Mr. Burbank, but mostly in a popular way by literary authors. Professor de Vries, however, has done justice to the significance of his labors from the scientific stand- point in his new book on Plant Breeding. p. c. THE OPEN COURT PUBLISHING CO. 378-388 Wabash Ave. Chicago Readers may secure copies of this book in Eastern and Southern states through the Baker & Taylor Co., 33-37 East Seventeenth Street, New York, or direct from the publishers. ntOPERTY LIBRART N. C. State. College I