Glass. Book. COPYRIGHT DEPOSIT ~2^ i ^^^w 9-,)' ', B -:> '^-m- ■■:,:' ■iiJti -\,= J/ -> « — -■ - <■ Plate I. -Mutant Forms of Drosophila Melanogaster {AmvdovhUa) . l.WildTypo. 2. Ebony 3. Yellow. 4. White Eye. 5. Bar Eye. 0. Eosin Miniature Black 7. \ estigial. 8. Buff Eye. 0, Cherry Eye. (From Drawing by E. M. Wallace.) GENETICS IN EELATION TO AGRICULTURE BY ERNEST BROWN BABCOCK PROFESSOR OF GENETICS, UNIVERSITY OF CALIFORNIA AND ROY EL WOOD CLAUSEN ASSISTANT PROFESSOR OF GENETICS, UNIVERSITY OF CALIFORNIA First Edition McGRAW-HILL BOOK COMPANY, Inc. 239 WEST 39TH STREET. NEW YORK LONDON: HILL PUBLISHING CO., Ltd. 6 & 8 BOUVERIE ST., E. C. 1918 o. -? Copyright, 1918, by the McGraw-Hill Book Company, Inc. APR 22 1918 THIS MAPLE FKESS TOKK PA. THOMAS HUNT MORGAN PREFACE Of all the sciences that contribute to the great, tertiary' composite which is known as agriculture none is more important economically than genetics. One may not overlook the fundamental relation borne by the primary sciences, mathematics, physics and chemistry, and by the second- ary sciences, botany, zoology, geology, meteorology and economics, to the production and distribution of raw materials. But we confidently assert that the science which underlies the improvement of plants and animals for agricultural purposes is destined to receive increasing atten- tion is agricultural education and in agricultural practice. Without doubt vast possibilities await realization through the more thorough and systematic development of our living economic resources. Such de- velopment is directly dependent upon the successful utilization of genetic principles in plant and animal breeding. The science of genetics is still very young, but it is firmly established and is developing rapidly. It claims the attention of the producer of today and invites the most serious study of the agriculturist of tomorrow. It lays claim also to the interest of the eugenist, the sociologist and the philanthropist and all students of biology. This text has been prepared in response to a real and widely recognized need. The experience of the authors in teaching the principles of breed- ing to undergraduate students has forced home the conviction that an adequate presentation in a single text of the facts and principles of genetics and their practical applications is a prime necessity. Those familiar with the literature of the subject will appreciate the magnitude of the task and, we trust, will be lenient in criticizing our choice of subject matter. It is impossible to include many things of mutual interest to genetics and agriculture if the work be limited to a single volume. We are keenly aware of many deficiencies and it is our desire to prepare a revised edition of the book in the near future. With this in view the suggestions of others are earnestly solicited. We take this opportunity to express gratitude to all who have rendered assistance, especially to those who have read portions of the manu- script or assisted in proof-reading and to all who donated or loaned photo- graphs or who assisted othei'wise with the illustrations. The onus of the work has been lessened in no small degree by the interest and en- couragement of our colleagues. The Authors. Berkeley, California, Feb. 18, 1918. CONTENTS PART I Fundamentals CHAPTER I The Methods and Scope of Genetics Page Introductory 1 Genetics defined 1 Content of genetics 1 Variation and Heredity defined 3 The problems of genetics 4 The methods of genetics 4 Prerequisites for genetics 10 The application of genetics 11 Genetics in agriculture 12 CHAPTER II Variation Introductory 14 Darwin and variation 14 The universality of variation 14 The variation concept 15 Classification of variations 15 Variation and development 20 Variation and environment 21 CHAPTER III The Statistical Study of Variation Introductory 32 Causes of fluctuations 32 Law of statistical regularity 32 Law of deviations from the average 34 The normal curve and its significance 34 Requirements of biometrical study 36 Biometrical terms 37 Requisites to reliability 37 Grouping variates — frequency table 38 Frequency graphs 39 The mean, standard deviation and coefficient of variability 41 Theory of error 44 ix X CONTENTS Page Multimodal curves 48 Correlation and the correlation coefficient 49 Regression 55 Employment and value of the statistical method 56 CHAPTER IV The Physical Basis op Heredity Introductory 57 Heredity a consequence of the genetic continuity of cells 57 The chromosomes 57 Somatic cell division 59 The production of germ cells GO Synapsis — its significance 63 Independent distribution of chromosomes 63 Number of chromosome combinations 65 Chromosomes and sex in Drosophila 65 Recapitulation of the mechanism of heredity 67 CHAPTER V Independent Mendelian Inheritance Introductory 68 Mendelism essentially statistical 68 Mendel and his discovery 68 The monohybrid 68 Mendelian terminology 71 The chromosome interpretation in a monohybrid 73 Sex-linked inheritance 74 The mathematical adequacy of Mendelism 77 The dihybrid 81 A case in maize 81 The chromosome interpretation 33 Mathematical consideration 83 A case in guinea pigs 85 A case in Drosophila 87 The trihybrid 90 A case in snapdragons 90 Multi-factor hybrids 94 Four-fold factor segregation in mice 94 Methods of dealing with genetic data 95 Validity of segregation ratios 100 CHAPTER VI Linkage Relations in Mendelism Introductory 104 Chromosome numbers and factors 104 CONTENTS xi Page A case of partial linkage in maize 105 Chromosome interpretation of linkage 109 Linkage in Drosophila 110 The factor groups 110 An illustrative case from the first group 113 Linkage in the second and third groups 113 Linear arrangement of factors 115 Application in Drosophila 116 The mode of inheritance in crossing-over 121 Experimental verifiratiou of theory of linear arrangement of factors 121 Interference 122 Linkage phenomena in other plants and animals 125 Mathematical relations in linkage of phenomena 127 CHAPTER VII Thk Nature and Expressiox of Mendelian Factors Introductory 129 Conception of factors as loci in chromosomes 129 Factors the genetic representatives of characters 129 Manifold etfects of factors 133 The variability of factor expressions 134 Duplicate factors 136 Multiple factors 137 CHAPTER VIII Allelomorphic Relationships in Mendelism Introductory 144 Character relationships in Pisum 144 Dominance defined 144 The extent of dominance 144 Intermediate expression in the hybrid 147 Variable character expression in the hybrid 149 Competitive action of factors 150 Mosaic expression of a character 152 The presence and absence hy])othesis 153 Multiple allelomorphism 155 In Drosophila 156 In other species 158 Arguments in favor of the conception of multiple alleloinorijliism 160 CHAPTER IX Types of Factor Interactions Introductory 163 Aleurone color factors in maize 163 Comb characters in fowl lt)4 xii CONTENTS Page A factor system in. stocks 166 Truncate winged Drosophila 168 The factor explanation of reversion 170 Darwin's hybridization experiments with pigeons 171 Factor analysis of plumage color in pigeons 172 CHAPTER X Factor Relations in Quantitative Inheritance Introductory 174 Meaning of quantitative inheritance 174 Tall and dwarf races 174 Bush and cupid sweet peas 175 Other factor differences affecting size 177 The cotton leaf factor 179 Corolla length in tobacco 181 Mathematical requirements of the multiple factor theory of size inheritance . . . 185 Castle's hooded rats 186 Hypothesis of factor variability 192 Arguments for the multiple factor theory . 194 CHAPTER XI Inheritance of Sex and Related Phenomena Introductory 196 The Drosophila or XY type of sex-inheritance 196 Sex-linked inheritance 197 Non-disjunction in Drosophila 198 Secondary non-disjunction 201 Non-disjunction not due to a sex-linked factor 204 Bearing of non-disjunction on chromosome constitution of sex 204 The WZ type of sex-inheritance 205 Analogy between the two types 209 Sex-determination in certain insects 210 Sex-determination in certain plants 212 Secondary sexual characters 214 The question of sex factors 215 Intersexual forms in Lymantria 216 Analogous case in tobacco 218 Conclusion 218 CHAPTER XII Species Hybridization Introductory • 219 Genetic compared with taxonomic differences 219 Species hybrids in Antirrhinum 220 Species hybrids in Cavia 223 CONTENTS xiii Page The forms of species hybrids 227 The vigor of species hybrids 230 SteriUty in species hybrids 234 Partially sterile hybrids of wheat and rj-e 236 Partially sterile hybrids in Nicotiana 238 Species hybridization in ffinothera 244 Conclusions 248 CHAPTER XIII Puke Lines Introductory 250 Discovery of pure lines 250 Conditions necessary for the existence of pure lines 255 Isolation of pure lines from mixed populations 256 The effect of selection within pure lines ... 257 Significance of the pure line principle in breeding 259 CHAPTER XIV . Mutations Introductory 263 Two classes of mutations 263 Chromosome aberrations 263 Factor mutations 263 The nature and causes of factor mutations 266 Factor mutations, both germinal and somatic 268 Vegetative mutations versus somatic segregation 272 " Mutations " in the evening primroses 276 PART II Plant Breeding CHAPTER XV Introduction, Historical Introductory • 287 The beginnings of plant breeding 287 Pioneers in plant breeding 288 More recent progress in plant breeding; classification of methods 291 Mass selection 291 Line selection 293 Hybridization 294 Clonal selection 298 Organization of plant breeding work 299 Seed and plant introduction 299 Collections of plant breeding material 299 Research on plant groups 300 xiv CONTENTS Page Organizations of plant breeders 300 Summary 301 CHAPTER XVI On Varieties in Plants Introductory 302 Extent of variety differentiation in plants 302 The origin of domestic varieties of plants 302 Origin of sweet pea varieties 303 Flower color in sweet peas 304 Form and size in sweet peas 304 Habit in sweet peas 308 Hybridization and selection in the sweet pea 309 Creation of varieties of the rose 310 Origin of varieties in the Boston Fern 312 CHAPTER XVII The Composition of Plant Populations Introductory 317 Reproduction in plants 317 Plants normally self -fertilized 317 Plants normally cross-fertilized 317 Discussion 318 Populations of plants normally self -fertilized 320 Populations as affected by crossing 321 Summary 324 CHAPTER XVIII Selection Introductory 325 Selection methods in maize breeding 325 Inbreeding in maize 325 The ear-to-row method 327 The Ilhnois Station experiments 327 The remnant system 335 Selection methods in close-pollinated plants 336 The plant-to-row method 337 Ineffectiveness of continued selection within pure lines 339 The practical importance of keeping varieties pure 341 CHAPTER XIX Hybridization Introductory 342 CONTENTS XV Page Purpose and plan 342 General method 343 Method of hybridiziuf? maize 344 Method of hybridizing wheat 345 Method of hybridizinjr alfalfa 348 Some of the difficulties attending hybridization 350 Conditions favorable for hybridization 351 Species hybridization 351 Svalof method of creating populations 352 CHAPTER XX Utilization of H-i'BRiDs in Plant Breeding Introductory 353 Purposes of hybridization 353 Increased production in Ft maize hj'brids 353 Crossing inbred strains or biotypes 354 Effect of inbreeding in strains of maize 355 Method of comparing yields 356 Crossing species, subspecies, varieties and strains 358 Superior qualities of first generation hybrids 359 Immediate effect of crossing on size of kernel 360 Centralized seed corn production 361 A method of producing hybrid corn seed 361 Application in other annual crop plants 362 Application in vegetatively propagated plants 364 Rapid growing tinker and ornamental trees 365 CHAPTER XXI Mutations in Plant Breeding Introductory 366 Occurrence of mutations 366 Mutations in crop plants 367 The search for mutations 370 Propagation of mutations 372 CHAPTER XXII Graft-hybrids and Other Chimeras Introductory 374 Definition of graf t-hybritl 374 Winkler's tomato-nightshade graft-hybrids 374 Baur's investigation of a natural chimera 378 Other natural chimeras ' 381 Two categories of variegation 381 The physiological behavior of graft hybrids 382 Modification of one graft-symbiont by the other 383 xvi CONTENTS CHAPTER XXIII Bud Selection Page Introductory 385 Efficacy and practicability of bud selection 385 Bud variation in plants 385 Bud selection in Coleus 386 Bud selection in horticultural practice 391 Performance records as a basis for bud selection 391 Bud mutations in Citrus •. . . . 392 Deciduous tree fruits 393 "Pedigreed" nursery stock 394 Bud selection in the potato 394 Certified seed potatoes 397 Other crops in which bud selection may apply 398 Limitations of bud selection 399 CHAPTER XXIV Breeding Disease Resistant Plants Introductory 400 The causes of plant diseases 400 The nature of disease resistance in plants 401 Disease resistance in natural species 401 Phylloxera-resistant grapes 402 Endothia-resistant chestnuts 405 Blight-resistant pears ^ 407 Breeding disease resistant varieties by hybridization 408 Creating rust-resistant wheats 411 Inheritance of disease resistance in other plants 413 Breeding disease resistant plants by selection 416 CHAPTER XXV Plant Breeding Methods Introductory. , 419 Need of systematic methods 419 Pedigree culture 419 The Svalof system 425 Variety tests — purposes; difficulties involved 427 EstabUshing varietal types 427 Determining best varieties for given locations 428 Strain tests — purpose; difficulties involved 432 Plant-to-row tests 433 Factors that affect experimental results 433 CHAPTER. XXVI General Considerations and Conclusions Introductory 437 CONTENTS xvii Page The relation of science to plant breeding — historical review 437 The future relation of genetics and plant breeding 440 Planning breeding operations in the light of scientific knowledge 440 PART III Animal Breeding CHAPTER XXVII The General Aspects of Animal Breeding Introductory 443 The history of animal breeding 443 The animal breeding industry 445 The art of breeding 447 The problems of animal breeding 447 The service of genetics 448 The service of genetics in education 450 The personal equipment of the animal breeder 450 CHAPTER XXVIII Variation in Domestic Animals Introductoiy 453 The sources of variation 453 Selection as a cause of variation 454 Variation by modifiability 454 Modifiability and breeding value 456 Modifiability and correlation 459 Variation by recombination 459 Mutation in domestic animals 462 CHAPTER XXIX Mendelism in Domestic Animals Introductory 465 Importance of experimental breeding 465 Mendelism in horses 465 Mendelism in sheep 475 Mendelism in swine 476 Mendelism in poultry 476 CHAPTER XXX Acquired Characters in Animal Breeding Introductory 480 The problem 480 The beUef in the inheritance of acquired characters 482 The argument against the inheritance of acquired characters 484 xviii CONTENTS Page The soma and germ plasm — experimental investigations .... 485 The isolation of the germ plasm 488 The inadequacy of affirmative evidence 489 The transmission of functional modifications 491 Parallel induction 492 The adequacy of other factors 493 The conclusion 494 CHAPTER XXXI * The Selection Problem in Animal Breeding Introductory 495 General views of selection 495 The American standard bred horse 495 Fecundity in fowls 497 Bantam fowls 499 Selection and breeding methods 500 Selection indices 502 CHAPTER XXXII Hybridization in Animal Breeding Introductory 508 Growing importance of hybridization 508'' Grading 508 Crossbreeding • 514 Species hybridization among domestic animals 515 CHAPTER XXXIII Disease and Related Phenomena in Animal Breeding Introductory 522 The inheritance of disease 522 The inheritance of predisposition to disease 523 The inheritance of defects 524 Defects in domestic animals 526 Immunity to disease 527 Breeding for immunity 529 CHAPTER XXXIV Sex in Animals Introductory 536 The determination of sex 536 Sex-determination in mammals 536 Sex-determination in birds 539 The sex-ratio 539 CONTENTS xix Page Causes of unusual sex-ratios 542 Metabolic theories of sex-determination 544 Inheritance of unusual sex-ratios 546 Secondary sexual characters 548 The effects of castration 548 CHAPTER XXXV Fertility in Animals Introductory 551 Factors influencing fertility 551 The Darwinian theory of fertility 552 Inbreeding not in itself harmful 553 Fertility as related to Mendelian factors 554 The chromosomes and fertility — Drosophila 555 Sterility in other animals . . 555 Sterility in hj^brids 556 FertiUty as related to heterozygosis 559 Fecundity in fowls 559 Conclusion 563 CHAPTER XXXVI Some Beliefs op Practical Breeders Introductory 564 Telegony 564 Harmful effects of hybridization 571 Infection of the male. Saturation 571 Maternal impression 572 Prepotency 573 The Mendelian interpretation 574 The relative factor interpretation 575 The hereditarj^ complex interpretation 575 Greater prepotency in the male 576 Conclusions 576 CHAPTER XXXVII Methods of Breeding Introductory 577 Methods means to an end 577 Phenotypic selection 577 Limitations of phenotypic selection . 578 Pedigree breeding 580 Breeding systems based on blood relationship 580 Inbreeding 581 Line breeding 582 Out-breeding 583 Other systems of breeding 583 Genotypic selection 584 XX CONTENTS CHAPTER XXXVIII Methods of Conducting Breeding Investigations Page Introductory 591 The need of records 591 Judging the individual 591 Pedigrees " . . 596 The coefficient of inbreeding 598 The coefficient of relationship 600 Marking individuals 602 Recording data 603 Cooperative breeding 606 CHAPTER XXXIX Concluding Remarks Introductory 607 The present lack of detailed knowledge 607 The need of research 608 The service of genetics 609 The need of other knowledge 611 Glossary 615 List of Literature Cited 622 Index 648 GENETICS IN RELATION TO AGRICULTURE PART I-FUNDAMENTALS CHAPTER I THE METHODS AND SCOPE OF GENETICS Soon after Mendel's report of investigations in heredity had been rediscovered, it became evident to most biological investigators that a flood of light had been thrown upon the problem of heredity, and the related subjects of variation, development, and evolution. The need for a new term, therefore, to designate this interrelated portion of bio- logical science led Bateson to coin the word, genetics, from the Greek root, TEN, "become." The derivation does not indicate, it must be admitted, very clearly the portion of biology to which the term genetics applies, but this vagueness has in it an element of desirability, for it is extremely difficult to define accurately the boundaries which delimit the province of genetics. Bateson himself has stated that genetics deals with the physiology of heredity and variation; and a favorite statement of authors has been that genetics is the science of the origin of individuals. But these statements — they can hardly be called definitions — must be qualified carefully in order that they may be understood. Accordingly it has seemed desirable to construct a definition of genetics in purely objective terms. The following definition is, therefore, proposed to ful- fill this need; it, too, will require some qualification: Genetics is the science which seeks to account for the resemblances and the differences which are exhibited among organisms related by descent. The Content of Genetics. — If genetics be defined in the above manner, it may be stated roughly that variation is that portion of genetics having to do with the differences beween organisms, whereas heredity has to do with the resemblances which they exhibit. But this statement does not define very accurately the exact meanings of the two terms; to do this it is necessary to consider certain fundamental facts. Organisms exhibit various degrees of difference and resemblance, and classification is made possible first, by resemblances between individuals and, second, by differences between groups of individuals. Further, the orderly interrelations which are exhibited by living beings in general has 1 2 GENETICS IN RELATION TO AGRICULTURE made it possible to group them into orders, families, genera, and species according to the degree of resemblance which exists among groups of individuals. But this is merely a view en masse of the differences be- tween organisms, for it is universally true that no two individuals are exactly alike. There are, therefore, for all practical purposes, two orders of difference between individuals; first, racial differences, those which separate groups of individuals, and second, individual differences, those which distinguish the individuals of a group from one another. Strictly, of course, there are all possible gradations from the one degree of dif- ference to the other, but conveniently it may be said that the former, the racial differences, are those which characterize different lines of descent, whereas the latter, the individual differences, distinguish indi- viduals within a given line of descent. The problem as to the origin of racial differences is a problem of evolution; the problem of the origin of individual differences is a problem of genetics, and we accordingly shall construct our definition of variation to apply to differences exhibited by individuals related by descent. Now all multicellular organisms which reproduce by sex exhibit the common characteristic of two distinct cycles of cellular development; gametogenesis, or development of the germ cells, and somatogenesis or development of the body. The resemblances which make it possible to group individuals into orders, famiUes, genera, and species are the result of the fundamental relation which exists between these two cycles, for it is a commonplace fact that the germ cells of any species can reproduce individuals of the same and no other species. This rela- tion of germinal constitution to the development of the soma is specific for all classes and grades of characters, but the order of specificity may be either racial or individual, just as the order of difference between individuals is racial or individual. The term variation carries with it the idea of deviation from type, and obviously the above statements, brief as they are, of the cycles in individual development leave room for several possibilities of deviation from type. Thus, if we look at the matter from one point of view, the guiding hand in determining the characters of the individual is the specificity of the germinal substance. But every individual develops under a certain set of conditions, the environment, which is independent of the germinal substance; and these conditions have a certain, usually merely modifying, influence in the development of the individual. There is, therefore, a possibility for differences to arise in individuals independently of differences in the germinal substance, differences which are specifically attributable to diversities in the environment, and which may have no effect on the germinal substance itself, just as the degree of heat, for example, may cause a variation in the end products THE METHODS AND SCOPE OF GENETICS 3 which a given chemical system yields. Differences in development may, also, occur because of actual diversities in the germinal substance, and these may arise from the intermingling of different kinds of germinal substance, such as obviously takes place in sexual reproduction, a cause of variation which has been ably advocated by Weismann and styled by him amphimixis; or they may arise from actual changes in the germinal substance, distinct from the intermingling of germinal elements which already exist, a form of variation which has been proposed and elaborated by de Vries under the name of the mutation theory. Accordingly the term variation in genetics is so defined that it includes differences in individuals related by descent, although many authors do not include within the term those differences which are due to environmental conditions of all categories. The following definition is framed in conformity to that already given for genetics. Variation is difference, whether in the expression of somatic characters or in the elements of germinal substance, among organisms related by descent. Heredity is commonly defined as the tendency of offspring to develop characters like unto those of their parents; according to Castle it is resemblance based upon descent. Thomson presents a very able dis- cussion of the concept, heredity, together with criticism of definitions which have been offered from time to time for the term. According to his definition, by heredity is meant nothing more nor less than organic or genetic relation between successive generations. The universal tendency of organisms to produce similar organisms is the cause of the maintenance of organic groups and group relations. But experimental research has demonstrated that sometimes new com- binations of germinal substance produce characters which have not been exhibited by parents. It is necessary, therefore, to define heredity in such general terms that it will include those exceptional characters which have never been exhibited by any ancestor. Now regardless of any external difference which may be exliibited by an individual, its germinal constitution bears a perfectly definite relation to those of its parents. For that reason the following definition is stated in terms of elements of the germinal substance, rather than in terms of somatic characters. Heredity is germinal resemblance among organisms related by descent. Finally, with respect to the content of genetics, emphasis should be laid upon the importance of a consideration of the various phases of development. In development are included all those changes and cycles through which the individual passes in attaining the adult condition. Obviously there is much in development which cannot be treated at all in an elementary text-book of genetics, for particular cycles or phases of 4 GENETICS IN RELATION TO AGRICULTURE development are treated as separate sub-divisions of biology, such as embryology, cytology, experimental morphology, and like subjects. While obviously there is much in all of these subjects which is irrele- vant to a treatment of genetics, nevertheless, rightly interpreted, there is little which is essential to any one of them which does not bear some more or less intimate relations to those phenomena which belong more strictly in the province of genetics. The reason for this is very apparent, the development of the individual is a consequence of the elaboration of the hereditary material, it is the fulfillment of the possibilities wrapped up in the germ cell; how then can it fail to possess much that is of very great significance to genetics? Assuredly the further advancement of the science of genetics will focus more and more attention upon the prob- lems of growth and differentiation in the individual ; for that reason these emphatic statements are made. The Problems of Genetics. — Obviously the problems of genetics are those which grow out of a study of resemblances and differences in individuals related by descent. Wilson has reduced the statement of the problems of inheritance and development to that oft-quoted question : "How do the adult characteristics lie latent in the egg; and how do they become patent as development proceeds?" Pearl has voiced very much the same thought in his statement that the critical problem of inheritance is the problem of the cause; the material basis; and the maintenance of the somatogenic specificity of germinal substance. There are four general methods of attacking the problems of heredity; namely, the methods of observation, experimental breeding, cytology, and experimental morphology. Each of these methods has its specific advantage and particular value as well as its definite limitations. In the following discussion each method is considered briefly with respect to its relation to the development of the science of genetics. The Method of Observation. — The method of observation, or de- scription as it is often called, requires special treatment because it employs the inductive mode of reasoning. Briefly the essential steps involved in the application of inductive reasoning to the problems of genetics may be stated as follows. The first step is the observation of the re- semblances and difl"erences between representative individuals of a given line of descent or, if problems of evolution are under consideration, of different lines of descent. The next step is a comparison of the ob- servations which have been made for the purpose of determining whether they show orderliness with respect to each other; in other words to de- termine whether they probably have a common causal basis. If they do show such orderliness, an attempt is made to formulate the principles or laws which govern them. Finally, the principles or laws thus for- mulated are applied to other instances not included in the original set of THE METHODS AND SCOPE OF (lENETlCS 5 observations in order to test their general validity. The weakness of the method in biology lies in the lack of rigid experimental control over the phenomena which are under observation, and also in the fact that often it is either very difficult or impossible to subject to experimental verification the principles or laws which have been thus formulated. For this reason, the method of observation as a means of formulating prin- ciples and laws must constantly be subjected to rigid scrutiny, lest unde- tected fallacies lead to the acceptance of conclusions which actually have no significance from a biological standpoint. But although the observational method has very definite limitations in the determination of genetic principles, nevertheless it has been the chief method of investigation in the formulation of some of the most stimu- lating theories of biological science. The marshalling of evidence by Darwin in support of the evolution theory depended almost entirely on an application of this mode of research to a vast array of more or less iso- lated cases. The mass of evidence, which he accumulated in order to demonstrate that natural selection by favoring the "survival of the fittest," to use Spencer's phrase, results in evolutionary progress in suc- ceeding generations, .will ever stand as a monument to his masterly skill in observation and interpretation. In addition to its utilization in the development of the evolution theory, the observational method has been employed widely in the field more strictly included in genetics. Sir Francis Galton employed a refined type of the observational method in his study of heredity. His object was to establish a law of organic resemblance within a single species, distinctly a problem of genetics. In order to do this he employed a system of more exact observation based upon accurate determinations in a large number of instances and mathematical reduction of the data thus collected. This system has since undergone notable development, particularly at the hands of Karl Pearson, and, as biometry, it is often accorded recognition as a distinct branch of biology. As one of the re- sults of his studies, Galton announced the law of ancestral inheritance which states that on an average each parent contributes one-quarter or 0.52, each grandparent one-sixteenth or 0.5*, and so on to the total heritage of the individual, which equals 1.0. The other notable result of these studies, the law of filial regression states essentially that on the average any deviation from racial type is transmitted to the offspring in a lessened degree, so that;, in general, offspring differ less from the type of the race than their parents; specifically they exhibit a deviation from the racial mean only two-thirds as great as the parents. Mere observation, be it ever so precise, is subject to very decided limitations when employed as a method of analyzing the general problems of evolution and heredity. To be convinced of this, one need only con- 6 GENETICS IN RELATION TO AGRICULTURE sider the opinions which have been entertained by those who have em- ployed this method in the solution of biological problems. Thus Darwin believed that minute continuous variations are transmitted and form a basis for evolution and that the more striking discontinuous var- iations are of little moment in the origin of species. These are beliefs which rigid experimental investigation has failed to establish, and which are, therefore, highly improbable. In fact it has been clearly demon- strated that minute differences between individuals are for the most part not transmitted, and that distinct new characters which appear suddenly are often heritable. Similarly, the inheritance of acquired characters, so readily accepted by men with minds as keen as those of Darwin and Spencer, has failed to receive confirmation when subjected to rigid experimental enquiry. Definite knowledge on points such as these is of tremendous importance in making for progress toward the solution of the general problems of genetics, but such progress is slow and uncertain by the employment of the observational method of attack alone. It is for this reason that the favor of geneticists has swung so strongly toward a more rigid method of experimentation. However, the observational method is not unique in possessing limita- tions. No single method is known invariably to give correct results. It is necessary to combine all available methods in order to insure the most certain and rapid approximation to the truth. But the difficulty with the observational method, particularly that part of it known as biometry, has been in the manner of its employment in the elucidation of genetic phenomena. It has been employed, as Pearl points out, both as a method of research and as a method of stating the results of experience. The former manner of utilization is unquestionably of great value in genetic research, its particular value residing in the fact that it has substituted exact methods of expression for vague and indefinite statements. It has performed a service of tremendous value to biology in the introduction of the probable error concept as an index of the degree of reliance to be placed in the results of determinations arrived at by other methods. The latter manner of utilization, however, as a method of stating the results of experience, the employment of which is characteristic of the biometrical school, is subject to serious objections. However, it is worthy of note that the method of observation will ever remain a valuable aid to the extension of knowledge, particularly in directions in which, by their very nature, it is impossible to employ experimental methods of research. It is difficult to imagine, for instance, any notable advance in our knowledge of human heredity save by a proper employment of this method of investigation. The Method of Experimental Breeding.— The essential feature of all experimental breeding is the raising of pedigreed cultures of plants and THE METHODS AND SCOPE OF GENETICS 7 animals, for which reason it is sometimes called the pedigree method. The notable progress which has been made in genetics during the past few decades has come from the application of this mode of enquiry. It is the analytic method of the geneticist and it is often and not unjustly compared, both with respect to its utility and its limitations, to the test- tube method of analytical chemistry. From it have come many stimu- lating ideas of heredity and variation; the Mendelian theory of heredity; the closely related pure line theory of Johannsen; and the mutation theory of de Vries: few methods of research can boast a more honorable array of achievements. Of these achievements, the Mendelian theory is the accepted founda- tion of present ideas of heredity. For the application of Mendelian methods of analysis three essential breeding operations are necessary ; first,the raising of pedigreed strains of plants and animals to determine their behavior under controlled conditions; second, the hybridization of diverse races; and third, the intensive study of the hybrid progeny through successive generations. From this outline of the breeding methods which are employed, it may be concluded rightly that the Mendelian method, like the Galtonian, is essentially statistical. It differs radically, however, from the Galtonian method in that it substitutes the observation of con- trolled progenies for that of ancestral generations. Its particular ad- vantage lies in the fact that it is strictly verifiable. Moreover, it has had a different and more specific purpose in view, namely to state in definite terms how the particular individual will behave in heredity, rather than to arrive at a determination of average behavior in this respect. The important result of this method of analysis has been to demonstrate that the germinal material is made up of definite units or factors which stand in close relationship to particular characters of the soma, and to demon- strate how these elements of the germinal substance are transmitted from generation to generation. The two remaining products of the pedigree culture method, the pure line theory and the mutation theory, stand in close relationship to the Mendelian theory of heredity; because they may be interpreted in terms of the elements which constitute the germinal substance. Of these the pure line theory may be said to add another conception to those of the Mendelian theory, namely that elements of the germinal substance possess a high degree of stability. If this conception be accepted, it follows — and this is the central postulate of the pure line theory — that variation among individuals of like germinal constitution is a response to external or internal conditions which are not reflected in the germinal substance. Such variations, therefore, are of no consequence for the establishment of new hereditary characters. A large number of plants, among them barley, oats, rice, wheat, and practically all the legumes, 8 GENETICS IN RELATION TO AGRICULTURE are almost invariably self-fertilized. They consequently give rise auto- matically to populations which are composed entirely of pure lines. The pure line theory, therefore, has tremendous practical significance. The mutation theory adds yet another conception to those which have already been stated, namely that of occasional mutability of germinal elements. It is, therefore, directly contradictory to the pure line theory in its fundamental postulate; but the very great infrequency with which changes occur in germinal elements saves the pure line theory from inutility. Here the important result has been to establish firmly the occurrence of occasional, definite, discontinuous changes in germinal substance in consequence of which new characters are added to the heritage of the race. Much of the variability in individual characters which is exhibited by plants and animals appears to have had its begin- ning in mutational changes in the germinal substance. The mutation theory, therefore, is another consequence of genetic investigations which has far-reaching practical consequences. Fruitful as have been the results of the method of experimental breed- ing in prosecuting genetic research, students and investigators should not delude themselves as to the nature of the knowledge which it has yielded. It cannot stand alone as a mode of investigation, for even the present illuminating conception as to the structure and operation of the hereditary mechanism has been almost as much the result of cytological as of breeding investigation. But taking this conception in its present form, tremendous as has been the advance of recent years, this sort of knowledge cannot represent the ultimate goal of genetic research. Mendelism has given us the plan of heredity^the more intimate and fundamental knowledge of the material which is employed in the elabora- tion of that plan remain the task of some other mode of research. The Method of Cytology. — The method of cytology in genetic re- search is concerned primarily with questions of cell mechanism. It may be said to be directed toward the solution of two distinct problems, first the behavior of the hereditary elements in somatogenesis, the building up of the body, and secondly in the determination of the nature and operation of the mechanism which distributes hereditary elements from parent to offspring. These are matters of fundamental importance in genetic enquiry; it is unfortunate that the methods of dealing with the problems here presented are necessarily static and so little under the control of the investigator. Nevertheless even with these handicaps, the contributions of cytology to genetic interpretation are by no means inconsiderable. The determination of the equivalent distribution of the hereditary elements in the cell divisions of somatogenesis and the prob- able fact that every ultimate cell in the body normally possesses all the hereditary elements of the initial fertilized egg-cell have been established THE METHODS AND SCOPE OF GENETICS 9 as nearly as may be by cytological research. Moreover, the separation of homologous contributions of the parents in the formation of germ cells and the union of two homologous sets of hereditary elements for the production of new individuals represent another phase of the problems which have been solved by cytological research. Although obviously the dangers of misinterpretation in dealing with fixed and stained preparations of cells or sections of cells are very great, a fact which is disclosed by the diverse interpretations which different investigators have given of the same phenomena and structures, never- theless the importance of this field of research should not be under- estimated on that account. There are several reasons for reposing confi- dence in the results of cell investigations, and these come from two sources; from the confirmations of the growing field of what may be called experimental cytology, the observation of cell phenomena directly in living cells, and from the broad general result of cytological research that the mechanism which has been discovered is by nature such an one as might be expected from a yriori consideration of the results of Mendelian investigations. The close correspondence which exists between cell behavior as it is believed to exist from cell investigations and hereditary phenomena as they are known to exist from Mendelian investigations has given renewed confidence to students of heredity in the validity of their interpretive conclusions. The most important progress which has been made within the last decade in genetic science has been that of interpreting Mendelian phe- nomena of inheritance in terms of the behavior of the cell mechanism. Thus far this work has been carried to any degree of completeness in only one species, the common fruitfly, Drosophila ampelophila. In the extensive investigations which have been made with this species, Morgan and his associates have demonstrated how close a correlation exists all along the line between cell behavior and hereditary distribution of characters. Certain characters -are distributed independently of each other, the pairs of chromosomes separate independently of each other in the formation of gametes; certain characters display irregularities in distribution and expression associated with differences in sex, the chromo- some content of the two sexes is demonstrably different; four sets of characters exist the members of which tend to remain together in trans- mission in the combinations in which they occurred originally , the entire chromatin material is contained in four pairs of chromosomes; and finally irregularities in character distribution have been discovered, the chromosome constitution and distribution in such cases are correspond- ingly irregular. These facts the student will be better fitted to appreciate later on; they are given here to show how the results of one method of investigation are supported and strengthened by those of other methods. 10 GENETICS IN RELATION TO AGRICULTURE The Method of Experimental Morphology. — Under the heading morphology, we include those particular phases of development which are designated by the terms, ontogeny and embryology. The method of experimental morphology has for its task the solution of the problem of the development of the individual as it is related to problems of variation and heredity. The aim of this method is to determine how the characters of the adult become patent as development proceeds, the broad question of the origin of complexities within organisms. In the Mendelian method, the formal relations which exist between hereditary elements are dealt with, particularly their relations in dis- tribution and recombination. The characters of the adult organism are for the most part the basis of judgment. In spite of the general truth of this statement, however, Mendelian analysis has in many cases extended into the field of the physiological relations which exist between hereditary elements, not merely with regard to contrasted homologous hereditary determiners, but with regard to the physiological relations existing in development between entire sets of hereditary elements, and at times even between these and definite factors of environment. But for the most part the solution of such problems depends upon thorough experi- mental study of development in individuals of known genetic constitution. This portion of the problem remains almost untouched. If development be thought of as a series of successive physico-chemical reactions, the complexity of the problem may easily be judged. Certain of the simpler features of it, however, have been attacked and the results of these preliminary studies have indicated still other modes of approach, so that we may expect that when geneticists come to appreciate the light which may be thrown upon heredity by the experimental investigation of development, research in this field will be greatly stimulated. Already as Jennings has pointed out the main features of the process of develop- ment are clearly indicated; the hereditary elements of the chromosomes remain the same in each cell, the reactions and functions of any cell depend upon this chromatin system working in conjunction with the cytoplasmic matrix in which it is located. From this fact may be drawn the broad conclusion that differentiation within the individual depends upon cytoplasm differentiation. The difficulty of the question of how and why should not deter investigation. Prerequisites for Genetics. — The foregoing discussion of modes of research in genetics should indicate something as to the nature of the working equipment necessary for a study of the science. Since genetics is a biological science, intelligent study of it presupposes a thorough grounding in general biology such as is given in foundation courses in botany and in zoology. Inasmuch as practically all domesticated plants and animals belong to the higher orders, particular attention should be THE METHODS AND SCOPE OF GENETICS 11 given to the cycles of developments in these organisms, especially those phases which are comprised in development and reproduction. Of particular importance is a general knowledge of physiology, not so much on account of the direct utility which it has in the study of genetics as for the attitude toward life phenomena which it awakens in the student. Genetics, indeed, is essentially a sub-division of physiology in the broader sense. A knowledge of mathematics is a valuable asset because it is often necessary to subject the data of heredity and variation to mathe- matical treatment in order to interpret them properly. For the elemen- tary study of genetics, a knowledge of the methods of dealing with simpler algebraic problems is sufficient; for advanced study a knowledge of the differential and integral calculus is highly advantageous. Finally it may not be out of place to mention the fact that investigation in genetics is not confined to those who employ the English language. A reading knowledge of French and German is practically necessary for those who desire to pursue the subject very far. The Applications of Genetics. — Genetics has both scientific and prac- tical applications. As an example of its scientific applications, the part which it has played in shaping doctrines of evolution instantly comes to mind, for of necessity such doctrines must conform to the fundamental principles of genetics. The science of genetics and that of evolution are by their very natures constantly encroaching each upon the fields of research of the other. Thus experimental investigations of evolution are of vital interest to genetics, because they deal with the mode of origin of hereditary characters. Genetics, also, has its applications in branches of biology other than that of evolution, indeed throughout the entire realm of biology its influence is felt in shaping thought and direct- ing interpretation. There are few other sciences which possess so much of general interest as that of genetics. The practical applications of genetics are found in agriculture and in human affairs. Here genetics involves many things which are extra- biological. Thus in agriculture emphasis is placed upon the employment of the principles of genetics for the amelioration of plants and animals for man's use. Breeding, then, may be defined as the art of improving plants and animals by hybridization and selection. To make effective progress along this line methods of testing given individuals or races, both with respect to fixity of type and comparative value, have been de- vised. The methods of attack are very much the same as those which are employed in the experimental study of heredity and evolution, the primary aim of which is merely to discover underlying principles. Eugen- ics is concerned with the principles of genetics in so far as they may be applied in the improvement of the human race; but it includes much that is sociological, rather than biological. The applications of the prin- 12 GENETICS IN RELATION TO AGRICULTURE ciples of genetics, therefore, are always subject to such modifications as may be determined by practical considerations. Genetics in Agriculture. — Modern agriculturists, for the most part, appreciate fully the importance of producing only the best types of plants and animals; for in spite of the strange anomalies of economic conditions which at times appear to give actually a greater return for smaller total yields, the fact must remain that the larger view of the agriculturist's place in society requires of him as of all its other members the fullest possible returns compatible with economic principles and the require- ments for a permanent agriculture. But although the desirability of high production and quality is very generally recognized, it is a fact that very often this ideal cannot be attained except by the most careful and intelligent efforts. This is more often the case with plants than with animals, for plants are on the whole less independent of environmental conditions and therefore more susceptible to differences in them. Pro- ducers of crops are always in need of varieties which are better adapted to local conditions, but except in rare cases they are not fitted to develop such varieties. Here genetics comes very definitely to the aid of the plant breeder for its principles provide a safe guide for him in attaining his ideal. Already breeders of plants have realized a great saving of time and expense as a result of the application of principles derived from scientific investigations in their work. The animal breeder on the other hand has faced a somewhat different problem. The far greater comparative value of the individual in live- stock operations has led in animal breeding to the establishment of pure breeds of domesticated animals of remarkable excellence. Long applica- tion of the method of trial and error has developed a body of empirical knowledge which has achieved results nothing short of the marvelous. But while the old empirical methods have served their purposes well, nevertheless they cannot from their very nature give complete satisfac- tion. Knowledge is only secure when it rests upon a firm foundation of principle, and however excellent have been the results of empirical breeding from a utilitarian standpoint, they have not led to the discovery of fundamental principles. The principles of genetics provide a consist- ent interpretation of the results of breeding methods. To the novice a knowledge of such principles is an abundant aid in interpreting and organ- izing details of experience; by its help he can progress more safely and more surely in determining the methods of procedure which are es- sential to the fullest success in his breeding operations. The real service of genetics to animal breeding lies in the promotion of clarity of thought, and that is a thing of no little value. Although genetics thus far has contributed but little toward improve- ment of the existing methods of animal breeding, it is not a dream im- THE METHODS AND SCOPE OF GENETICS 13 possible of realization that in the future its contributions in this direction will be of considerable importance. The science of genetics is still in its infancy, it is still in the formative period of its existence. It has not yet been possible with any degree of satisfaction to analyze the heredi- tary constitution of any farm animal, even to the incomplete extent which has been accomplished in some plants and in some of the smaller animals. Obviously we cannot apply even the general principles of genetics intelligently in animal breeding until we are more thoroughly conversant with the facts of character behavior and factor relationship. Such facts can only be determined by means of carefully planned experi- mental investigations. A few investigations have already resulted in important extensions of our knowledge in this respect, others now under way promise to extend this knowledge considerably further. Systematic crossbreeding of cattle and sheep for definite commercial purposes is of proven value. The method of breeding for high winter egg production in fowls has been determined. Investigation of the inheritance of high milk production in cattle is under way. Geneticists are also seeking to analyze the extensive data with respect to certain characters such as color, fecundity, and speed which have been recorded in herd books. Progress in such work with the larger domestic animals is necessarily ex- ceedingly slow, but this should not deter investigators from organizing carefully planned experiments to extend knowledge in this direction. It is only in this way that genetics can take its proper place in practical animal breeding. The progressive agriculturist can well afford to en- courage every proper effort having as its aim the collection of genetic data. CHAPTER II VARIATION Organic differences, their nature and causes, have furnished abundant material for speculative enquiry since time immemorial. The great sig- nificance of the fact of organic individuality was not fully grasped until Lamarck founded his theory of evolution which postulated the progressive, imperceptible change of one species into another. It remained for Darwin to scrutinize all phases of organic life, past and present, wild and domes- ticated, in his search for a guiding principle which should explain the course of evolution. Darwin's hypothesis of natural selection assumes variability without enquiring into its causes, but this does not mean that Darwin was not concerned with the problem of causes. In both his "Origin of Species" and "Variation in Animals and Plants under Domestication" the causes of variability are often referred to and he suggested among others, the kind and amount of food, climatic changes and hybridization. Our respect for the great naturalist's keen percep- tion deepens when we realize that very little has been added as yet to our knowledge of the causes of variation. The Universality of Variation. — Individuality is common to all or- ganisms. No two trees, no two leaves, no two cells in a leaf are identical in every respect. Individuals sometimes appear exactly alike but even identical twins will be found to differ in some features. The shepherd knows his sheep individually and the orchardist his trees. Were there no differences in individuals there would be no changes in species and there could be no improvement of cultivated plants. "Variation is at once the hope and despair of the breeder," the hope because without it no improvement would be possible, the despair because very often, when improvement has been made, variation results in a tendency to fall below the standard previously reached. In the sugar beet, for example, a high percentage of sugar has been maintained by continually testing and selecting the "mother" beets for the next crop of seed. How- ever, this necessity for continual selection does not exist in respect to all important field crops although they are subject to the general law of variation. That this must be so is clear when we realize that many natural species as well as cultivated varieties of plants are really mix- tures of sub-species, varieties, or races and that upon being isolated these distinct forms reproduce their own particular type. This is most easily demonstrated in plants normally self-fertilized^ yet in all naturally 14 VARIATION 15 cross-fertilized plants and in higher animals this same endless diversity among individuals is even more marked. The Variation Concept. — As we have implied in the above remarks the term, variation, may be used in very different senses in referring to different phenomena. Thus variation within a species or variety means that the group in question is heterogeneous. Among individuals varia- tion may consist of differences between members of the same generation or between parents and offspring. Even when thus restricted, however, the term is apt to prove ambiguous. Hence it is necessary to give some thought to the sources, nature and causes of these individual differences in order that we may use clear cut expressions which shall always convey to one another a concept of the same particular sort of organic difference. Classification of Variations. — 1. Heritahility. — Character differences either represent something specific in the germ or they are merely the effect of external stimuli upon the individual soma. In the first case they are inherited, although they will not reappear necessarily in all later generations or in all the progeny. In the second case they will not be inherited. This is a fundamental distinction and may well serve as our primary basis of classification. According to heritahility variations are either germinal or somatic. Under germinal variations we recognize two sub-classes, combinations and mutations. Purely somatic variations will be referred to hereafter as modifications. Modifi.cations are non-heritable differences between the individuals of a race caused by the unequal influence of different environmental factors. Such variations frequently approximate continuity and, when studied statistically, display the normal variability curve, which w;ill be explained in the next chapter. Combinations are heritable differences between the individuals of a race or between the offspring of a pair of parents caused by segregation and recombination of hereditary units. They also frequently display the normal variability curve. Mutations are heritable differences between parents and offspring which do not depend upon segregation and recombination. These three categories, as Baur has shown, are not to be recognized and separated merely according to appearances. The cause of any individual differences can usually be established only by careful breeding experiments; but by this means the separation of the three categories is always possible as the boundaries between them are quite sharp. Modi- fications are somatic effects of environmental differences and should not be confused with germinal changes which, are sometimes induced by natural or artificial means and which result in the production of muta- tions. Within this first category must be included all place-effects in plants and somatic environmental effects in animals. Modifications 16 GENETICS IN RELATION TO AGRICULTURE comprise a large portion of what are commonly spoken of as fluctuations due to environment, hut all cases of fluctuating variation are not modifica- tions inasmuch as variations due to combinations frequently display the normal variability curve also. Modifications .are not heritable. The second category, variation by combination of hereditary units is often confused with modification, as already stated, because of the fact that variations caused by segregation and recombination when studied statis- tically often display the normal variability curve. This is especially apt to be the case in quantitative characters (those of size or weight) and segregation and recombination may be the cause of gradations in color intensity. In autogamous (self-fertilized) organisms hybridization between races is sufficiently rare to be negligible n this connection, i.e., in such species the fluctuating variations are caused by the environment. But in allogamous organisms (those in which two individuals are neces- sary to accomplish sexual reproduction) fluctuating variations may be caused either by the environment, by segregation and recombination of factors, or by both causes acting together. We shall take up the third category, mutations, in a later chapter. For the present it is sufficient to remember that mutations are no doubt the least frequent of the three classes, that easily distinguishable mutations are comparatively rare, but that there may also occur true mutations of such moderate extent, as compared with the population, that their existence would only be detected by breeding tests, since their progeny would exhibit a different range of fluctuation from that of the population. 2. Nature. We may next enquire into the nature of variation as it affects the organism. Upon this basis we may distinguish between four classes: morphological, physiological, psychological and ecological. Morphological variations are differences in size and form (Fig. 1). In general morphological variations have more significance for the biolo- gist than for the agriculturist. However in many products of the farm, size and conformation are of decided importance. Two sub-classes under morphological variations are meristic and homeotic variations. Meristic variations are differences in number of repeated parts such as the petals in a flower, the leaflets in a compound leaf or number of phalanges. Homeotic variations are differences caused by the replace- ment of one part by another, as the production of an antenna in place of an eye in an insect. Physiological variations are differences in quality and performance. Examples of qualitative variations are difference in degree of hardness of bone, flavor of meat, richness of milk, difference in normal color (Fig. 2), resistance to drouth, frost or alkali. Variations in performance constitute the most important group for the producer. Differences in performance are sometimes, though not necessarily, associated with VARIATION 17 certain details of structure. For example, note the prominent milk veins on the udder of Tilly Alcartra as shown in Fig, 231. Psychological variations are differences in mental traits. That mental and nervous conditions have very definite effects upon physical con- FiG. 1. — Morphological variation in number, form and size of leaflets in the blue elderberry, Sainbucus glauca. ditions is well known, but the problem of distinguishing ])etween pur- poseful action and automatic response, between manifestations of reason and manifestations of instinct, is set for the students of animal behavior. While variations in mental characteristics have an important place in eugenics and merit the attention of livestock breeders, yet the inheritance 18 GENETICS IN RELATION TO AGRICULTURE of pyschological characters must be more extensively investigated before the subject can be considered with profit in a fundamental study of genetics. Ecological variations are those differences between individuals that result from their fixed relation to the environment. These differences are especially noticeable in plants and are known as place-effects or place variations. This category includes some of the phenomena of Fig. 2. — Substantive variation due to chlorophyll reduction in certain areas of the leaves of Elasagnus pungens. variation in crop yield and hence is of immediate significance to agricul- ture. Fig. 3 illustrates place-effects in a common weed. 3. According to differences between them there are two general classes of variations: first, the slight differences in every character which are always to beobserved even among individuals of identical heredity; second, unusual, striking differences commonly known as sports. The first class are called normal, indefinite fluctuating or continuous variations and the second, abnormal, definite and discontinuous variations. It should be noted, however, that all discontinuous variations are not necessarily definite or even distinguishable. Continuous variations when examined statistically are found to conform to the law of statistical regularity. VARIATION 19 That is, if measured and plotted the graph will approximate the normal curve of variability (Chapter III). Continuous variations are either heritable (combinations) or non-heritable (modifications) and, as was stated above, the only certain method of determining the class in which a Fig. 3.— Place-effects in common mustard (Brassica campestris) due to soil differences (herbarium specimens). given case may fall is the breeding test. Discontinuous variations are essentially discrete differences whether they be large or small. They are also either heritable or non-heritable and there is no correlation between size and heritability. Thus the extremely large and small 20 GENETICS IN RELATION TO AGRICULTURE mustard plants shown in Fig. 3 considered by themselves are discontinu- ous variations, but they are almost certainly due entirely to environ- mental differences and seed from the small plant if grown under optimum conditions would produce plants of normal size. On the other hand, it is known that many minute differences in organisms are heritable. 4. According to direction variations are classed as orthogenetic and fortuitous. Orthogenetic variations are those differences found in indi- viduals related by descent which form progressive series tending in a definite direction. Many remarkable illustrations are found among paleontological records of the evolution of animals. Occasional examples are found among short-lived or vegetatively propagated species. The remarkable series of variations of the Boston fern described in Chapter XVI is a good example. Fortuitous variations are chance differences occurring in all directions. 5. According to cause variations are either ectogenetic, differences arising from conditions acting upon the organism from without; or autogenetic, differences resulting from strictly internal relations between germ and soma. Variation" and Development. — Somatogenesis, in sexually produced multicellular organisms, includes the entire history of cellular multipli- cation and specialization from the first cleavage of the fertilized (or parthenogenetic) egg to the completion of all adult features. From the standpoint of individual development it includes gametogenesis, for the production of sexual glands and of secondary sexual characters are merely phases of differentiation. Cell growth and cell function depend directly upon the activity of the living substance within the cell. The nature and degree of this activity depends upon two sets of determining causes acting simultaneously. First, there are the specific hereditary determiners or genetic factors, which react with the other elements of the protoplasm and, under favorable circumstances, condition normal development. Second, there are all the conditions external to the cell which stimulate or inhibit protoplasmic activity. These " developmental stimuli " are chem- ical and physical changes wrought by energy from without the organism or caused by its own physiological activities. Chemical stimuli are exerted mainly through the medium of the circulating liquid which surrounds each living cell. Normally this fluid contains the elements essential for maintenance of life as well as various waste products. It may also bear toxic substances that suppress or inhibit the cell functions and in higher animals it contains the secretions of the ductless, sexual and other glands that profoundly affect development. Physical stimuli are exerted chiefly from without and upon the organism as a whole. They include changes in temperature, light and density of medium, the effects of electric and radiant energy, force of gravity, etc. Obviously, so many VARIATION 21 interrelated causes acting simultaneously, ^ach being independently capable of inducing a change in the end product, may cause an infinite number of differences in substance and in degree of development. Variation and Environment. — External stimuli affect the develop- ment of characters in three ways: (1) they modify the development of inherited characters; (2) they actually condition the production of charac- ters whose hereditary determiners ai'e present in the germ-plasm; (3) they may cause germinal variations which result in the appearance of new heritable characters. The following arc illustrations of these effects with reference to particular environmental factors. Fig. 4. — Sedum spectahile. The three shoots (taken from a single plant) were planted in small pots on March 12, 1904, and placed in different greenhouses: /, in blue light; II, in mixed white light; III, in red light. Photographed on Sept. .30, 1914. {After Klebs.) 1. Environment Modifies Development of Inherited Characters. — (a) Light and Function. — Klebs reports the results of growing the Showy Sedum (Sedum spectahile) in white, red and blue light. The diverse effects of the three kinds of light are clearly shown in Fig. 4. Although the visible differences between the three plants were very pronounced the experiment was carried much further. During 1905-06 observations were made on the numbers of stamens in the flowers of plants similarly propagated under white, red and blue light and under various conditions of temperature, moisture, and food. About 20,000 flowers were examined 22 GENETICS IN RELATION TO AGRICULTURE Substance White Red Blue Ash 13.20 n.04 22.29 0.16 5.82 5.33 13.20 15.40 18.02 0.33 3.66 6.15 18.60 Sugar 2.40 Calcium malate Free nitrogen ...... Starch 18.10 0.59 1.20 Crude protein 7.64 and six distinct types were found, according to the variation in number of stamens. These had the following average numbers of stamens: (1) 9.68, (2) 8.45, (3) 6.54, (4) 5.05, (5) 9.47, (6) 7.33. Finally, Klebs subjected similar plants from white, red and blue light to chemical analysis in order to secure further evidence of the physiological effects of light of different wave lengths. Table I shows the composition of the leaves in three plants like those shown in Fig. 4. They were in their respective greenhouses from June 6 to September 7. The percent- ages shown are per 100 g. of Table I.-Chemical Composition of Three . substance. In compar- Plants op Sedum Spedabile Grown in . ,, , •. White, Red and Blue Light. ^^g these percentages it should be remembered that the plant in white light pro- duced 1324 flower buds and the plant in red light 405, while the plant in blue light produced none. This ex- plains the higher percentage of ash, nitrogen and protein in the last. On the other hand, the amounts of starch and sugar found in the plant from white light are decidedly larger than the one from blue light. In short, according to Klebs, in comparison with normal white light, the production of organic substances, such as starch and sugar, is diminished under the influence of blue light as microchemical and macrochemical tests distinctly show. In consequence of this di- minished assimilation of carbon dioxide the rosettes become purely vegetative. In red light the carbon assimilation is greater than in blue light but less than in white. These experiments prove that the transfor- mation of a plant "ripe to flower" into a vegetative one is possible on the one hand by an increase of temperature and of inorganic salts and on the other hand by a decrease of carbon assimilation. (6) Temperature and Pigmentation. — Many experiments in the rearing of moths and butterflies under controlled temperatures prove that degree of pigmentation is profoundly influenced by the temperature at which the pupae are kept. Some species exhibit seasonal dimorphism in the wild state. By taking pupse of the common European form of the swallowtail butterfly, Papilio machaon, and subjecting them to a tempera- ture of 37° to 38°C., Standfuss obtained the characteristic summer form which occurs in Palestine. Again it has been shown by temperature experiments that many variations found among insects in nature are merely aberrations due to temperature effects. Goldschmidt by arti- ficially controlled temperatures has produced a series of forms of the VARIATION 23 diurnal peacock butterfly, Vanessa io, which show the fading out of the "peacock eye" mark (see Fig. 5). (c) Food and Structure. — Woltereck was able to prove that the form Fig. 5. -The diurnal peacock-butterfly (Vanessa io), above, and below, forms produced by subjecting the pupse to unusual temperatures. (After Goldschmidt.) (hence the structure) of the fresh water crustacean, Hyalodaphnia, varies directly with the food supply. These minute animals produce many generations' during a season and the successive generations from the same 28-VI 30-VII 15-IX Fig. 6. — Morphological cycle of head-height and shell-length in Hyalodaphnia. Roman numerals designate months. {After Woltereck, from Goldschmidt.) water exhibit a morphological cycle, the earher and later generations having shorter heads and the generations produced from midsummer to autumn having longer ones. Fig. 6 is a reproduction of Woltereck 's diagram of the morphological cycle in Hyalodaphnia showing variation 24 GENETICS IN RELATION TO AGRICULTURE in head and shell length as found on successive dates from June 3 to January 3. By raising these animals under constant temperature condi- tions and varying the strength of the nutrient solution, Woltereck proved 35 '40 50 55 60 70' 75 m 9 80 85 90 95 Fig. 7. — Schematic curves of head-height in Hyalodaphnia as grown in media of three different food values. {After Woltereck from Goldschmidt.) that the relative size of body parts varied with the food. In Fig. 7 the percentages of head height to\shell length are plotted as abscissas and the numbers of individuals as ordinates. Animals from three strengths ^^1 ^^^1 ^^^^bK$^ '^^^^I ■■ BH ^nk Ifl ■ .' ^'i V f.^ ^1 ^B - \ ^H ji V ^' if'^l ttPyi^^^B K.' -1 fe *tjl > Er> * ''/*'^^B MSpS.'iSDj^^^H 1 ■vfl i 9 Fig. 8. — a, Typical wild pigeon, Scardafella inca; b, the form dialeucos; c, hraziliensis; d, ridgwayi; e, S. inca after three moultings in a moist atmosphere. (After Beebe from Goldschmidt.) of nutrient media were measured, the curves of those from the weaker, the medium and the richer media being shown at Wi, m2 and ms respectively. V ART AT ION 25 (d) Moisture and Plumage Color. — Beebe experiineiited witli the pigeon, Scardafella inca. This species, as found in North and Central America, is very constant in color of plumage, but in the moist tropics the following darker colored forms occur : in Honduras, dialeucos; in Venezuela, ridgwayi; in Brazil, braziliensis; and these differ in the amount of pigment in the feathers. By subjecting birds of the northern type to an especially moist atmosphere, Beebe caused them to be so influenced that with each new moulting, whether natural or artificially induced, they always de- veloped darker feathers. Thus a wild bird having ])igm('nt in 25.9 per cent, of its area, would have after the second moulting under experimental conditions, 38 per cent, and after the third, 41.6 per cent. Thus during the experiment the typical form assumed the appearance of the three other forms and finally developed plumage markings which have never been seen in nature. Fig. 8 shows the type form, inca, the three geographical variants, and the darkest artificially produced form. Fig. 9. — Plants of Scilla, stalled alike hul (lie pot uii tlie riiiht was kept in a ilaik room. {From Ganono-) 2. Environment Conditions Development of Inherited Characters. — (a) Light and Metabolism. — In a general sense light conditions life in all normally green plants. It certainly conditions normal development in such plants. Potatoes sprouted in a dark room develop no chlorophyll in the stems and the rudimentary leaves are abortive. In many bulbous plants, however, the influence of moisture and heat are sufficient to induce leaf growth and even development of the inflorescence, but it is all done at the expense of the food stored up in the bulb as is shown in Fig. 9. 26 GENETICS IN RELATION TO AGRICULTURE (6) Temperature and Flower CoZor.—B aur reports an experiment with a red variety of the Chinese primrose, Primula sinensis rubra. If plants of this variety are raised by the usual method until about one week before time to bloom and then some of the plants are put in a warm room under partial shade (temperature from 30° to 35°C.) and the re- mainder in a cool house (temperature from 15° to 20° C), when they bloom those in the warm temperature have pure white flowers while those in the cool temperature have the normal red color of the variety. Moreover, if plants are brought from the warm into the cool temperature the flowers which develop later on will be normal red in color. Thus it cannot be said that this primula inherits either red or white flowers. What it really inherits is ability to react in certain ways under the influence of temperature. (c) Food and Fertility. — It is well known that the kind of food supplied to the larvae of bees determines whether the females shall be fertile (queens) or infertile (workers), (Fig. 10). The striking differences in Fig. 10. — The three forms of bees: a, drone; b, queen; c, worker. The two latter develop as the result of difference in the food supplied to the larvae. {After Harrison.) structure and instincts of the two classes of females are all conditioned by the food provided for the larvae. Each larva inherited the capacity to react in either way according to the stimulus received. (d) Moisture and Structure. — Morgan reports a variety of the pomace fly, Drosophila ampelophila, with abnormal abdomen (Fig. 60); "the normal black bands of the abdomen are broken and irregular or even entirely absent. In flies reared on moist food the abnormality is extreme ; but even in the same culture the flies that continue to hatch become less and less abnormal as the culture becomes more dry and the food scarce, until finally the flies that emerge later cannot be told from normal flies. If the culture is kept well fed (and moist) the change does not occur but if the flies are reared on dry food they are normal from the beginning." 3. Environment May Cause New Heritable Characters. — As yet there is a dearth of evidence which can be accepted as scientific proof that external stimuli actually cause germinal variations. At the same time there is an abundance of data which falls into the class of circum- stantial evidence in favor of such a doctrine. Moreover, there are a few VARIATION 27 cases in which new heritable characters have been artificially produced by carefully controlled external stimuli. Hence some germinal variations are apparently caused by known environmental conditions and we are justified in recognizing this third category of developmental differences due to environmental effects. Considerable evidence of permanent changes in both morphological and physiological characters has been secured from experiments with the culture of bacteria and yeast, in unusual culture media, in the presence of toxic solutions, or under extreme temperature conditions. The sig- nificant results of four investigators who worked independently, Hansen, Barber, Wolf and Jordan, have been reviewed and discussed in regard to their bearing on genetic theory by Cole and Wright. The four investi- gators mentioned above used refined methods and three of them began by isolating a single organism from whose progeny they obtained dis- Fio. 11. — 0, Portion of leaf of parental Scrophularia showing branching lateral vein; D, branching vein replaced by two laterals in leaf of a seedling grown from seed produced by an injected ovary. Also note difference in size and margin of leaves. (After Mac Dougal.) tinct strains or biotypes which remained constant for hundreds of test- tube "generations." It must be admitted that in most of these cases no specific influence can be named as the direct cause of the inherited variation. But there is no longer any doubt that permanent, discon- tinuous variations do occur spontaneously in these lowest organisms, and it is highly probable that certain incidental, external forces play an im- portant part in inducing such variations. Direct experimental attack upon the germ cells themselves has been made with plants by a number of investigators, notably by Mac- Dougal, who injected very dilute solutions of potassium iodide, zinc sulphate, sugar, etc., directly into the ovaries of various plants imme- diately before fertilization. Consequently somatic changes have been produced which were inherited throughout several generations. By means of check experiments and observations it was found that these germinal variations were not caused by the wounding of the ovary and it is thought that they must have been induced in some way by the presence of the foreign chemical solution in the ovary. Fig. 11 shows a mor- phological change which appeared in a seedling of an unnamed species 28 GENETICS IN RELATION TO AGRICULTURE of Scrophularia as a result of ovarial injection. Having tested this species sufficiently to determine that it was a simple one, MacDougal treated several ovaries with potassium iodide, one part in 40,000 and se- cured seed. No other species of Scrophularia grew near the cultures. From this seed only three plants were raised. "One formed a shoot fairly equivalent to the normal, finally producing flowers in which the anthocyans were of a noticeably deep hue. The two remaining plant- lets were characterized by a succulent aspect of the leaves and by a lighter and yellow color of the leaves and stems. The flowers on one of the derivatives, as they may be called, were so completely lacking in color as to be a cream-white, this derivative being designated as albida, while the other showed some marginal color and a rusty tinge and was designated as rufida Seeds of the original two derivatives were sowed in the greenhouse. But one plant of albida, the most extreme departure, survived, while four of rufida were secured." MacDougal compared these second generation seedlings with seedlings from the original stock of the species, noting differences in size and margin of leaves, length of petioles and number of marginal glands. He found that the differences shown by the first generation appeared again in the second generation. Striking as these results appear it must be admitted that it would be difficult, on accoufit of the small numbers of individuals differing from the parent type, to prove satisfactorily to the biome- trician that they were not mutations which would have occurred regard- less of the ovarial treatment. What appear to be germinal variations in the tomato have been induced by intensive feeding. T. H. White tested the effect of dried blood, dis- solved phosphate rock, sulphate of potash and iron filings all in excessive amounts, and (with the exception of the iron) in various combinations, on the Red Cherry tomato. The lack of data on control cultures of seedlings from the same parent as the experimental cul tures makes it impossible to compare the actual amount of permanent variation produced. T. H. White states that measurements ''show that the plants of the sixth gen- eration grown under the influence of the dried blood are one-third larger in height, length of leaf and size of fruit, than those of the second"; (see Fig. 12). The author concludes that " there can be no doubt . . . that, in the case of Red Cherry treated with dried blood, there is permanent variation to the third generation." If these results are corroborated by more carefully planned and rigidly controlled experiments they will add the weight of scientific proof of a principle in plant breeding long since recognized on empirical grounds, to wit, that the introduction of wild plants into intensive cultivation induces variation. Furthermore, it suggests a possible means for rapid permanent improvement of wild forms with which hybridization may be impracticable. VARIATION 29 In experiments on lower animals, e.g., the protozoa, the same difficulty is met with as has been encountered in bacteria and yeasts, in that it is manifestly impossible to distinguish between somatic and germinal variations. Moreover, in most of these experiments, as with most of those on higher animals, the necessary conditions for rigid scientific analysis have been lacking. Either the same strain as was subjected to artificial conditions was not grown for comparison under natural condi- tions or else the conditions themselves were not sufficiently well con- FiG. 12. — Leaf and cluster of fruit of Red Cherry tomato of the second generation (right); same of the sixth generation (left) of continuous treatment with excessive amount of dried blood. {Photo by T. H. White.) trolled to permit of certain analysis. It is interesting to note that the pomace fly, Drosojjhila ampelophila, which has produced more mutations so far as we know than any other organism, was subjected to the effects of ether on a grand scale and under controlled conditions by Morgan, but that not a single mutation was observed to result from this treat- ment. However, mutations have subsequently appeared again and again in cultures of " wild " flies not only of this species but also of other species of Drosophila. Thus it appears that germinal variations fre- quently occur independently of external stimuli. It also seems that a tendency to produce mutations may be inherited. 30 GENETICS IN RELATION TO AGRICULTURE With animals the best known experiments on the artificial production of germinal variations are those of Tower who worked with the Colorado potato beetle, Leptinotarsa decemlineata, and related species. Like other arthropods these beetles are more directly under the influence of tempera- ture changes at least than are warm-blooded animals. Tower first de- termined the period in ontogeny when external stimuH will affect the germ cells. He found that in Leptinotarsa the germ cells do not become susceptible to external stimuH until after the time in ontogeny when the color pattern of the individuals subjected to the stimuli can be influenced. He found that eggs were most susceptible just before and during maturation and this observation is in agreement with those of Fischer, Standfuss, Weismann and others who have conducted similar Fig. 13.- -A, Leptinotarsa decemlineata and three mutants; B, tortuosa; C, pallida; D, defectopunctata . (After Tower.) investigations. Tower concluded that certain individuals from the germ cells of a stimulated parent "show intense heritable variations, whereas those not acted upon do not show these changes. " Most of the inherited variations involve changes in the pigmentation of the body parts. In certain cases there was an actual change in the color pattern (see Fig. 13). It is to these results that Tower attaches the greatest significance inasmuch as most similar experiments have not succeeded in causing pattern changes. In spite of the elaborateness of Tower's methods con- siderable skepticism exists regarding the validity of his conclusions, and this has not been lessened by the non-appearance of confirmatory data. In a recent paper he reports the production of very striking germinal modifications in L. decemlineata as a result of subjecting a morphologically homogeneous race to an extreme change in environment. However, it is still a question whether the material used may not be heterogeneous as regards the germinal factors that condition certain physiological characters. Stockard's investigations on the effect of alcohol on the progeny of guinea pigs have shown that the germ cells as well as the somatic tissues VARIATION 31 of the alcoholized animals are injured. This case is considered further in Chapter XXX. On the whole it must be admitted that the experimental induction of heritable variations is still largely an unworked field. The complex conditions to be considered and consequent obstacles to be overcome are appreciated by no one more fully than by those who have attempted such investigations. For, as Tower has said: "It is evident that the problem of germinal change is one of difficulty, and involves more of indirect than of direct methods of investigation. There is little reason to expect that present biochemical methods can give a solution, but they may give valuable suggestions for further indirect investigation. It seems not improbable, however, that this problem like so many others in biology, must await the solution of the larger question of what life is before it will be possible to express in exact terms the nature of germinal changes. Our present status, with several methods of production and much knowledge of the behavior of induced germinal changes available, is a basis from which great advances in knowledge and in operation may reasonably be expected." CHAPTER III THE STATISTICAL STUDY OF VARIATION In the present chapter we shall consider the applicatior^ of purely statistical methods in the analysis of biological phenomena especially the phenomena of variation. The treatment given here does not pretend to be exhaustive or rigorous, but it presents the commonly used method of recognized biometricians, from several of whom valuable suggestions have been received. We shall have occasion to refer to the utilization of statistics in the study of heredity by the "biometrical school," but the application of statistical methods in the analysis of specific genetic problems will be deferred until later chapters. Causes of Fluctuations. — Continuous variations, or the slight differ- ences normally found in organisms, are generally referred to as fluctuating variations or fluctuations. It is frequently assumed that "fluctuating variability" is due entirely to differences in environment. But, as was stated in the preceding chapter, either the modifications in development due to environment, or individual differences which are caused by seg- regation and recombination of genetic factors, may display the normal curve of variation when examined statistically. Hence fluctuations may be due to either of two causes and before conclusions may be drawn from the study of frequency distributions and statistical constants, the causes of the variations studied must be clearly differentiated. The only way to accomplish this is to make one set of conditions or the other as uniform as possible. If the object be to examine modifications, only pedigree material should be used and, on the other hand, if variations due to recombinations are to be considered, the environmental conditions must be as uniform as possible or else due account must be taken of exist- ing irregularities. Certain technical requisites to the biometrical method will be mentioned later. This difference in the nature of fluctuating variations according to their cause is of such fundamental importance that it should be clearly understood at the outset. Law of Statistical Regularity. — This fundamental principle, which is also known as the law of probability or law of chance, may be most simply introduced by means of an illustration. Suppose two persons, blindfolded, were each to pick about 500 beans from a bag containing a million beans of any standard variety. The average weights of the beans picked out by the two persons would be almost identical even though the 32 THE STATISTICAL STUDY OF VARIATION 33 individual beans varied considerably in size. Furthermore if one were to obtain the average weight of the whole million, it would not differ, essentially, from the average weights of the smaller groups. The prin- ciple involved here may be stated in various ways. Wold expresses it M=I5.4 // U 13 /+ 13 /(. n 19 19 iO i/ JA Fig. 14. — Frequency distribution of 500 Broad Beans arranged in classes according to width. as follows: "If a number of different events are equally possible as regards constant conditions {that is, if there is no persistent reason why one should occur rather than another), and all are repeatedly given opportunity to occur, they will in the long run occur with equal average frequency .^^ While this is a satisfactory general statement of the law of probability, the same 34 GENETICS IN RELATION TO AGRICULTURE principle has been expressed by King in terms, which fit well the imaginary case under discussion, as follows: ''A moderate^ large number of items chosen at rundown from among a very large group are almost sure, on the average, to have the characteristics of the large group." It must not be inferred that any partial group of individuals no matter how large, will give exactly the same results as would be obtained by the use of the entire mass. But the averages will be close and the probability of in- accuracy due to accidental errdr diminishes as the numbers increase because individual errors tend in the long run to counteract each other. Law of Deviations from the Average. — If, now, one lot of 500 beans be measured to the nearest millimeter and then arranged in columns from left to right according to width beginning with the narrowest beans, the result will be very similar to Fig. 14. It will be noticed first that the middle classes contain the most beans while the classes on the extreme left and right are very small. The black vertical line M indicates the average width or mean of all the beans and the column with the most beans in it represents the most frequent width of beans and is called the mode. The columns nearest the average value on either side contain the most beans and the further the column is from the average the fewer the beans in it. Thus we see that the majority of the beans show only slight deviations from the average while a few exhibit wide deviations therefrom. Statistical study has proved that it is a general rule with fluctuations that individuals showing extreme deviations in either direction for a given character are comparatively rare, while individuals exhibiting smaller deviations, and hence occupying a position inter- mediate between the two extremes are especially frequent. In other words, continuous variations usually appear in frequencies such that, if we represent these frequencies graphically, we obtain a polygon which resembles more or less the normal variability curve. Such a polygon is produced by connecting the ends of the columns in Fig. 14. The Normal Curve and its Significance. — ^The normal variability curve is a theoretical curve which pictures the result of expanding the binomial (a + 6)" when a = b = 1 and n is assumed to be indefinitely great. By the binomial theorem (a + 6)1 =1 + 1 (a + 6)2 = 1 + 2 + 1 (a + 6)3 = 1 + 3 -F 3 + 1 (a + 6)4 =1+4-1-6 + 4 + 1 (a + 6)5 =1+5 + 10 + 10 + 5 + 1 (a + 6)6 =1+6 + 15 + 20 + 15 + 6 + 1 {a + hy =1+7 + 21+35 + 35 + 21+7 + 1 (a + 6)8 =1 + 8 + 28 + 56 + 70 + 56 + 28 + 8 + 1 (a + 6)9 = 1+9 + 36 + 84 + 126 + 126 + 84 + 36 + 9 + 1 (a + 6)10 = 1 + 10 + 45 + 120 + 210 + 252 + 210 + 120 + 45 + 10 + 1. THE STATISTICAL STUDY OF VARIATION 35 From Fig. 15 it is evident that as n becomes larger the straight lines of the polygon more closely approximate the normal curve. The normal curve is perfectly symmetrical because it represents the distribution of an indefinitely large number of items and it assumes all causes to be of equal strength or value. It is assumed that certain biological frequency polygons should simulate this curve for these reasons. It is probable that the environment of any organism is made up of a large number of factors each of which may vary around a mean independ- \ Fig. 15.- -Polygons representing expansion of the binomials (a + b)^ and (a -j- b)^° as compared with the normal curve. ently of the others. Now if a frequency polygon is to be made regarding a character of a population composed of individuals alike in zygotic constitution, such as a field of potatoes of the same variety, the differences found in the development of any character are due wholly to these en- vironmental factors. Hence it is likely that the mean of the distribution is made up of observations on individuals upon which an equal number of favorable and unfavorable forces have acted and the deviates are those upon which a greater or less number of favorable or unfavorable forces have acted. But in sexually reproduced allogamous species the in- dividuals are not alike in zygotic constitution. Moreover, the causes affecting a given character may have an unequal mass effect according to ecological conditions. Either of these factors may cause a high degree of asymmetry in a polygon of variation. Graphs in which the mode is rather far removed from the mean are called skew polygons or curves. 36 GENETICS IN RELATION TO AGRICULTURE The significance of the normal curve as an index of variation is based on the conception that the area within the curve represents an indefinite number of individuals and that the constants of the curve indicate the distribution of these individuals with respect to a given character. If in any curve (Fig. 16) the perpendicular erected at M divides the area of the curve into two equal parts, this line is the median and the point M represents the average or mean of all the values from which the curve is constructed. The perfect symmetry of the normal curve causes the median to coincide with the mean and the mode; but in actual cases Fig. 16. — A normal curve divided into quartiles by the perpendiculars erected at M,Qi,Qz- Fig. 17. — A normal curve of exactly the same area as the curve in Fig. 16, but with flat- ter slope and correspondingly greater breadth. The distribution pictured by this curve pre- sents a greater range of variation than in Fig. 16 as is indicated also by the value of Q. 'these three values will not coincide because the curve will not be sym- metrical. If a perpendicular be erected in either half of the curve at such a distance from M that it divides the area enclosed by the median, the base and half of the curve into two equal parts, the distance of such a perpendicular, Qi or Q^ from M is the quartile, q. Then in the normal curve q = MQx = MQ^. Now the slope of the curve is an index of the amount of variability. The steeper the slope supposing the area (the number of individuals) to remain the same, the nearer to the median will be the position of the quartile and hence the position of the quartile is also an index of variability (Fig. 17). Since curves constructed from actual distributions are never symmetrical, in practice the index taken is — n However, the measure of variation in common use is the standard deviation, ^)4 2 heads and 2 tails = 6(K)* 3 heads and 1 tail = 4(3.^)* 4 heads and tail = liH)*. THE STATISTICAL STUDY OF VARIATION 45 The coefficients that appear are what they are because precisely those combinations are possible. There is but one combination in which there are no heads, there are four combinations consisting of 1 head and 3 tails, there are six combinations possible of 2 heads and 2 tails, there are four combinations of 3 heads and 1 tail, and again but 1 with no tails. But this is simply the expansion of the binomial (1 + !)■*. The prob- ability that when n coins are tossed exactly m of them will be heads and the rest tails, therefore, is given by the m + 1st term of the binomial expansion (1 + 1)". When n is small a symmetrical frequency polygon is obtained somewhat similar to that given by plotting the yields of individual oat plants. When n is very large more points are obtained Fig. 19. — A normal curve or curve of error showing the relationship between the quar- tile, i.e., the probable error of a single variate, and the standard deviation. Q = .6745 X M. 52 GENETICS IN RELATION TO AGRICULTURE change as the other character changes. The general features of such a table are shown in Fig. 24. The intersection of the two means Mx and My, divides the table into quadrants, which are numbered 1, 2, 3, and 4. The signs of the deviations from the mean of x and y are opposite in the 1st and 3d, while they are the same in the 2d and 4th quadrants. Now the deviation from M of every individual in the table is Vx — Mx in terms of x and Vy — My in terms of y. As these deviations are to be considered relatively, their products are taken. The products of unlike signs are negative, 1st and 3d, and of like signs, positive, 2d and 4th. After arranging the x and y individuals in arrays, if the larger number fall in the 1st and 3d quadrants, we learn that there is negative correlation or a tendency for one character to diminish as the other / \ 1 \ 1 1 1 1 1 Fig. 25. -Interpretation of the correlation table. Shape of and amount of correlation. 'swarm" indicates nature increases. If the majority fall in the 2d and 4th quadrants, we conclude that there is positive correlation or a tendency for one character to in- crease as the other increases. If the individuals are uniformly distributed in the four quadrants we find no evidence of interdependence i.e., zero correlation. These typical distributions are illustrated by the three diagrams in Fig. 25. Comparing the two correlation tables (Figs. 22 and 23) with these diagrams it is evident that the correlation between yield of plant and number of culms is definitely positive, while the nature of correlation (whether positive or negative) between average height of plant and number of culms cannot be inferred from mere observation of the table but that it is very low indeed is clear from the widely scattered distribution. The Coefficient of Correlation. — The interpretation of a correlation table is based upon the fact that the table shows deviations with respect to two characters for each individual or class of individuals. We must remember that the x and y deviations of each class from the mean are multiplied in order to understand how the distribution in the table can indicate plus, minus, or zero correlation between the characters. The product of the two deviations for any individual or class is its product- THE STATISTICAL STUDY OF VARIATION 53 moment, and the summation of all the product-moments divided by 7i is the average product-moment. This measure of absolute correlation is expressed by the formula Av. prod. -mom. = Mdxdy) No. of culms per plant > x G. = i 2 3 4 5 6 7 /„ f-'l'u f.d'-y :i(d'xrf'v) 2- 3- 8-9 /x f.d'. SA'"- 28 18 19 66 20 42 50 -2 -100 200 134 -134 134 58 59 26 14 38 40 50 106 109 80 42 -2 ■100 •106 80 84 21 27 200 106 80 164 63 32 25 400 Wy = -17 400 = -.0425 wK= .0018 -^173 400 = -.4325=i/'x 1871 421 = 1.0525 400 j^871 .8654 , where the two pairs of double chromosomes, one larger and one smaller, . are diagrammed and the nucleolus, the large black body of the previous steps, is shown cast out and degenerating. The daughter chromosomes of each pair now separate from each other until at E they have moved nearly to the opposite poles of the spindle and are beginning to fray, out and seemingly to lose their identity. At this stage actual division of the cell body has begun. Finally at F, the chromosomes have completely lost all appearance of their identity, the chromatin material is distributed throughout the nucleus as in the origi- nal cell shown at A, and the nucleolus has been reformed in each nucleus. Division of the cell-body has resulted in two daughter cells each of which, so far as chromomeres are concerned, contains exactly the same chromatin elements as the original cell. There are many variations in this process particularly in the order of occurrence of the steps, but these variations in nowise modify the essen- tial fact of mitosis which is that the chromatin material of the cell is converted into a thread which splits throughout its entire length into two halves so that the daughter nuclei receive exactly equivalent portions of chromatin material. This precise division of the chromatin is brought about by a division of each chromomere so that not only do the daughter nuclei receive equivalent portions of chromatin but these portions are also equivalent qualitatively to the entire chromatin content of the mother cell. By this method then each of the cells of the body finally comes to possess not only the whole number of chromosomes contrib- uted by the two parents, but also the entire set of chromatin elements which it received from them. The extreme care with which the cell mechanism partitions the chromatin material in each successive cell division is in itself eloquent testimony of the fundamental importance of this material. The Production of Germ Cells. — In the production of germ cells a different set of phenomena occur which result in a reduction of this num- ber of chromosomes to one-half that characteristic of the somatic cells. THE PHYSICAL BASIS OF MEN DELI SM 61 Preceding the actual reduction division the chromatin material passes through a complex series of steps which may be included under the term synapsis. (This term is sometimes applied in a specific sense to the pairing of homologous chromosomes and sometimes to the contraction of the chromatin threads in the conjugation stage.) The essential steps in the prereduction process are shown in outline in Fig. 29. At A is diagrammed a "resting" nucleus at the completion of the multiplication divisions in the germ plasm. As a result of the exact type of mitosis which has been outlined above it contains the full number of chromosomes characteristic of the species. The chromatin of the nucleus next becomes Fig. 28. — Diagram of mitosis in a species having four chromosomes in its cells. A, The "resting" cell. B, Formation of the spireme-thread. C, Longitudinal division of the spireme-thread and transverse segmentation into four chromosomes. D, Separation of the daughter chromosomes formed by longitudional splitting of spireme-thread. E, Beginnings of nuclear reconstruction and division of the cell body. F, Cell division complete and daughter nuclei in the "resting" stage. organized into threads of chromomeres which pair as shown at 5. In this diagram the paired threads are taken to represent homologous chromo- somes, and the opposite chromomeres in a pair of threads are considered as the homologous chromomeres of the two chromosomes. The paired threads contract and fuse along their entire length giving the figure diagrammed at C in which the two loops represent two pairs of homolo- gous chromosomes in the conjugation stage, the essential step in synap- sis. Following this stage the two contracted loops of chromatin split lengthwise and unravel in somewhat the manner shown in D. These filaments contract again forming the intertwined pairs of chromosomes shown at E, and the nuclear membrane thereupon begins to disappear. Further contraction and the formation of a spindle results in the reduc- 62 GENETICS IN RELATION TO AGRICULTURE tion figure shown at F, the significant feature of which is the fact that each of the daughter nuclei resulting from this division receives only two chromosomes instead of the four which the original cell at A contained. Since the original cell contained one pair of larger and one pair of smaller chromosomes, the daughter cells which are formed each receive one larger and one smaller chromosome. Cytological investigation is not yet in agreement as to the interpre- tation of synapsis especially as to the manner in which the phenomena therein concerned are connected with preceding mitotic divisions. Con- sidering certain cytological investigations and the results of research in Fig. 29. — The reduction division as represented for a species whose diploid number is four. A, "Resting" nucleus of a primary germ cell. B, Formaton of paired threads of chromomeres. C, Conjugation of homologous chromosomes (synapsis). D, Loosening of the synaptic knot. E, Condensation of the chromosomes and disappearance of the nuclear membrane. F, Homologous chromosomes about to pass to opposite poles, thus giving each secondary germ cell a member of each pair and one-half the somatic number. heredity together, it appears that the threads which^'pair in stage B rep- resent pairs of chromosomes with homologous chromomeres occupying corresponding positions along their entire length. Likewise the contrac- tion stage at C is taken to represent a conjugation of the members of pairs of chromosomes which later again separate. Other cytological evidence indicates that in some forms the conjugation of pairs of homologous chro- mosomes is brought about in another way. However, the essential fact is the same in either case. In the reduction figure the members of each pair of chromosomes are distributed to the opposite poles of the spindle so that the daughter nuclei received only one member of each pair. The significance of synapsis lies in the conjugation of homologous THE PHYSICAL BASIS OF MEN DELI SM 63 chromosomes. In the mitoses which have preceded this particular divi- sion, the chromosomes were each time conceived to be reformed from the identical group of chromomeres which they contained originally. In synapsis, however, as shown at B there is a certain amount of intertwin- ing of the paired threads and in the unraveling of the chromosomes after the contraction stage there is likewise a twisting of the filaments about each other. The indications are, therefore, that in synapsis there is a possibility of interchange of chromatin material between the members of a pair of homologous chromosomes. In all cases, however, in order to uphold our conception of the definite organization of the chromosomes with respect to the chromomeres which they contain, this interchange of material must involve exactly equivalent portions of the two chromo- mil" Fig. 30. — Diagram of chromatin interchange between homologous members of a pair of chromosomes. {After Muller.) somes. The chromosomes of the reduction division shown at F may not, therefore, be identical with the four originally present in A, but may represent various combinations of portions of both members of a par- ticular pair of chromosomes. The results of such interchange between members of homologous pairs of chromosomes is shown in Fig. 30. At the left is shown a pair of chromosomes one in outline the other in full black. In the middle the steps in chromatin interchange are diagrammed and finally at the right this interchange results in a pair of chromosomes each of which is made up of parts of both members of the original pair of chromosomes. Various combinations may result depending on the points at which interchange takes place, but in every case the exchange involves corresponding portions of the two chromosomes. Independent Distribution of Chromosomes. — In Fig. 31 are illus- trated diagrammatically the chromosomes of Drosophila, with particular reference to their size and form relations and to their characteristic pairing in the cell. One member of each of these pairs of chromosomes was contributed by the female parent and one member by the male parent. In the reduction divisions these chromosomes are separated so that each germ cell contains one member of each pair of chromosomes. The simplest condition which could obtain is that of independent distribu- 64 GENETICS IN RELATION TO AGRICULTURE tion in each pair of chromosomes such that the particular member of one pair which went to a given pole of the reduction spindle would have no influence on the distribution of the members of any other pair. Such independent distribution of chromosomes appears to be actually the type m m m m Fig. 31. — Diagram showing consequences of independent segregation of chromosomes in Drosophila ampelophila. followed in reduction. As a consequence the germ cells contain various combinations of chromosomes with respect to their original parental deri- vation. In Fig. 31 the types of combinations of maternal and paternal chromosomes and their mode of derivation in Drosophila are shown diagrammatically. Two germ cells, one from the female with the chro- mosomes in outline, and the other from the male with the chromosomes in full black, unite to form the female zygote shown in the middle of the figure. The combinations of maternal and paternal chromosomes which THE PHYSICAL BASIS OF MENDELISM 65 result in the production of germ cells in such an individual are shown diagrammatically in the lower portion of the figure. There are eight different ways in which the chromosomes may be grouped in the reduc- tion figures and on the basis of chance any one of these types is as likely to occur as any other. As a result there are sixteen possible combina- tions of chromosomes in the germ cells with respect to the original derivation of the chromosomes, whether from the female or from the male parent. This of course represents only the total number of pos- sible combinations of entire chromosomes. By exchange of chromatin material between homologous chromosomes resulting in the formation of combination- chromosomes the number of actual combinations is greatly increased. The number of chromosome combinations resulting from independent distribution is that number possible when each pair of chromosomes is considered separately, and every combination has an equal chance of occurrence. With a form having but two pairs of chromosomes there would be only four possible combinations, three pairs would give eight, four pairs sixteen, and in general the number of possible combinations is given by the expression 4" in which n is the number of pairs of chro- mosomes in the individual in question. In tobacco which has 24 pairs of chromosomes the number of possible combinations in the germ cells reaches the enormous total of 16,772,216. This means that in the for- mation of zygotes in a self-fertilized tobacco plant the actual parental combinations, i.e., combinations identical with those of the germ cells which united to form the individual in question, occur only twice in over sixteen million times, and this proportion is still further lessened when the interchange of chromatin material between homologous chromosomes is taken into account. The condition of independent distribution although simple in itself results in a rapid increase in complexity with the increase in the number of pairs of chromosomes involved. Chromosomes and Sex in Drosophila. — The relation between inherit- ance and the chromosome mechanism is perhaps most simply displayed in the inheritance of sex in those animal forms in which the sexes occur in approximately equal proportions. Thus in Drosophila as indicated in Fig. 32 there are three pairs of autosomes which are alike in both the male and the female. The remaining pair of chromosomes, however, differ, for the female possesses two X-chromosomes whereas in the male a single X-chromosome is paired with a F-chromosome and these differ- ences are characteristic of all normal males and females of this species. The bearing of these differences on the inheritance of sex is shown diagram- matically in Fig. 32, Beginning with the parents, the diploid number is shown in the circles representing the female and the male. In the female the three pairs of autosomes are outlined and the X-chro- 66 GENETICS IN RELATION TO AGRICULTURE mosomes only are drawn in black to indicate that they are the ones pri- marily concerned in the determination of sex. Similarly in the male the three pairs of autosomes which are exactly like those in the female are outlined but the X-chromosome and the F-chromosome are drawn in Fig. 32. -Diagram to show chromosome relations in the inheritance of sex in Drosophila ampelophila. black. The reduction division in the female results in a separation of the members of each pair of chromosomes, so that every secondary germ cell (or egg) contains two large curved autosomes, a small autosome, and an X-chromosome. Consequently as far as chromosome content goes the eggs are all exactly alike. In the male, however, the separation of THE PHYSICAL BASIS OF MEN DELI SM 67 the members of the chromosome pairs results in sperms half of which contain an X-chromosome and half a F-chromosome in addition to the three autosomes. The reduction division in the male insures an equality in numbers for the two kinds of sperm cells and the chances that either kind of sperm will fertilize an egg-cell are equal. By this arrangement the numerical equality of the sexes is maintained. When, later, the egg cells of the female are fertilized by the sperm cells of the male, as shown in the lower portion of the figure, half of them being fertilized by sperm cells which contain an X-chromosome will give females, and half uniting with sperm cells which contain F-chromosomes will produce males. The inheritance of sex in Drosophila provides a beautiful illustration of the parallel behavior of the chromosome mechanism and a somatic differ- ence, in this case sex. To recapitulate, the essential phenomena of cell behavior which fur- nish the mechanism for the distribution of hereditary factors are these. 1. Every species is characterized by a definite number of chromosomes, each of which is made up of a definitely organized group of chromomeres. The chromosomes occur in pairs, in each of which one member is derived from each parent. In ordinary somatic mitosis the distribution of chro- matin is such that each daughter cell receives a full complement of chro- mosomes which are equivalent qualitatively to those of the mother cell. 2. In germ cell formation the homologous chromosomes conjugate during synapsis, then separate, and pass into a division figure in which entire homologous chromosomes are opposed to each other. The re- sulting reduction division gives daughter cells with half the number of chromosomes characteristic of the species, the half number being made up of one member of each pair of chromosomes. During synapsis there is an opportunity for the members of a pair of chromosomes to ex- change chromatin material. When such interchange takes place equiva- lent portions of chromosomes both qualitatively and quantitatively are involved. In the reduction division segregation within one pair of chro- mosomes is entirely independent of that of any other pair so that the combinations of parental chromosomes in the germ cells represent all those to be expected on the basis of chance distribution. The student should constantly endeavor to harmonize this conception of the distributing mechanism of the chromatin material with the Men- delian interpretations of hereditary phenomena which will be presented in what follows, to the end that he may obtain a clear and definite idea of the interrelations between the known facts of heredity and cell behavior. CHAPTER V INDEPENDENT MENDELIAN INHERITANCE Essentially Mendelism is an attempt to explain the result of heredity on a rigid, statistical basis. Morgan has stated that the cardinal feature of Mendelism is the fact that when the hybrid forms germ cells the factors segregate from each other without having been contaminated one by the other. The presence or absence of any contamination of factors is still a debatable subject as will be apparent from later discussions, but for all practical purposes the absence of such contamination may be re- garded as an established fact. The other implications of this statement that the two germ-cells which unite to produce the individual each con- tribute an homologous set of hereditary units or factors which determine the characters of the individual and that these units again separate from each other in germ-cell formation are the fundamental conceptions of Mendelism. When the units are considered pair by pair one member of each of which has been derived from each parent, it is clear that the im- portant feature of Mendel's discovery lies in the segregation of the members of each pair in germ-cell formation. The statistical laws of segregation of characters were first announced by Johann Gregor Mendel, Augustinian monk and later Pralat of the Konigskloster at Briinn, Austria. In 1865 after 8 years of thorough and painstaking research which is even today a model of genetic inves- tigation, he read the results of his investigations before a meeting of a local scientific society, the Natural History Society of Briinn, and the following year the paper was published in the transactions of this society. Unfortunately, however, the announcement of the work was made at a time when the scientific world was not in a position to appreciate its full significance and was busy with other things. The results, therefore, were neglected until in 1900, the independent investigations by the three botanists, Correns, von Tschermak, and de Vries, led to similar conclusions and to the rediscovery of Mendel's paper. By that time experimental research had so far advanced that the importance of Mendel's work was immediately recognized and it was not long before a vast series of investigations had been reported in confirmation of it. The Monohybrid. — The operation of Mendelism is best followed by considering an actual experiment. Mendel crossed tall and dwarf peas and obtained hybrid plants, all of which were tall like the tall parent. 68 INDEPENDENT MENDELIAN INHERITANCE 69 When the progeny of these tall hybrid plants were grown three-fourths of the plants were tall, like the original tall variety, and one-fourth were dwarf, like the original dwarf variety. Although like the tall plant in appearance, therefore, the tall hybrid plants which were produced by crossing a tall and a dwarf plant displayed their hybrid nature in the kind of progeny they produced. To distinguish them from the tall parents which produced only tall plants, they are accordingly called tall hybrids. Continuing this experiment, Mendel found that the dwarf segregants of the second generation bred true, they produced only dwarf plants; but of the tall plants one-third only bred true, and the other two-thirds proved to be tall hybrids, for three-fourths of their progeny were tall plants and one-fourth dwarfs. The progeny of the original tall hybrid plants, therefore, when subjected to this analysis was found to consist of 1 tall : 2 tall hybrid : 1 dwarf. The experimental results of the hybridization of tall and dwarf peas may accordingly be diagrammed as in Fig. 33. Tall X • Dwarf 1 Tall hybrids 1 Tall 2 Tall hybrid 1 Dwarf I . ■ . I Tall 1 Tall 2 Tall hybrid 1 Dwarf Dwarf i i Tall Tall 1 Tall : 2 Tall hybrid : 1 Dwarf Dwarf Dwarf Fig. 33. — Results of hybridization of tall and dwarf peas. Mendel studied hybrids involving several different pairs of contrasted characters and found that in every case one member of each pair of characters was expressed unchanged in the hybrids, whereas the other member of the pair became latent and its presence could be detected only by growing the progeny of the hybrid. Those characters which were expressed unchanged in the hybrid Mendel termed dominant, the latent characters he called recessive. In the above experiment, for example, tallness was dominant and dwarfness, recessive. Mendel saw that the dominant character, therefore, in these experiments possessed a double significance, that of parental character in which case a uniform progeny of dominants is produced and that of a hybrid character in which case one-fourth of the offspring display the contrasted recessive character. In the above experiment the parental dominants are the tall parents and the hybrid dominants are the tall hybrids. The condition of dominance for a character, therefore, is determined by the fact that in the hybrid that character is expressed to the exclusion of its contrasted character. Dominance is by no means a universal phenomenon, but in Mendel's 70 GENETICS IN RELATION TO AGRICULTURE T T t t T T t t 11 II 1 T T t t experiments it so happened that one member of each of the seven pairs of characters displayed complete dominance. The explanation for the appearance of the recessive character in the second generation and in subsequent generations rests on the fact that the contrasted characters are represented in the germ cells by units or factors. The factor for tallness may be represented by T and the factor for the contrasted character dwarfness by t. The relations which exist when plants bearing these dif- ferent factors are crossed are shown in Fig. 34. In the tall race of plants the gametes all bear the factor T, so that since any individual of this race arises from the union of two germ cells its genetic constitution with respect to this particular factor is TT. Similarly the genetic con- stitution of any plant of the dwarf race is represented by tt and it pro- duces germ cells each of which bears the factor t. When tall and dwarf plants are crossed, the hybrid receives a factor T from one germ cell and a factor t from the other, so that the tall hybrids which are produced are of the genetic constitution Tt. In the production of germ cells and in the union of these germ cells to produce the individuals of the second generation is seen the opera- tion of Mendelian principles. The contrasted units T and t separate in the germ cells of the offspring so that a particular germ cell receives only one of these factors, either T or t. Half the germ cells consequently bear the factor T and half bear the factor t, and this is true of both pollen grains and ovules. When a tall hybrid plant is self -fertilized, therefore: a T ovule may be fertilized by a T pollen grain producing a TT plant, tall, a T ovule may be fertilized by a t pollen grain producing a Tt plant, tall hybrid, a t ovule may be fertilized by a T pollen grain producing a Tt plant, tall hybrid, a t ovule may be fertilized by a t pollen grain producing a tt plant, dwarf. Ovules. / / T T T t '' TALL T/ VLLHYBR D T t t t V VLL HYBf »» | »! f < l f» | Bc AbC ABc X Abc ABc aBC ABc X aBc ABc abC ABc abc ABc ABC AbC X ABc AbC X \AbC IbC Ahc AhC aBC AbC X aB 4" - 2" Number of kinds of genotypes in F2 Number of kinds of homozygous genotypes Number of kinds of heterozygous geno- types 3" 2" 3" - 2" From this table it is clearly apparent how rapidly Mendelian problems increase in complexity with increases in the number of factor differences. With only five pairs of factors the number of individuals necessary to represent the F2 population is 1024 and in order to be sure to have all classes represented it would be necessaty to grow four or five times as many individuals as this. In such an experiment there would be 243 different genotypes distributed among thirty-two phenotypes. Natu- rally the chances of selecting a homozygous individual would vary ac- cording to the phenotype within which such selection was made, but the average chance of selecting a homozygote would be one in thirty-two, and the chance of selecting such an individual in the class displaying all five dominant characters would be only one in 243. The practical diffi- culties of dealing with large numbers of factor differences are there- fore of considerable importance in planning and carrying out Mendehan experiments. Methods of testing the "goodness of fit" of Mendelian ratios depend upon the application of the mathematical theory of probabilities. It is beyond the province of this book to enter into any exhaustive treat- ment of this subject, the present discussion is intended merely to point out the mathematical requirements which must be fulfilled, if no factors are present which tend to disturb the ratio constantly in a given direc- tion. For most problems of this kind it is sufficiently accurate to con- sider the standard deviation of a Mendelian ratio = ±y/N{K —N where N represents a particular term of a Mendelian ratio and i^repre- INDEPENDENT MENDELIAN INHERITANCE 101 sents the sum of all the terms of such a ratio. This gives for the probable error E,, of a given term .V of a Mendelian ratio the value En= ± 0.6745 V N{K -N) n In this formula n = the total number of individuals classified. The actual application of this formula may be illustrated by the use of data from East and Hayes given in Table XI. The totals in this table give observed frequencies as shown in Table XVII. Table XVII. — Goodness of Fit in a Mendelian Experiment Phenotypes Observed Observed ratio Theoretical ratio E Probability Purple starchy .... Purple sweet White starchy White sweet 1,861 614 548 217 9.190 3.032 2.706 1.072 9 3 3 1 ±0.104 ±0.074 ±0.074 ±0.046 1:3.45 1:1 1:142.26 1:2.57 Totals 3,240 16.000 16 The results are expected to be in agreement with a 9:3:3:1 ratio; therefore these observed results are first reduced to the form of a ratio per 16 by dividing each term by j^e of the total number of individuals, 3240 or by "Yr' ~ 202.5. By this method the observed ratio in Table XVII was calculated. To obtain the probable error for the purple starchy class values are substituted in the above formula as follows : £"9 = ± 0.6745 V 9(16 - 9) _ 3240 = ± 0.104 The observed deviation 0.19 is approximately twice the value of the probable error. For practical purposes a deviation less than three or four times the probable error is not considered significant. A deviation of the above magnitude in comparison to the probable error occurs about once in four times. In Table XVII the values of the probable error have been calculated for all four of the terms of this ratio. One term lies considerably within the probable error and its probability has been put down as 1:1. This is not strictly correct but serves the purposes of these calculations. It will be noted that there is one serious deviation, that of the white starchy class which could occur only once in 142 times. This deviation is not serious enough, however, to lead us to reject the hypothesis of two factor differences for this case, but it may indicate that other dis- turbing forces are in operation in this experiment. 102 GENETICS IN RELATION TO AGRICULTURE A better method of testing goodness of fit has been suggested by Harris. The formula employed is X2= 2 (o - c)' In this formula o = the observed frequency of any class; c, the cal- culated frequency of that class; and 2 indicates that all values of the type ^^ — are added together. When this formula is applied to the case treated above the values obtained are as given in Table XVIII. The value of X^ is 8.14. The number of phenotypic classes is four. To deter- mine the significance of this value it is necessary to refer to Elderton's tables for calculating goodness of fit. The value for P, the probability, for this case derived from such a table is 0.0437. The chances that the deviations shown in this ratio are merely due to random sampling are about one in twenty-three, again confirming our previous statement that some unknown slightly disturbing forces may be operating in this case. The deviation, however, is not enough to establish this certainly, for such a deviation might be expected to occur in about 4 per cent, of cases. Table XVIII. — Goodness op Fit in a Mendelian Experiment Phenotypes Observed Calculated (o - c)2 Purple starchy Purple sweet. . White starchy. White sweet . . . o 1,861 614 548 217 3,240 c 1,822.5 607.5 607.5 202.5 3,240.0 0.81 0.07 5.83 1.43 14 =X2 Mathematically the method suggested by Harris is preferable. It has also the advantage that it gives a measure of the goodness of fit of the ratio as a whole; which particular terms are most seriously at variance ((J _ C)2 may be determined by simple inspection of the values of . For determining the significance of X^, it is necessary to have available Elderton's table for test of goodness of fit. These are given in Pearson's tables for statisticians and biometricians. It must ever be held in mind that forces which tend to disturb Mendelian ratios may not neces- sarily be of significance as bearing upon the essential feature of the analy- sis, namely, that a given number of independent factors are concerned INDEPENDENT MENDELIAN INHERITANCE 103 in a certain experiment. There is always a chance that biological conditions of necessity may disturb a ratio, for after all a ratio is only the end point of a series of phenomena which we pretend to describe step by step. Unless constantly guarded against, such biological con- ditions as differences in viability, variations in phenotypic expression, etc., may result in selective elimination of a certain number of zygotes at some time previous to classification, or in error in the classification of some individuals. CHAPTER VI LINKAGE RELATIONS IN MENDELISM Thus far Mendelian experiments have been considered in which the different pairs of factors segregate independently, and it has been shown that such cases may be explained very simply on the assumption that different pairs of chromosomes carry independent factors. However, there are several different species of plants and animals in which the number of known factor differences exceeds the number of pairs of chromosomes. Since it is reasonable to believe that only a small proportion of the possible number of factorial differences in any species has been analyzed, the conclusion appears justifiable that the number of factors in any species of plant or animal greatly exceeds the number of pairs of chromosomes; in fact our present evidence leads us to believe that the number of hereditary units in any organism must reach into the thousands. If the chromosome view of heredity is valid, therefore, each chromosome must carry a very great number of factors. In the present chapter it is proposed to discuss that class of Mendelian phe- nomena which depend upon factors which tend to remain together during segregation rather than to undergo independent assortment. Assuming that such factors are borne by the same chromosome, it will be shown how the chromosome mechanism provides an adequate physical basis for all the relations exhibited by such factors. Linkage and factor coupling are terms applied to that type of inheritance in which the factors tend to remain together in segregation. Linkage of factors is definitely an exception to one of the principles which Mendel laid down, namely, that of independent character segregation. Nevertheless by common consent the term Mendelism has been extended to include all phenomena of inheritance based on the unit factor hypothesis. For a long time only a few cases of linkage were known, and these were regarded in effect as anomalies. But the advocates of the chromosome theory of heredity have zealously prosecuted the study of linkage because of the many ways in which linkage relations parallel chromosome behavior. Moreover as the number of definitely recognizable factors within a species increases it becomes more and more important to determine the relations which the factors display among themselves. Linkage relations among factors, therefore, are of primary importance, and have been the direct means of giving us a clear and illuminating picture of the consti- 104 LINKAGE RELATIONS IN MEN DELI SM 105 tution of the hereditary material and of the operation of the chromosome mechanism in the distribution of the herecHtar}^ units. Purple Aleurone and Waxy Endosperm in Maize. — A typical example of the relations which obtain for linked factors is given in the experiments which involve purple aleurone color and waxy endosperm in maize. We have shown in Chapter V that aleurone color in maize in certain cases depends on a single factor difference, so that in F2 segregation is in the ratio 3 purple : 1 white. For waxy endosperm, when contrasted with starchy endosperm, Collins has shown that starchiness is dominant and that in Fz segregation is in accordance with the normal monohybrid ratio, 3 starchy : 1 waxy. The factors involved in these two cases are C for aleurone coloration and its recessive allelomorph c for colorless aleurone, and W for starchy endosperm and its recessive allelomorph w for waxy endosperm. Collins found that when purple starchy corn, CCWW, is crossed with white waxy, ccww, Fi is purple starchy, CcWw; but jP2 does not segregate in the expected dihybrid ratio 9 purple starchy: 3 purple waxy : 3 white starchy : 1 white waxy. The data which he actually obtained from six ears are given in Table XIX. The calculated Table XIX. — F2 Segregation of Cross Purple Starchy X White Waxy (After Collins) Ear number Number of Purple Purple White White grains starchy waxy starchy waxy 152 183 112 20 22 29 301 579 372 62 63 82 302 536 343 52 53 88 303 627 409 57 62 99 325 650 434 55 61 100 380 161 104 17 18 22 Observed totals . 2,736 1,774 263 279 420 Calculated .9:3:3:1 1,539 513 513 171 Calculated 22.6 ( ?rossing-over . . . 1,775 276 276 409 ratio based on independent segregation evidently falls far short of agree- ment with the observed results, even though, when each pair of characters is considered separately, the agreement with the monohybrid ratio is very satisfactory. Thus for purple and white the observed totals are 2037:699 giving a ratio of 2.98:1.02, and for starchy and waxy the observed totals are 2053:683 giving a ratio of 3.00:1.00. The latter ratio is so close that it would be perfect if only one kernel were shifted from the starchy to the waxy class. Taking each pair separately, there- 106 GENETICS IN RELATION TO AGRICULTURE fore, the factors evidently segregate in the normal Mendelian fashion, but the excess of purple starchy and white waxy kernels indicates that the factors C and W which came from one parent and c and w which came from the other have been distributed to the same gametes more often than would occur on the basis of independent segregation. The ordinary gametic ratio for independent segregation in a hybrid of the genetic constitution CcWw is lCW:lCw:lcW:lcw. In this particular case, however, the gametes were produced in about the ratio 3ACW:lCw:lcW:3Acw. The factors, therefore, display partial linkage, i.e., the parental com- binations of factors tend to remain together more frequently than they tend to form new combinations. The factor W breaks away from C to form a new combination with c only once in about 4.4 times, instead of once in two times as is the case for independent segregation. Neces- sarily whenever W breaks away from C to form a new combination with c, w forms a new combination with C. This accounts for the symmetrical relations displayed in the gametic ratio. In order to show that the two fac- tors are linked, in this case we represent the genetic constitutions of the parents as {CW){CW), purple starchy, and {cw){cw), white waxy; not CCWW and ccww respectively, which is the form used to indicate inde- pendent relations between the factors. Correspondingly the Fi is {CW){cw), not CcWw, and the series of gametes which it forms is written 3.4(CTF) :l{Cw) : l(cTF) :3.4(cw;). The method of deriving an F2 ratio from such a gametic series is shown in the checkerboard in Fig. 49. Here it is necessary to take into account not only the genetic constitutions of the gametes, but also the coefficients which represent their relative frequency. Summing up the totals for like phenotypes from this checkerboard, we find the F2 grains are distributed in the following ratio : 50.28 with purple aleurone and starchy endosperm 7.8 with purple aleurone and waxy endosperm 7.8 with white aleurone and starchy endosperm 11.56 with white aleurone and waxy endosperm. The calculated results based on this ratio are given in Table XIX. They show very close agreement with numbers actually observed, but in judging the significance of this agreement it must be borne in mind that a gametic ratio was arbitrarily selected which would give the closest possible agreement with the observed results. LINKAGE RELATIONS IN MEN DELI SM 107 When the factors enter the hybrid in (hfferent relations, the segre- gation ratio is different. Thus when purple waxy, {Cw){Cw), is crossed with white starchy, {cW){cW), the Fx is purple starchy as in the previous case. The resemblance, however, is not complete except as to phenotypic 3.4 {CW) 1 {Cw) 1 {cW) 3.4 {cw) 3.4 {CW) 1 {Cw) 1 {cW) 3.4 {cw) 11.56 {CW){CW) purple starchy 3.4 {CW){Cw) purple starchy 3.4 {CW){cW) purple starchy 11.56 {CW){cw) purple starchy 3.4 {Cw){CW) purple starchy 1 {Cw){Cw) purple waxy 1 {Cw){cW) purple starchy 3.4 {Cw){cw) purple waxy 3.4 {cW){CW) purple starchy 1 {cW){Cw) purple starchy 1 {cW){cW) white starchy 3.4 {cW) {cw) white starchy 11.56 {cw){CW) purple starchy 3.4 {cw){Cw) purple waxy 3.4 {cw){cW) white starchy 11.56 {cw) {cw) white waxy Fig. 49. — Fi checkerboard of cross between purple starchy and white waxy maize. 1 {CW) 3.4 {Cw) 3.4 (cTF) 1 {cw) 1 {CW){CW) purple starchy 3.4 {CW){Cw) purple starchy 3.4 {CW){cW) purple starchy 1 {CW){cw) purple starchy 3.4 {Cw){CW) purple starchy 11.56 {Cw){Cw) purple waxy 11.56 {Cw){cW) purple starchy 3.4 {Cw) {cw) purple waxy 3.4 {cW){CW) purple starchy 11.56 {cW){Cw) purple starchy 11.56 {cW){cW) white starchy 3.4 {cW){cw) white starchy 1 {cw){CW) purple starchy 3.4 {cw){Cw) purple waxy 3.4 {cw){cW) white starchy 1 {cw) {cw) white waxy 1 {CW) 3.4 {Cw) 3.4 {cW) 1 {cw) Fig. 50. — F2 checkerboard of cross between purple waxy and white starchy maize. expression, for its genetic constitution is {Civ){cW), instead of (CW){cw) as in the first cross. It produces a series of gametes in the ratio 1{CW) :SA{Cw) :3.4(cF) :l(,cw). In this ratio the numerical proportions of the gametes are reversed. This is due to the fact that here the original factor combinations, C and 108 GENETICS IN RELATION TO AGRICULTURE w, and c and W, although the reverse of those in the former case, tend to remain together in the same ratio. When Fi plants of the genetic constitution, {Cw)(cW), are selfed segregation occurs in F2 as shown in the checkerboard in Fig. 50. When like phenotypes are collected into classes, the following distribution is obtained : 39.72 with purple aleurone and starchj^ endosperm 18.36 with purple aleurone and waxy endosperm 18.36 with white aleurone and starchy endosperm 1.00 with white aleurone and waxy endosperm. This ratio is strikingly different from that obtained for the former cross, although exactly the same characters are involved. Unfortunately data supporting this part of the analysis have not yet been presented in a satisfactory manner, but the results so far as reported do show a positive linkage between the factors. Moreover other cases which we shall discuss in this chapter demonstrate beyond doubt that the relations described above hold rigidly for cases of factor linkage. The different results obtained when factors enter a cross in different combinations are, there- fore, simply due to the fact that the original combinations tend to be preserved in segregation in a definite fixed proportion of gametes. To give a chromosome interpretation of linkage we assume that the factors linked are borne in the same chromosome. Thus the factor for purple aleurone color is one of the chromomeres occupying a definite locus in a particular pair of chromosomes in a purple starchy race of corn and the factor W for starchy endosperm occupies a different locus in these same chromosomes. In Fig. 51 the chromosome behavior in linkage is shown graphically. In the hybrid one member of a pair of chromosomes bears the factors C and W, the other member c and w. During synapsis these chromosomes conjugate, and when the threads representing the two chromosomes separate after conjugation they may in consequence of their twisted condition break at certain points and, reuniting, the free ends of different threads may join together. In a cer- tain percentage of cases this breaking of the filaments may occur between C and W, so that the chromosomes afterward reconstituted will contain the factors C and w, and c and W rather than the original combinations. More frequently the chromosomes will untwist without exchanging chro- matin material or after having exchanged it in such a way as not to dis- turb the original factor combinations. Exchange of chromatin material between homologous chromosomes is called crossing-over. This term is also applied to the formation of new combinations of linked factors, and these new combinations are called cross-overs. In this particular case the end result is that for the factors C and W and their allelomorphs LINKAGE RELATIONS IN MEN DELI SM 109 crossing-over occurs in 22.6 per cent, of cases. Accordingly the gametes are formed in the ratio : 38.7 percent. {CW) : 38.7 per cent, (cw) : 11.3 per cent. (Cw) :11.3 percent. (cTT) 77.4 per cent, non-cross-overs 22.6 per cent, cross-overs. It follows, therefore, that linkage may be interpreted as due to as- sociation of factors within the same chromosomes and that crossing-over or breaking apart of linked factors may be regarded as a consequence of I Fig. 51.— Diagrammatic representation of crossing-over and results. At the left, the two original chromosomes. In the middle, the twisted condition of the chromosomes in synapsis and their subsequent Reparation. At the right, the four types of chromosomes which result and their proportions. Fig. 52. — Diagrammatic representation of crossing-over and its results when the factors enter in the opposite combination from that shown in Fig. 51. chromatin exchange between homologous chromosomes during synapsis. The factors may be thought of as the purely passive objects with which the chromosome mechanism deals, they are Hnked together because they are borne in the same chromosome, they show breaks in linkage in a certain percentage of cases because in synapsis breaks occur between the loci which they occupy in the chromosome such that new combinations of the factors are formed. The chromosome relations are the same even when chromatin interchange results in no new combinations of factors, 110 GENETICS IN RELATION TO AGRICULTURE it is only when there are factor differences between the homologous chro- mosomes that the operation of the mechanism can be detected and some conception gained of its mode of operation. Linkage in Drosophila. — To Morgan and his associates through their investigations with mutations of Drosophila aynpeloplila we owe directly practically our entire conception of the linkage relations dis- played by factors. No other single species has provided such a wealth of data or proved so favorable for genetic investigations. This body of data is still growing very rapidly and is adding new conceptions all the time, but even at this time it is no exaggeration to say that the Zoological Laboratories of Columbia University, like the old garden of the Konigs- kloster at Brtinn, have yielded results which will be accounted among the epochal advances in genetics. Mendel's work showed that the char- acters of the organism were dissociable elements of its makeup which could be recombined and shuffled about in genetic experiments. From this starting point the factor conception of heredity, which assumes that characters of the individual may be referred to the action of definite factors in the hereditary material, was developed by a host of investi- gators. Morgan's work has also furnished an overwhelming body of evidence supporting the factor conception of heredity, but its most im- portant contribution to genetics has been in the establishment of the relations existing between the factors of heredity and the chromosome mechanism of the cell. The Four Groups of Factors in Drosophila. — According to the chro- mosome theory of heredity a factor is located at a particular locus in the chromosome mechanism. Consequently since linkage depends upon factor relations within the same chromosome it follows that the factors should display linkage relations such that they would be thrown into groups corresponding to the number of pairs of chromosomes. In Droso- phila the linkage relations existing among over a hundred factor mu- tations have been studied. The factors fall into four groups correspond- ing to the four pairs of chromosomes in Drosophila, and furthermore the relative sizes of these groups corresponds roughly to the relative sizes of the different pairs of chromosomes. There is a large group of sex- linked factors all of which display the same type of inheritance as white- eye color, which has already been described. This group corresponds to the X-chromosomes. There are two large groups of factors which correspond to the two large pairs of autosomes, and finally there is a small group, consisting as yet of only two factors, which corresponds to the small pair of autosomes. The following list of the groups of factors in Drosophila, although incomplete, gives some idea of the number and kinds of factors which have been studied in this species (Table XX). The type of behavior shown in linkage in Drospohila may be illus- i LINKAGE RELATIONS IN MEN DELIS M 111 "t3 h2 M W C3 « ^ lU o a £> o cu " ^ _::; ^ j3 .g O O O -3-dT3«o**'*'* T m CO OS "J 03 -5 >. >i 'S o o o >> >)>>>> >i >) >M o 'ft ; g a o a o O J3 03 S C fi 03 a) i-< ~ 'S ^ -^ ■'^ :-! Ow a».s ° oj ojo ftP.^ '-'b»b. O O fl CD C dC bO UD O (UTSCoflOOOOfl ^^^^fS^S^^^^^^^^UWW^W^^HH^^ ^ o g i3D,t;o3^;3oo ■S 5 S a) — -<?33'=50pHfe!Ka}!»HH> 03 o O 03 a ^ X -a i»i -^c ic r^ u •►-* it; -G >> >i 'C >) 'iT >> o •►r; o ^ -^ x^ a >>'>^ ° s s &►' ^ i; i. ,'?,'? i; '^ (y d C O ^ a» o o o o o — - — o o •►" >. >i o s h a -^ a; ^ t. 0) a ^ >^ 112 GENETICS IN RELATION TO AGRICULTURE trated by the cross yellow white female by gray red male. The genetic formula of a yellow white female is {ywX){ywX), and of a gray red male {YWX)Y. In these formulae y stands for the factor for yellow body-color and w for white eye-color, and the capital letters for the corre- sponding dominant allelomorphs present in the wild type. When two such flies are bred together the Fi consists of females of the genetic con- stitution {YWX){ywX) and males of the genetic constitution {ywX)Y. The Fi females, therefore, have gray bodies and red eyes and the males have yellow bodies and white eyes. Gray red females of the genetic constitution {YWX){ywX) produce four kinds of eggs in the following proportions : {YWX) 49.45 per cent. (ywX) 49.45 per cent. 5.9 per cent, non-cross-over gametes. (YwX) 0.55 per cent. (yWX) 0.55 per cent. 1.1 per cent, cross-over gametes. When such a female is bred to a yellow white male, genetic constitu- tion {ywX)Y, which produces only two kinds of sperms, (ywX) and Y, the progeny in both sexes obviously will be in the ratio 49.45 gray red: 49.45 yellow white: 0.55 gray white: 0.55 yellow red. Table XXI shows the results which have been secured, mostly from matings of this type. The relations shown when the factors enter in the reverse combinations may be determined by mating a gray white female, genetic constitution {YwX)(YwX) to a yellow red male, genetic constitution (yWX)Y. This gives F^ gray red females of the constitution {YwX)(yWX), and gray white males of the genetic constitution {YwX)Y. In this case the Fi females produce four kinds of eggs in the following proportions: {YwX) 49.45 per cent. : (yWX) 49.45 per cent. 98.9 per cent, non-cross-ever gametes. (.YWX) 0.55 per cent. -.{ywX) 0.55 per cent. 1.1 per cent, cross-over gametes. Table XXI. — Factor Linkage in Gray Red Female Drosophilas of the Type {YWX)(ywX) Number of flies classified Non-cross-overs Cross-overs Percentage Reported by Gray Yellow red white Gray white Yellow red of cross- ing-over Dexter 14,939 1,818 854 3,424 8,093 6.672 93 14 2 7 81 5 10 1.16 Morgan and Cattell Morgan and Cattell Morgan and Bridges .... 1,075 513 1,807 729 334 1,600 0.77 0.82 0.50 Totals 21,035 11,488 9,335 116 96 1.01 LINKAGE RELATIONS IN MEN DELI SM 113 When, therefore, the F\ females are mated to yellow white males of the genetic constitution (ywX) Y the progeny will give directly in its phenotypic ratio the proportions in which the gametes are produced as follows : 49.45 gray white: 49.45 yellow red: 0.55 gray red: 0.55 yellow white. The actual experimental results from this type of mating are summarized in Table XXII. The actual linkage value obtained is 1.13 which is substan- tially the same as that shown in the previous table. On the basis of sum- marized data of counts of 81,299 flies, Morgan and Bridges fix the value for crossing-over between these two loci at 1.1 per cent. This is the value we have used in deriving the above gametic ratios. Other factors have been studied in the same way and give different percentages of crossing- over. Thus Morgan and Bridges report the value for crossing-over between white and miniature based on counts of 110,701 flies at 33.2 per cent., between white and vermilion from 27,962 flies at 30.5, between vermilion and bar from 23,522 flies at 23.9, and so on for the whole series of factors in the first group. Table XXII. — ^Factor Linkage in Gray-red Female Drosophilas of the Type {YwX){yWX) Number of flies classified Non-cross-overs Cross -overs Percentage Reported by Gray white Yellow red Gray red Yellow white of cross- ing-over Dexter 1,348 3,258 9,027 13,633 440 1,841 4,292 889 1,412 4,605 16 4 86 3 1 44 1.41 Morgan and Cattell Morgan and Cattell 0.15 1.44 Totals 6,573 6,906 106 48 1.13 Factors in the second and third groups display the same type of linkage relations as those in the first group. As an example we may take the recessive factors black and curved which lie in the second group. A black curved female of the genetic constitution (6c„)(6ct,) crossed with a gray normal male (BCv) (BCv) gives in Fi gray long females and males of the genetic constitution (BCv)(bCv). When such Fi females are crossed back to black curved males the results as reported by Sturtevant and Bridges are given in Table XXIII. The observed percentage of crossing-over between the loci B and Cv in the second chromosome in this experiment amounts to 24.04 per cent. When gray curved females {Bcv){Bcv) are mated to black normal males (6C^)(6C„) the Fi flies are gray normal as in the previous case, but genetically they are of the constitution (Bc„)(6C»). Such females 114 GENETICS IN RELATION TO AGRICULTURE mated to black curved males, according to the same investigators, gave the results tabulated in the last two columns in Table XXIII. Here the percentage of crossing-over amounted to 22.74 per cent., a value substantially in agreement with the results of the reverse factor tests. Table XXIII. — Crossing-over Between B and Cv in Drosophila Phenotype Black curved cf (bcr) (bcv) mated to Gray normal 9 (BC„)(6c„) Non- cross-overs Cross-overs Gray normal $ (Sc„) (bCv) Non- cross-overs Cross-overs rGray normal Gray curved Black normal Black curved Totals Percentage of crossing-over 610 652 184 226 2,292 2,148 644 663 1,262 410 4,440 1,307 24.04 22.74 No Crossing-over in the Male. — The above results show clearly that crossing-over in the female between the loci B and C„ of the second chro- mosome results in the production of approximately 23 per cent, of cross- over gametes irrespective of the particular combination of the factors concerned. Sturtevant and Bridges, however, have shown that there is no crossing-over in the male so that males of the genetic constitution {BCv)(bCv) produce only two types of sperm in the ratio l{BCv) :l(6c„) and males of the genetic constitution (Bc„)(bCv) produce sperms in the ratio, l(5c„) :1(6C,). It is not known just what the absence of crossing-over in the male depends upon. In the case of factors in the X-chromosome or first group, crossing-over would involve exchange of chromatin material be- tween the X- and F-chromosomes. Since these differ strikingly it is not surprising that interchange of chromatin does not take place between these chromosomes for it is difficult to see how the difference could be preserved if crossing-over should occur. But the other chromosomes are alike in both sexes, nevertheless no matter how high the percentage of crossing-over in the female, none whatever has been observed in the male. This has been found true for factors lying in the third group as well as for those lying in the second group, and it is without doubt a general phenomenon. The knowledge that no crossing-over occurs in the male has often been turned to advantage in experimental work. When gray curved flies, {Bcv){Bcv) are mated to black normal^ {bCv)(bCv), the Fi flies are LINKAGE RELATIONS IN MEN DELI SM 115 gray normal and of the genetic constitution {Bcv)(bCv). When these are interbred to obtain the F2 generation the results are shown in the checker- board in Fig. 53. According to this checkerboard, the F2 will consist of flies in the ratio 2 gray normal : 1 gray curved : 1 black normal. No black curved flies are obtained in this cross in F2, and it is of interest to note that no matter what the amount of crossing-over in the female the flies in F2 will always be in the ratio 2:1:1. In F3 black curved flies may be obtained from a certain percentage of matings of black normal or of gray cvu'ved flies. The failure of the double recessive class to appear in F2 has been much used by Morgan in determining the factor group to which new mutations (Be) (bC) 11.5 (BC) 11.5 (fiC„)(Sc„) gray normal 11.5 {BC,){bC,) gray normal 38.5 (Be.) (Be,) gray curved 38.5 (Bc„)(6C.) gray normal 38.5 (.bC.){Bc„) gray normal 38.5 (6C.)(6C.) black normal 11.5 (be,,) {Be.) gray curved 11.5 {bC,){bC^) black normal 38.5 {Be,) 38.5 {bCv) 11.5 (be,) Fig. 53. — F2 obtained by crossing gray curved and black normal flies. belong. For this purpose black-pink flies are crossed with the new mutant type. Since black lies in the second group and pink in the third, if the new factor belongs to either of these groups it will fail to show the corre- sponding double recessive form in F2. Whether it belongs to the sex- linked group is of course readily determined from the sex relations ob- tained in such an experiment, and if the test shows that the factor in question belongs to none of these three groups, by exclusion it must be- long to the fourth. Linear Arrangement of Factors. — It was an old idea of Roux brought forward to explain the division of the chromatin while in the form of along thin thread that the individual elements of the chromatin are arranged in a linear series in the chromosomes. Later Janssens developed the idea that in synapsis homologous chromosomes twist about each other 116 GENETICS IN RELATION TO AGRICULTURE and in separating tend to break apart at places, and in reuniting exchange chromatin material. Morgan has taken these two ideas and applied them to the results of the Drosophila investigations. The twisting of the 0.0 ! Yellow, Spot 1.1 j White, Cherry, Eosin 2.4 Abnormal 6,3 Bifid 12.5 Lethal H 14.6 ' Lethal sb 16.7 Club 19.5 23.7 26.5 Lethal Ilia Lethal sa Lethal III 33.0 Vermilion 36.1 Miniature 38.0 Furrowed 43.0 55.1 57.0 59.5 66.2 Sable Radimentary Bar Fused Lethal sc 16.6 Streak Dachd 34.7 Black 35.0 40.0 52.0 60.4 Jaunty Purple Vestigial Curved 84.1 Arc 90.0 91.9 Speck Morula 0.0 Sepia Bent Eyeless 25.0 Pink, Peach 40.0 Kidney 55.0 Ebony, Sooty 73.0 Beaded 85.0 Rough Fig. 54. — Plotted relations of factors in Drosophila ampelophila. chromosomes in synapsis is held to be the physical evidence of the known interchange of factors between chromosomes which takes place in crossing- over. Moreover, if the factors are the individual elements of the chro- LINKAGE RELATIONS IN MEN DELI SM 117 matin threads which twist about each other and these elements are held to occupy invariable loci in the chromatin thread, then the percentage of crossing-over between any two loci may be taken as an indication of the distance between the factors. For obviously if the chromatin thread is as likely to break between any two chromomeres as between any other two, then the farther apart two factors lie in the chromatin threads repre- senting homologous chromosomes, the greater is the chance that crossing- over will occur between them. The results of the application of this idea to the linkage relations existing in Drosophila are shown in Fig. 54. In this chromosome map of Drosophila the factors have been plotted in a linear series according to their relative position in the chromosomes as determined by linkage rela- tions. The evidence as yet is not sufficient to give an accurate picture of the arrangement of all the factors, but the number of factors plotted and the relations which they display provide further evidence of the corre- spondence between the chromosomes and the factor groups. Morgan has taken 1 per cent, of crossing-over as the unit for expressing linkage relations. Expressed in such units the first chromosome, which contains all the sex-linked factors, has a length of 66.2. The second and third groups, as far as determined, have lengths of 91.9 and 85.0, respectively. These lengths in general correspond fairly well to the known relative sizes of the two large pairs of autosomes when compared with each other and with the X-chromosomes. In the fourth group but two factors are known and their loci are so close together that thus far no crossing-over has been observed between them. Accordingly no definite value can be fixed for their linkage relations. From a knowledge of the small relative size of the third autosome Muller, at the time he an- nounced the discovery of the first factor in the fourth group, predicted that factors in this group would show very close linkage values. This prediction has been upheld satisfactorily and it is further evidence that the chromosome theory of heredity works. The demonstration that factors lie in a linear series in each group provides a unique method of predicting the results of factor behavior. Obviously if a factor is known to belong to a particular group, it is possible to predict confidently that it will display independent segre- gation with factors belonging to other groups. But further than this when the loci of a number of factors in a given group have been plotted accurately, with a new factor it is only necessary to determine the linkage relations with two of the plotted loci in order to determine its locus. When its locus has been determined, its linkage values with any other members in the group may be predicted from its distance in units from those factors. To illustrate, in Group I, if the position of miniature were unknown, it might be tested with vermilion and sable. It would 118 GENETICS IN RELATION TO AGRICULTURE give about 3 per cent, of crossing-over with vermilion and about 7 per cent, with sable. Knowing the position of the vermilion locus at 33.0 and the sable locus at 43.0, we would be able from these data to fix the locus position of miniature at about 36.0. With this value determined we could confidently predict for example that miniature and white would show somewhat less than 35 per cent, of crossing-over or miniature and bar about 21 per cent. The ability to make such predictions is a unique product of recent investigations in heredity. How experimental results support the hypothesis of linear arrange- ment of factors may be illustrated by what Morgan calls a three-point experiment, i.e., an experiment involving three different factors in the same chromosome. We may take three factors which are in widely separated loci in the chromosome, white at locus about 1.0, miniature at about 36.0, and bar at about 57.0. The summaries which Morgan and Bridges have given of the data involving these three loci are in- cluded in Table XXIV. White and miniature give directly 33.2 per cent, of crossing-over, and miniature and bar 20.5 per cent. Since the distance between white and miniature plus that between miniature and bar is equal to 53.7, this latter value should represent the distance between white and bar. But direct experimental determinations of the per- centage of crossing-over between white and bar give a value of 43.6 per cent., which is 10.1 per cent, short of the calculated value. Table XXIV. — Crossing-over for the Loci W, M, and B' in Drosophila. Data Obtained by Mating Females op the Constitution {wmB'X){WMh'X) With Triple Recessive Males (wmb'X)7. Character combinations Number of flies classified Number of cross- overs Per cent, of crossing- over White miniature. . . . Miniature bar White bar 110,701 3,112 5,955 31,071 636 2,601 33.2 20.5 43.6 The reason for this should be plain from a consideration of Fig. 55 which shows diagrammatically how the chromosomes behave in a three- point experiment. On the left in the two upper groups are represented the two chromosomes with the factors in the original positions in which they were derived from the parents. On the right the homologus chromo- somes are shown twisted about each other, and dit A, B, C, and D the types of chromosomes which are obtained after chromatin interchange in synapsis. The numbers below refer to the relative frequency of pro- duction of the four types of chromosome pairs in this three-point experi- ment based on the data of Table XXIV. In A (Fig. 55) no exchanges LINKAGE RELATIONS IN MEN DELI SM 119 of chromatin have occurred which affect the relations of the factors to each other, so that this type of separation after synapsis gives the non- cross-over gametes {wmB'X) and {WMh'X). Types B and C involve single breaks in the chromosomes followed by chromatin interchange in reunion. They are the single cross-overs and give the cross-over gametes /'-\ B 46.3 28.15 15.45 10.1 Fig. 55. — Diagram to show crossing-over in a three-factor experiment. {wMh'X) and {WmB'X) and {wmb'X) and (WMB'X) respectively. Finally in type D the chromosomes have broken and exchanged material at two points. This type is called double crossing-over and results in the production of gametes of the genetic constitutions {wMB'X) and (Wmh'X). In this last case, although chromatin interchange has occurred between the two chromosomes, the relations between the loci W and B' remain unchanged. 120 GENETICS IN RELATION TO AGRICULTURE The occurrence of double crossing-over accounts for the low per- centage of crossing-over between white and bar as compared with the sum of the values given by white and miniature and miniature and bar. The value for crossing-over between W and M is given by B -\-^ = 28.15 + 5.05 = 33.2 per cent. and similarly between M and B' C + ~ = 15.45 + 5.05 = 20.5 per cent. consequently the distance between W and B' as measured by adding together the values for W and M and M and B' gives the equation B -\- C -\- D = 53.7. Since double crossing-over of the type D does not involve a rearrange- ment of the loci, W and B', however, the actual crossing-over obtained experimentally must fall short of the computed distance by a value equal to D as given by the equation B + C = 43.6 per cent. The lowering of the percentage of crossing-over when extreme distances are involved is, therefore, a logical consequence of the relations existing between linked factors. Obviously double crossing-over occurs much less frequently in short distances than in long ones. Consequently since a factor map is designed to give the total values for crossing-over between the different loci, such a map is prepared so far as possible from experi- ments involving short factor distances. If such data are not at hand simple methods of interpolation are used to locate the loci. It should be noted in passing that variations in linkage values some- times occur among members of a given set of factors. Bridges has pointed out that in some cases at least the percentage of crossing-over depends somewhat on the age of the female, and Plough has detected definite effects of extremely high or low temperatures on the percentage of crossing-over between factors of the second chromosome in Drosophila, although crossing-over in the first and third chromosomes was not in- fluenced by the changes in temperature. Besides such variations, how- ever, definite factors have been discovered (Sturtevant) which lower the percentage of crossing-over. Muller has shown that such a factor exerts a particularly disturbing action in the third chromosome in which it is located. But even in cases of variation in linkage values the order of the factors in the chromosome is not disturbed. The relations shown, there- fore, in cases involving variations in linkage are in harmony with the conception of linear arrangement of factors in the chromosomes. LINKAGE RELATIONS IN MEN DELI SM 121 The most striking confirmation of the hypothesis of linear arrange- ment is found in the case of "deficiency" in the A'^-chromosome, which was investigated by Bridges (see p. 155) and in which the location of forked spines within the deficient region "was detected and proved as a result of deliberate search among those genes which had previously been mapped closest to bar!" The Mode of Interchange in Crossing-over. — Factor interchange conceivably might take place by interchange of isolated factors here and there along adjacent threads or it might follow as a consequence of inter- change of relatively large sections of chromatin between chromosomes. The sectional mode of chromatin interchange appears to have more cyto- logical evidence in its support and Plough's recent studies on the effect of temperature on crossing-over corroborate Muller and Bridges' inference that crossing-over takes place in the fine thread stage of synapsis, which would be the most favorable stage for sectional interchange. But breed- ing investigations of themselves clearly establish this hypothesis. Thus Muller made up females which contained twelve sex-linked mutant factors. These females received from one parent the factors for yellow body color, white eye color, abnormal abdomen, bifid wings, vermilion eye color, miniature wings, sable body color, rudimentary wings, forked spines, and from the other parent the mutant factors cherry eye color, club wings and bar eyes. Using the system of writing the genetic formulae which has been followed in this text, these females were of the genetic constitution (ywA%Civmsrfb'X)iYw<=a'BiCiVMSRFB'X). Muller found in tests of 712 individuals arising from gametes from such females, that the proportions of crossing-over between factor loci in the formation of gametes occurred according to the figures given in Table XXV. The results show that in this experiment there was no crossing-over in 54.4 per cent, of cases; single crossing-over in 41.7 per cent., and double crossing-over in 4.2 per cent. No example of triple crossing-over was found among these flies, but a few such cases have been observed. The values agree satisfactorily with those calculated from the three-point experiments involving the loci W, M, and B' in this same chromosome. If we consider the double cross-overs which were obtained in this experiment we find abundant evidence in support of the sectional mode of chromatin interchange. It is difficult to visualize the relations from the numerical data, consequently Fig. 56 has been prepared to illustrate diagrammatically the types of double crossing-over obtained in these experiments. In all but one case the points of crossing-over are far re- moved from each other, and even in the exceptional case the distance between the points of crossing-over may have been as great as nineteen units distance. 122 GENETICS IN RELATION TO AGRICULTURE Table XXV.— Classification of Factor Combinations Transmitted by Females OF Drosophila having the Genetic Constitution {ywA'biCivmsrfb'X)iYvfa'BjCiVMSRFB'X) No crossing-over 186 200 386 Crossing-over between the loci Number of yellow flies Number of gray flies Totals Yellow and white White and abnormal Abnormal and bifid Bifid and club - Club and vermiUon Vermilion and miniature Miniature and sable Sable and rudimentary Rudimentary and forked Forked and bar Total single cross-overs Double crossing-over Y and W:Ci and V YandW-.M and S Y and W:S and R Y and W:R and F TF and A' : Cj and F W and A' : R and F A' and Bi-.C I and V A' and Bi : S and R Bi and Ci : M and S Bi and Ci : S and R Ci and 7: 7 and M Ciand V:SandR Ci and V : R and F Ci and V : F and B' Total double cross-overs 2 3 4 17 46 7 18 28 5 5 11 27 51 9 19 38 5 1 15 44 97 16 37 66 5 1 296 Totals 30 Interference. — Interference is merely a consequence of the sectional mode of chromatin interchange between homologous chromosomes. The term is used to designate the observed fact that when crossing-over takes place at a particular point in the chromosome the regions for some dis- tance on both sides are protected from coincident crossing-over. The operation of interference is well illustrated in Muller's data, although the numbers are not sufficient to warrant a quantitative determination of its effect. With long distances interference decreases, which is in accord- ance with expectation. Even for relatively long distances, however, as for the loci W, M and B' which we have already considered in detail LINKAGE RELATIONS IN MEN DELI SM 123 there is still some evidence of interference. Based purely on the laws of chance, if crossing-over occurs between W and M in 33.1 per cent, of cases and between 71/ and B' in 20.5 per cent., then the chance of coinci- dent crossing-over is equal to the product of the independent chances of crossing-over. This gives a value of 6.8 per cent, which is slightly greater than the value 5.05 per cent, calculated from the experimental data. A three-point experiment involving shorter distances, however, gives a clearer idea as to the extent of interference. Morgan and Bridges 35 A 5 5 B, 29-V 31 -M I I 48.5R I I SO B^ I I I I i I I I I I i i 11 21 1111271821 Fig. 56. — Diagram showing types of double crossing-over in females of Drosophila heterozygous for twelve sex-linked factors. The figures below indicate the number of times the type occurred in 712 cases. (The loci indicated in the " map" at the left are only approx- imately correct according to recent data of Morgan and Bridges, but they are sufficiently accurate for the purpose of this diagram.) have reported such an experiment involving the loci for vermilion, sable, and bar with the results given in Table XXVI. From this table the total percentage of crossing-over between vermilion and sable is 9.8 per cent, and between sable and bar 13.8 per cent. The expected percentage of double crossing-over for these values obtained by taking 9.8 per cent, of 13.8 per cent, would be 1.35 per cent. The observed amount of double cro.ssing-over, 0.25 per cent., is only about one-fifth of this value. That interference is normally to be expected from the method of chromatin interchange in synapsis may be seen clearly by a consideration of Fig. 57. Thus if the chromosomes have a modal length in loop twisting about each other in synapsis, then a crossing-over at point B 124 GENETICS IN RELATION TO AGRICULTURE Table XXVI. — ^Linkage of Vermilion, Sable, and Bar in Drosophila '■ Non- cross-overs Single cross-overs between Characters Vermilion and sable Sable and bar Double cross-overs Gray red normal 755 734 724 845 608 800 665 641 110 92 97 87 80 95 81 74 140 151 131 126 123 129 107 108 4 Gray vermilion normal 1 Sable red normal 4 Gray red bar 4 Gray vermilion sable 3 Gray vermilion bar 1 Sable red bar 1 Vermilion sable bar 3 Totals 5,772 716 1,015 21 Percentages 76.7 9,53 13.49 0.28 A D F Fig. 57. — Diagram to illustrate interference in crossing-over. would protect the loci A, C, and D on either side of it from crossing-over because there would be no close twisting of the chromosomes at these points. As we move on toward E, however, the frequency of double crossing-over would become greater and greater until at E where the modal length of loop was attained double crossing-over values approach- ing those expected on the basis of pure chance would be obtained. Muller has actually shown that such conditions are fulfilled in his twelve-point experiments and he has been able to plot a curve showing that the observed frequency .of double crossing-over gradually increases until when the modal length is reached the curve coincides with that based on pure chance. Thus we see again how another point of attack has lent support to the conception that the factors are arranged in a linear series and that the linkage relations of factors are referable to the mechanical consequences of relative positions in the linear series. Bridges points out that interference stands in about the same relation to linkage as linkage does to free Mendelian assortment. Also that the development of the idea of interference is an illustration of the advantage of the chromosome hypothesis. The existence of this phenomenon was LINKAGE RELATIONS IN MENDELISM 125 originally deduced by Miiller and Sturtevant from a consideration of linkage as a chromosome hj'pothcsis. Linkage Phenomena in Other Plants and Animals. — Our extended discussion of linkage relations has been based practically entirely on Drosophila anipelophila because the factor analysis in this species pro- vides us with a body of data incomparably superior to that provided by any other species. Nevertheless there are other scattered cases of linkage in many species of plants and animals. In plants Bateson first described the phenomenon of linkage in sweet peas for the characters round pollen and red flower color. Later the factor for hooded standard was found to be linked to the factors for these two characters. Later Punnett discovered linkage in a second group of factors consisting of those for green axils, cretin flower shape, and sterile anthers. Gregory has described a group of five factors in the Chinese primrose, those for red stigma, red flower color, long style, dark stems, and light corolla tube. In garden peas, Vilmorin and Bateson have both reported linkage between the factors for round as opposed to wrinkled seed and tendrilled as opposed to non-tendrilled or "acacia" leaves, and Hoshino has suggested coupling between red flower color and a factor for late flowering. Very recently O. E. White has investigated or com- piled the data on thirty-five factor differences in Pisum and has presented data for four linked groups of characters. In the garden snapdragon, Baur found linkage between the factors for red flower color and for the "picturatum" type of color pattern in the flower and also clear cut evidence of linkage between some other factors. Surface has shown that in oats the factors for pubescence on the back of the lower grain, pu- bescence on the back of the upper grain, and black grain color are closely linked. Enough cases have also been reported for other plants to demon- strate that linkage relations are of general occurrence in plants. In animals, Castle and Wright have suggested that linkage occurs in rats between the factors for red eye-color and pink eye-color. A clear case also has been established by Tanaka in the silkworm moth in which a series of factors for larva pattern are linked to factors for yellow and white cocoon color. Besides these cases there are a large number of cases of sex-linked inheritance in many animal forms. These will be discussed in Chapter XI. It is clear from what we have stated above that aside from our knowledge of linkage in Drosophila, we have not progressed far in the investigation of linkage relations. Several factors have contributed to this condition. Most of the forms which have been used in genetic in- vestigations have a larger number of chromosomes than Drosophila, a fact which considerably complicates such investigations. Most genetic data have been obtained from experiments which involve but few fac- 126 GENETICS IN RELATION TO AGRICULTURE tors. If the chromosome number is large, the chances" of such experi- ments showing factor Hnkage are sHght. Finally there are experimental difficulties in the way of securing an adequate body of data for most animals and for practically all plants. It is necessary to conduct most technical investigations in heredity with relatively meager financial sup- port, consequently the expenditures necessary to obtain sufficient data of this kind would be prohibitive for most of the larger animals and plants. Moreover, on account of the time required to raise a sufficient number of generations and to classify the individuals a considerable time must elapse before a body of data can be gathered in any species sufficient to submit it to the critical tests necessary to establish the chromosome theory. Drosophila with its prolific breeding tendencies, short life cycle, and ease of handling provides a form far superior to any other thus far investigated for the elucidation of factor relations in general. It is safe to say that our ideas of linkage for some time to come will be largely determined by the results of the Drosophila investigations. Particularly is this true because thus far none of the linkage phenomena exhibited by other animals and by plants have yielded evidence contradic- tory to the chromosome theory. The number of factors which have been investigated in several species exceeds the number of pairs of chromo- somes, nevertheless in no single case has there been a clear demonstra- tion that the number of independently Mendelizing factors exceeds the number of pairs of chromosomes. Moreover, those cases of linkage which have been discovered are largely of factors for wholly unrelated characters, just as in Drosophila. Added to this the ratios are of the same diverse orders of magnitude and the linkage relations in general show no essential difference from those which are displayed by Drosophila. It would be nothing short of inconceivable, in fact, that the conclusions reached from the Drosophila investigations are not applicable in all their essential features to plant and animal forms in general. On the basis of the sweet pea and Primula investigations, the English school of geneticists, represented particularly by Punnett and Trow, has developed a theory of linkage very different from that outlined in this chapter, which is called reduplication. According to this hypothesis segregation occurs in a series of cell divisions preceding the reduction divisions, and for linked factors gives gametic series mostly of the form For coupling (n — 1):1:1: (n — 1) For repulsion 1: (n — 1): (n — 1): 1. In these ratios n is some power of two. Interaction of two such series may give secondary reduplications which give different values for the terms of the ratio. This theory of linkage cannot, however, lay claim to the experimental support which the chromosome theory has obtained, LINKAGE RELATIONS IN MENDELISM 127 nor is it based on any known cytological phenomena. The series of ratios which lent original support to the theory appear to be no more fre- quent than should be the case on the basis of chance, and many which are supposed to fall into the series have been placed there on evidence which is entirely inadequate. The large series of linkage values which have been obtained in Drosophila demonstrate clearly that all intermediate ratios can be obtained, and since all other conditions are satisfied by the chro- mosome theory it seems unreasonable to give it up for an hypothesis which has no cytological support and an uncertain amount of experimental support. Moreover, it may be safely stated that all cases of linkage thus far reported may be explained according to the chromosome theory of linkage. The mathematical relations existing in linkage phenomena are of interest because they provide a method of determining the genetic relationships involved in certain cases of somatic correlations. If two factors are linked in inheritance it follows that a larger proportion of the population will display the corresponding two characters than would be the case, if the factors were inherited independently. Consequently character correlations of this type are an index to factor linkage. In Tables XXVII and XXVIII the results of various strengths of factor linkage and the consequences with respect to the gametic and phenotypic ratios are given. These tables show clearly that the only satisfactory method of deter- mining the presence of linkage and its value is to cross back the hetero- zygous individual to individuals recessive for both factors. In such crosses the phenotypic ratio corresponds exactly to the gametic ratio, and it is, therefore, possible to determine the percentage of crossing-over by this method with a much greater degree of precision than from ordinary F2 populations. When the two dominant factors enter the cross from opposite sides it is practically impossible to determine the linkage values by simply mating Fi individuals together, for comparatively large differences in linkage value may affect the phenotypic ratio so slightly that the deviations, in small populations at least, might be ascribed merely to the operation of the laws of chance. The significant feature of such ratios is the small proportion of double recessives which appear. Thus with crossing-over values exceeding 20 per cent., this class practically disappears in experiments involving the usual number of individuals in a population. Moreover, such matings in species which dis- play crossing-over only in the sex-homozj^gotes as shown in Table XXVIII give the ratio 2:1:1 for all percentages of crossing-over when the domi- nant factors enter the cross from one parent only. A careful consider- ation of these two tables will show clearly how difficult it is to determine linkage values precisely except by properly planned experiments, and in this difficulty lies the reason for many errors of interpretation. 128 GENETICS IN RELATION TO AGRICULTURE Table XXVII. — ^Linkage Relations — Crossing-over in Both Sexes Pl Percentage of crossing-over Gametic ratio Phenotypic ratio per thousand AB:Ah:aB:ab AB Ah aB ah 1 99.0:99.0:1 500.025 249.975 249.975 0.025 2 49.0:49.0:1 500.100 249.900 249.900 0.100 3 32.3:32.3:1 500.225 249.775 249.775 0.225 4 24.0:24.0:1 500.400 249.600 249.600 0.400 5 19.0:19.0:1 500.625 249.375 249.375 0.625 6 15.7:15.7:1 500.900 249.100 249 . 100 0.900 X 7 13.3:13.3:1 501.225 248.775 248.775 1.225 .0 8 11.5:11.5:1 501.600 248.400 248.400 1.600 9 10.1:10.1:1 502.025 247.975 247.975 2.025 10 9.0: 9.0:1 502.500 247.500 247.500 2.500 20 4.0: 4.0:1 510.000 240.000 240.000 10.000 30 2.3: 2.3:1 522.500 227.500 227.500 22.750 40 1.5: 1.5:1 540.000 210.000 210.000 40.000 50 1: 1 : 1 :1 562.500 187.500 187.500 62.500 40 1.5:1:1 1.5 590.000 160.000 160.000 90.000 30 2.3:1:1 2.3 622.500 127.500 127.500 122.500 20 4.0:1:1 4.0 660.000 90.000 90.000 160.000 10 9.0:1:1 9.0 702.500 47.500 47.500 202 . 500 A 9 10.1:1:1 10.1 707.025 42.975 42.975 207.025 e 8 11.5:1:1 11.5 711.600 38.400 38.400 211.600 X 7 13.3:1:1 13.3 716.225 33.775 33.775 216.225 05 6 15.7:1:1 15.7 720.900 29.100 29.100 220.900 5 19.0:1:1 19.0 725.625 24.375 24.375 225.625 4 24.0:1:1 24.0 730.400 19.600 19.600 230.400 3 32.3:1:1 32.3 735.225 14.775 14.775 235.225 2 49.0:1:1 49.0 740.100 9.900 9.900 240.100 1 99.0:1:1 99.0 745.025 4.075 4.075 245.025 Table XXVIII. — Linkage Relations — Crossing-over only in the Sex HOMOZYGOTE, NoN-SEX-LINKED FACTORS Pi Percentage of crossing-over Gametic ratio Phenotypic ratio per thousand AB:Ah:aB:ah AB Ah aB ah Ah X aB n 100 -71.100 -n_^ n n 500 250 250 1 99.0:1:1:99.0 747.5 2.5 2.5 247.5 2 49.0:1:1:49.0 745.0 5.0 5.0 245.0 3 32.3:1:1:32.3 742.5 7.5 7.5 242.5 4 24.0:1:1:24.0 740.0 10.0 10.0 240.0 X 5 19.0:1:1:19.0 737.5 12.5 12.5 237.5 pO 6 15.7:1:1:15.7 735.0 15.0 15.0 235.0 Q 7 13.3:1:1:13.3 732.5 17.5 17.5 232.5 8 11.5:1:1:11.5 730.0 20.0 20.0 230.0 e 9 10.1:1:1:10.1 727.5 22.5 22.5 227.5 X 10 9.0:1:1: 9.0 725.0 25.0 25.0 225.0 cq 20 4.0:1:1: 4.0 700.0 50.0 50.0 200.0 30 2.3:1:1: 2.3 675.0 75.0 75.0 175.0 40 1.5:1:1: 1.5 650.0 100.0 100.0 150.0 50 1 :1:1: 1 625.0 125.0 125.0 125.0 CHAPTER VII THE NATURE AND EXPRESSION OF MENDELIAN FACTORS In previous chapters the formal relations which exist in the trans- mission of factors from parent to offspring have been discussed. It has been shown that these relations may be ascribed to the locus positions which factors occupy in the chromosomes. This single assumption taken together with the known behavior of the chromosome mechanism in its cycles explains very simply the two great categories of inheritance with respect to distribution of factors, namely independent segregation and linkage. Obviously, however, these are merely formal considerations, it is of considerable importance to know something about the factors themselves and the physiological interactions which they display with one another in the development of characters in the individual. It is to this problem that this and following chapters are addressed. It is true that as yet we know next to nothing about the factors themselves with respect to their physical and chemical constitution, we know them merely by their actions. We regard them as the loci ar- ranged in a linear series in the chromosome, we know they have certain characteristic effects in development and by these effects we recognize them. It is important to note that our knowledge of their behavior even is based on factor differences, not op a study of the factors them- selves. Thus we know that a certain locus in the germinal substance in Drosophila is concerned with the production of red eye color because when it is changed in a particular fashion, the eye color developed is no longer red, but white. We have no means of knowing how profound the relation of this factor to the other factors in the system is, nor can we judge as to the nature of the change in the locus by which the course of development was shifted from red to white in the production of eye color in Drosophila. Nevertheless a few things at least are known con- cerning the effects of factors in development and even in this vague field more and more facts are being discovered all the time. Factors are the genetic representatives of certain characters. Thus if a fly has a genetic constitution containing, among other factors for eye color, the factor w, then it will develop white eyes. In this particular case the eye color is practically the only character affected. Similarly in corn, if a mutation occurs in one of the basic aleurone color factors, for example, a change in the chromogen factor C to c then that corn thus 9 129 ISO GENETICS IN RELATION TO AGRICULTURE developing is white as respects aleurone color. Here apparently only aleurone color is concerned. Similarly in other cases much more insig- nificant changes may be connected with definite factor differences. Thus a forked condition of the spines in Drosophila is dependent upon a definite factor difference, a recessive factor in this case. One could go on and recount indefinitely factors which cause only very slight character changes. Any character change, therefore, however slight, may be based on genetic factor differences. The only valid genetic test is the pedigree breeding method, at the same time giving due con- sideration to environmental influences which may obscure or temporarily cover entirely the underlying genetic differences. Very great somatic differences may also be dependent upon differences in single factors in individuals. Perhaps the most striking of these are large size differences such as are found in beans, peas, and even in animals at times. Thus in beans the main difference between pole and bush beans is dependent upon a single factor difference. The difference between tall and dwarf varieties of peas is of a similar nature and has been fully discussed above. Certain types of dwarfing in man appear to depend upon single factor differences and in Drosophila there are factors which determine the production of giant races and others of dwarf races. Moreover, factor differences show striking relations to one another. Thus in Drosophila there are factors for eye color which change the shade of red in the eye, some resulting in a darker and many in lighter shades, but there is also a single factor difference which results in white eyes or in other words in the entire loss of color in the eyes, and even further there is a factor for an eyeless condition, which when a part of the genetic constitution of a fly results in the production of mere rudiments of eyes or even none at all. Very frequently single factors may cause such profound changes as to alter the entire appearance of the individual and interfere more or less with all its functions. Such, for example, is the case with fasciated forms in plants, some of which at least are dependent upon simple factor differences. A striking case of this type has been reported by O. E. White in tobacco. In this fasciated variety the number of leaves is greatly increased, from 24 to as high as 80, the stem is flattened and exhibits a characteristic fasciated condition, and the flowers are very abnormal. The abnormality of the flowers extends to every part, the numbers of sepals, petals, stamens, and ovary locules are increased, and striking deformities of these parts give evidence of the disturbing effect of the factor. The abnormal effects of the factor are not confined to external characters, but cytological studies show that the division figures, par- ticularly in reduction, show marked irregularities which may be expressed in an increase in the number of chromosomes, or in a breaking down of THE NATURE AND EXPRESSION OF MENDELIAN FACTORS 131 cells during division, or in various other peculiar phenomena. The ab- normal variety also displays a certain degree of sterility, probably asso- ciated with abnormal cell division. In spite of all the differences both external and internal which this mutation displays when compared with the normal variety from which it arose, its behavior in inheritance shows clearly that only a single factor difference is involved. When crossed with the normal type, the Fi is intermediate, and in F2 segregation is in approximately the ratio 1 abnormal: 2 intermediate:! normal. The F2 homozygous segregants are exact duplications of the original pure forms, the normal segregants are in every respect as normal as the normal parent and the abnormal segregants are no less abnormal than those of pure abnormal races. The heterozygous forms are throughout clearly dis- tinguishable from abnormal homozygotes on the one hand and normal homozygotes on the other. Taken as a whole it would be difficult to find a better example of the profound effects which may result from a single factor difference. Lethal factors also exist which affect vital organs and result in the death of individuals homozygous for them. Excellent examples of such disturbing factors are those which affect the production of chlorophyll in plants. A number of species of plants at times produce races in which under experimental conditions approximately one-fourth of the seedlings are yellow or white instead of green and hence die soon after germination. Svich strains are particularly common in cereals, and in maize in almost any variety when a large number of self-fertilized ears are tested, a number of strains may be found which produce seedlings about one-fourth of which die as soon as the food supply of the endosperm is exhausted on account of deficiency in chlorophyll production. Since the homozygous recessive forms of albino strains die soon after germination, it follows that such strains must be propagated by means of the heterozygous individuals. The operation of such a scheme is illustrated in the following case. The original self-fertilized ear gave on germination 3 fully green seedlings to one which was pure white and which died shortly after germination. If we call the albino factor g in this case and its normal allelomorph present in the green plants G, we may assume that this ear was produced by a heterozygous green plant of the constitution Gg. Half the pollen grains of such a plant carry the factor G and half the factor g; and likewise in the ovules half bear the factor G, and half g. By self-pollination of such a plant, random fertili- zation of the ovules by the pollen grains results in grains in the ratio lGG:2Gg:lgg. Although grains of these different genotypes are indis- tinguishable in appearance, those of the genetic constitution GG and Gg produce fully green plants, while those which are gg produce albino seedlings which are incapable of independent existence on account 132 GENETICS IN RELATION TO AGRICULTURE of their lack of chlorophyll. Those green plants of the genetic constitu- tion Gg when self-fertilized produce grains one-fourth of which again give albino seedlings as in the previous generation. The green plants of the genetic constitution GG, however, since they are homozygous for the factor G produce nothing but green plants in succeeding generations. R « m « 08 @H @ (D t Fig. 58. — Disturbance of phenotypic ratio by a recessive sex-linked lethal factor. Com- pare with Fig. 36. Morgan has demonstrated the existence of a number of lethal factors in DrosophUa. These factors result in the death of the individuals at some time before they reach the adult stage. They are particularly found among sex-linked factors, because sex-linked recessive factors have no normal allelomorphs in the male. The results of the presence of sex-linked lethal factors is shown diagrammatically in Fig. 58. As in corn, the strains are propagated by means of heterozygous individuals. Since such individuals can only be females in the case of sex-linked THE NATURE AND EXPRESSION OF MENDELIAN FACTORS 133 factors in Drosophila, it follows that the strains must be propagated by means of females heterozygous for the factors. The diagram shows how such lines are maintained. A heterozygous female produces eggs half of which bear the normal factor L, and half bear the lethal factor /. When mated to a normal male the X-chromosome of which bears the normal factor L, half the daughters are normal homozj^gotes and half are heterozygous for /. Half the males receive an A'-chromosome bear- ing the factor L, and consequently are normal and half receive an X- chromosome bearing the factor /. These latter die before reaching the adult stage, consequently a heterozygous female produces flies two- thirds of which are females and one-third males. The unusual sex- ratio provides a convenient test for heterozygous females and by this means the strain may be continued. Some of the consequences of the presence of lethal factors when linked with other factors are of importance because of the disturbances to which they give rise in Mendelian experiments. An illustration of such effects may be taken from Lethal III in Drosophila which is located at about the locus 26.5 in the X-chromosome. It is about 25 units distance from the locus for white eyes. If now a white-eyed female heterozygous for Lethal III be crossed with a red-eyed male, as shown diagrammatically in Fig. 58 all the females will be red-eyed but only half will be homozygous for the normal factor L3. These females, homozygous for L3, produce flies in the ratio of 1 red 9 : 1 red cf : 1 white 9 : 1 white d^ when mated to their brothers. The other half of the Fi females, on the other hand, will be heterozygous for L3 and conse- quently, since crossing-over takes place in 25 per cent, of cases, they produce gametes in the ratio S{wkX):3(WL3X):l{Wl3X):l(wL3X). When such a female is mated to an Fi male fly the ratio is distinctly different from that obtained with the other females, in this case 4 Red 9 :3 Red d^:4 White 9 :1 White cf. The ratio of sexes in this latter case is 2 female: 1 male and the same is true in Fi. The sex ratio gives an immediate clue to the disturbing factor and leads to a true explanation of the cause of the disturbance. Manifold Effects of Factors. — In a preceding section of this chapter it has been shown how far reaching may be the effects of single Mendelian factors, and in the present account it is intended to deal specifically with what Morgan has termed the manifold effects of single factors. Careful study has revealed the fact that although factors are restricted in their conspicuous results to certain characters, nevertheless they may have other less noticeable results which are none the less definite and constant. Baur has observed for example in Antirrhinum that the factor which produces pure white blossoms also yields plants which are distinctly weaker in growth and are smaller than those which possess 134 GENETICS IN RELATION TO AGRICULTURE the normal allelomorph for this factor. Plants possessing the recessive factor may be recognized in the seedling stages by a peculiar coloration of the edges of the leaves and even better by the characteristic epidermis of the leaf blades. Manifold effects of factors are probably very common but very little definite work has been reported along this line. Morgan, however, has called attention to some cases in Drosophila. Thus there is a factor for club wings, and in strains of this type flies appear the wing pads of which fail to unfold after emergence. But this character is not constant, in fact about 80 per cent, of the flies in a pure strain have normal wings. Subsequent study, however, has shown that in such stocks the absence of spines on the side of the thorax is a constant differential test. These differences are shown in the accompanying figure (59). By employing the absence of spines as the differential test it is possible to classify mixed populations of "normal" and "club" flies accurately without paying any attention to wing characters. The Variability of Factor Expressions. — Factors also vary in the effects which they produce. We have pointed ovit that in pure strains of club- winged Drosophila (Fig. 59) only about 20 per cent, of the flies exhibit the unfolded wing pad characteristic of tlie club mutation. On the other hand, the absence of spines on the side of the thorax determined by the same factor appears to be an invariable characteristic of the club-winged flies. Sometimes this variability in factor expression may be traced to a defi- nite environmental condition. This is certainly true of the red Primula which produces red flowers under ordinary temperature conditions, but which when placed under abnormally high temperatures produces white flowers. The production of chlorophyll in some strains of corn, likewise, depends on generally favorable environmental conditions. This has been demonstrated by Miles for the yellow-green type of chlorophyll reduction. Plants heterozygous for this factor produce grains three-fourths of which produce fully green plants on germina- tion, but the other one-fourth produce pale yellowish seedlings with a tinge of green. The yellowish seedlings die under ordinary conditions, but in particularly favorable surroundings they continue to live and soon develop the normal chlorophyll coloration. If self -fertilized, they produce only yellowish plants which must again be given very favorable condi- tions for the production of the normal green leaf color. In Drosophila a number of environmental relations have been de- scribed. Thus Morgan has studied in considerable detail the influence of environment on the development of abnormal abdomen. Flies with the dominant factor for abnormal abdomen should all exhibit the char- acteristic type of deformed abdomen shown in Fig. 60; but this is not the THE NATURE AND EXPRESSION OF MEN DELI AN FACTORS 135 case, for pure mutant stocks constantly show a high percentage of flies with normal abdomens. This variability in abdomen characters has Fig. 59. — Club-winged Drosophila. At a characteristic unfolded wing pads. At c the absence of spines on the side of the thorax is shown in comparison with the normal con- ditions, b. {From Morgan.) Fig. 60. Mutant type of Drosophila ampelophila called abnormal abdomen (the wings have been cut off); a, female; b, male; c, female that approaches the normal type. Development of this character is dependent upon moisture. (From Morgan.) been found to depend upon the condition of the food. When the food is moist a high percentage of flies have abnormal abdomens, but when the larvae are raised on dry food nearly all of them have normal abdomens. 136 GENETICS IN RELATION TO AGRICULTURE On account of these relations the expected Mendehan behavior of this factor in crosses with normal flies is obscured in cultures grown on dry food, but with moist food Mendelian expectations are completely fulfilled. Moreover, the variability in the expression of the abnormal condition of the abdomen is not connected with any variability in the factor itself but is merely an expression of a variable reaction of the factor to the environment. Normal flies possessing the factor for abnormal abdomen when given moist food produce offspring just as abnormal as those from abnormal flies. The factor itself is invariable just as in a chemical system the elements which are in the system are invariable but may produce different results according to the dilution, temperature, and other conditions under which the reaction is going on. The reduplicated stock in Drosophila shows similar relations to en- vironmental conditions. The characteristic feature of this mutation is the production of extra legs or parts of legs. At normal temperatures very few flies show this condition, but when strains are grown at 10°C. a high percentage of them show supernumerary legs. As with ab- normal abdomen and moist food, so Miss lloge has shown that with temperatures below 10° these flies satisfy Mendelian expectations when crossed with normal strains, but at ordinary temperatures of cultivation the phenomena are entirely obscured. Duplicate Factors. — A number of cases are known where similar or identical effects are produced by factors located in different loci in the germinal substance. A case in point which has been subjected to excel- lent analysis is that for capsule form in the common shepherd's purse (Bursa) . When the form having flattened triangular capsules is crossed with that having top-shaped seed pods, the Fi plants produce triangular capsules. When the F2 is grown approximately 15 produce triangular capsules to one which produces top-shaped capsules. Such a result may be explained by assuming that two recessive factors, c and d, combine to produce the top-shaped capsule. The top-shaped race then is of the genetic constitution ccdd, and the contrasted tri- angular-shaped race is CCDD. The factors C and D are fully dominant and produce identical results, namely plants bearing the typical tri- angular-shaped seed pods. Consequently selfing Fi plants of the genetic constitution CcDd gives F2, 15 plants with triangular pods to 1 with top-shaped pods. The checkerboard for this case is shown in Fig. 61. If this analysis is valid for the inheritance of capsule form the F3 and subsequent generations should display a characteristic type of behavior as shown in the checkerboard. In each square is given the ratio in which the particular genotype should segregate in F3. Thus it will be seen that THE NATURE AND EXPRESSION OF MENDELIAN FACTORS 137 7 families should breed true for triangular capsules. 4 families should give 15 triangular :1 top-shaped. 4 families should give 3 triangular: 1 top-shaped. 1 family should breed true for top-shaped capsules. Shull applied this test to his cultures and obtained substantial agreement with theory throughout. Fig. 62 gives a graphic summary of his experi- mental results. c/— CD Cd Y cD f CD- Cd- cD- cd- CD. CD 1:0 Cd-CD 1:0 cD-CD 1:0 cd CD CD-Cd 1:0 Cd-Cd 1:0 cD-Cd 15:1 cd- Cd 3:1 CD. cD 1:0 Cd-cD 15:1 cDcD 1:0 cd^cD 3:1 CD. cd 15:1 Cd-cd 3:1 cD • cd 3:1 cd ■ cd 0:1 Fig. 61. — Checkerboard diagram to visualize the genetic relations in a dihybrid Ft family of Bursa bursa-pastoris X Heegeri, in respect to the capsule-characters. The capsules figured in each square indicate by their outline their phenotype, and by their oblique ruling their genotype, the gene C being represented by lines from upper right to lower left, and D from upper left to lower right. Homozygotes are densely lined, heterozygotes more sparsely. The ratios indicate the expectation in Ft when a plant having the genotypic constitution indicated in the same square, is self-fertilized. {After Shull.) When three duplicate factors are concerned in a hybrid the ratio in 7^2 is 63 : 1, with four factors, 255 : 1, and so on. The first case of dupHcate factors was that described by Nilsson-Ehle in wheat. Here the red color of certain races of wheat depends on the presence of three dominant Mendelian factors so that such races are to be represented by the genetic formula RRSSTT and the contrasted white race by rrsstt. The Fi of a cross between two such races is of a pale red color intermediate between the parental red and white, and in F2 all shades of red are found from very pale to about the same depth of color as the parent 138 GENETICS IN RELATION TO AGRICULTURE race. In the actual experiment among seven families comprising a total of 440 plants only one produced white grains, but the F3 generation demonstrated the adequacy of the three-factor analysis. The inter- : I 1 I tit t it":? - J- s t § J- I S 1 I 1 ' I 1 ^ T >^ I I 5^ t 4 § t 1 S 4- t S t t § t 1 s ill J- 4 § ill i s 1 X ^ 1 ^ 4 -L 1 ^ 1 -. '- L 1 i. J S ^ ^,^ I S 1. s ■ --^ t^ 1 t s+l i / t_ .- 1^1^ 1 {)<' wfwv § ^ ' ' ' 1 ^ ^ - ' ^ T~- ■ " r ■ '^ ~ u- ^ n ^~ •'— 7 r •• ; .1 ^ ; V ;i , F C ! , Hi \ 11 '/ -^ rr -' r \ T" -^\ i - y '.\ . ':^ \ JA J i ' \ T\ 7 , \ 1 ' ■ ' , \ \ / It 1 ! ; ^ t T ' ■ "1 v i \ --.-• ■ :Jr-] \: \ \ '.'- 1 \ ^^ ]:--;■' >-'T / 1 ■ ' i ^^ -' |\ ' ■ ' I ,^' |! \/ ' * ^_j^N. ^ 1 — ^ "?" 1 'in ' -F^; ^< N^^^^ ':;}: ! 1 i ■ ' 4^v 1 ',T y' 1 ^,. f ' \.y( y 1 ^v \ < y ^v 1 EI / ■^ ^/r' 1 1 1 1 1 . 1 1^ ■ jp t _ ■ ■ UTLT Fo t ' .'■K ■'' ~ 1 ', >L 1 I ^^ I i N t IN i 4 I^i J- 4 ^>. lis ^ t 4 s .076-68.528 250-73.860 .340-73.096 .028 - 83.036 .020-86.880 .616-90.228 .716-93.030 .320-95.436 .428-97.396 .140-93.760 .256-99.628 .876 - 100.00 100.090 Fig. 62. — Resume of ratios found in 142 families in the first five generations following the cross;between Bursa bursa-pastoris and B. Heegeri. Each square represents a possible family, the position of a family being determined by the percentage of plants with tri- angular capsules as indicated at the base of the figure. {After Shull.) mediate shade" of red produced in Fi and the varying shades produced in segregation depend on the cumulative effect of the color factors. In- THE NATURE AND EXPRESSION OF MENDELIAN FACTORS 139 stead of displaying complete dominance for any one member of the factor system as Shiill found for the triangular capsule factors in Bursa, the factors here have a certain effect in color production which is additive, RrSsTt RrSsTt RrSsTt RrSsTt RrSsTt rrSsTt RrSsTt RrSSTt rrSsTt RrSsTt RrSSTt rrSsTt RrSsTt RrSSTt rrSsTt rrSsTT RrSSTt RrssTt rrSsTT RrSsTT RrssTt rrSSTt RrSsTT RrssTt rrSSTt RrSsTT RrssTt RRssTt RrSsTT RrSstt RRssTt RRSsTt rrssTt RrSstt RrSStt RRSsTt RrSSTT rrssTt RrSstt RrSStt RRSsTt RrSSTT rrSstt RrSstt RrssTT RRSsTt RRSsTT rrSstt rrssTT RrssTT rrSSTT RRSsTT Rrsstt rrSStt RRSstt RRssTT RRSSTt rrsslt Rrsstt RRsstt RRSstt RRSStt RRSSTt RRSSTT Fig. G3.—-F2 squares of the checkerboard of a cross of red (RRSSTT) X white (rrsstt) wheat arranged in classes according to the depth of color displayed by the phenotypes. i.e., two factors produce twice the depth of red coloration in the grain that one produces and all six are necessary for the production of the 140 GENETICS IN RELATION TO AGRICULTURE full color of the parent red wheat. Consequently there are six shades of red in an F^ population possessing various frequencies with respect to the proportionate number of individuals which display a particular shade of color as shown in the foregoing diagram (Fig. 63). Factors which display summation effects have been conveniently called cumu- lative factors. Besides dominant factors which produce similar or identical somatic effects a large number of recessive factors are known which display the same phenomena. The first example of this type which was worked out was that in sweet peas described by Bateson. In sweet peas there are a number of different whites which phenotypically cannot be distin- guished from one another. The fact that they are genetically different is shown when they are crossed together, for then instead of producing white sweet peas the Fi plants bear colored flowers, the particular color depending upon the genetic constitutions of the whites which were crossed. Since the simultaneous action of two dominant factors, neither one of which by itself can produce any color, is necessary for color pro- duction, Bateson has proposed to call such factors complementary factors. The same relations have been found to exist in the production of aleurone color in grains of corn. Certain white varieties of corn are known which when crossed together give red or purple corn according to the genetic constitutions of the races which were crossed. As with dominant duplicate factors this sort of phenomenon gives peculiar Mendelian ratios in F^ because of the fact that many of the genotypes are indistinguishable phenotypically. Thus for example we may repre- sent a purple corn by the formula CCPP, these factors being particularly concerned in the production of aleurone color. A mutation in the locus C would give a white corn of the genetic constitution ccPP, and likewise a mutation in the locus P would give a white corn of the genetic constitution CCpp. Phenotypically these two varieties of white corn are indistinguishable, but from a genotypic standpoint the factors for white are located in different chromosomes in the two varieties. Accord- ingly when two such white varieties are crossed, the Fi is of the genetic constitution CcPp. Since C and P are both completely dominant over their allelomorphs c and p such a corn will be purple because the com- plete set of factors necessary for the production of purple aleurone color has been brought together by crossing these two genetically different whites. The checkerboard for the F2 of such a cross is shown in Fig. 64. It will be observed that the phenotypic ratio in F2 is 9 purple:? white. This is merely a modification of the typical 9:3:3:1 F2 ratio, for in this cross the last three classes are phenotypically alike, although geno- THE NATURE AND EXPRESSION OF MENDELIAN FACTORS 141 typically different. Of the nine purples, only one breeds true in Fz, and of the remaining eight purples, four give families which segregate in the ratio of 3 purple : 1 white, and four give families showing segrega- tion in the ratio of 9 purple: 7 white. All the whites, although of dijfferent genotypes, produce entirely white families. All these relationships are shown clearly in the checkerboard. In Drosophila a large number of similar cases of like somatic effect have been found to be dependent upon different factors. Here the linkage values of the different factors with other factors have been determined very precisely, and moreover the mutants have for the most part arisen directly from the cultures, so that the relationships have been established much better than in any other form. CP Cp cP cp CCPP Purple 1:0 CCPp Purple 3:1 CcPP Purple 3:1 CcPp Purple 9:7 CCPp Purple 3:1 CCpp White 0:1 CcPp Purple 9:7 Ccpp White 0:1 CcPP Purple 3:1 CcPp Purple 9:7 ccPP White 0:1 ccPp White 0:1 CcPp Purple 9:7 Ccpp White 0:1 ccPp White 0:1 ccpp White 0:1 CP Cp cP cp Fig. 64. — Checkerboard of F2 of cross white (ccPP) X white (CCpp) maize, showing phenotypes and F3 segregation as well as genotypes. For body color at least three similar mutant factors result in almost identical darker forms. The first of these to be discovered was the black factor which is located in the second group of factors. The factor for ebony body color is in the third group, and sable is a sex-linked factor. Although so nearly alike that a mixed population could not be certainly classified these particular races do show slight differences in coloration. Similarly nearly identical results are obtained from three different jaunty factors which cause the wings to turn up at the ends. Morgan has also pointed out other such similarities in effect of different factors which affect eye and wing characters, color, etc. Sometimes a dominant and a recessive factor give identical pheno- typic results. For an illustration of this we may again turn to aleurone 142 GENETICS IN RELATION TO AGRICULTURE color in corn. Taking into account the white dominant factor for aleurone coloration, the following genotypes may be obtained: WWccPP = white wwCCPP = purple wwccPP = white wwCCpp = white wwccpp = white WCP WCp WcP Wcp wCP wCp U'CP WCP WCp WcP Wcp wCP wCp wcP wcP WWCCPP White 0:1 WWCCPp White 0:1 WWCcPP White 0:1 WWCcPp White 0:1 WwCCPP White 1:3 WwCCPp White 3:13 WwCcPP White 3:13 WwCcPp White 9:55 WWCCPp White 0:1 WWCCpp White 0:1 WWCcPp White 0:1 WWCcpp White 0:1 WwCCPp White 3:13 WwCCpp White 0:1 WwCcPp White 9:55 WwCcpp White 0:1 WWCcPP WWCcPp White White 0:1 0:1 WWccPP WWccPp White White 0:1 0:1 WwCcPP White 3:13 WwCcPp White 9:55 WwccPP White 0:1 WwccPp White 0:1 WWCcPp ' WWCcpp White ! White 0:1 1 0:1 WWccPp White 0:1 WWccpp : WwCcPp White \ White 0:1 9:55 WwCcpp White 0:1 WwccPp White 0:1 Wwccpp White 0:1 WwCCPP White 1:3 WwCCPp White 3:13 WwCcPP White 3:13 WwCcPp White 9:55 wwCCPP Purple 1:0 wwCCPp Purple 3:1 wwCcPP Purple 3:1 wwCcPp Purple 9:7 WwCCPp White 3:13 WwCCpp White 0:1 WwCcPp White 9:55 WwCcpp White 0:1 wwCCPp Purple 3:1 wwCCpp White 0:1 wwCcPp Purple 9:7 wwCcpp White 0:1 WwCcPP White 3:13 WwCcPp White 9:55 WwccPP White 0:1 WwccPp White 0:1 wwCcPP Purple 3:1. wwCcPp Purple 9:7 wwccPP White 0:1 wwccPp White 0:1 WwCcPp White 9:55 WwCcpp White 0:1 WxvccPp White 0:1 Wwccpp White 0:1 wwCcPp Purple 9:7 wwCcpp White 0:1 wwccPp White 0:1 wwccpp White 0:1 Fig. 65.- -Fi checkerboard for cross of white (WWCCPP) X white (wwccpp) corn. In the Fs segregation ratio the purple is given first as in Fig. 64. The student will be able to figure out many different relations which exist when such races are crossed. In this section only one will be con- sidered as an illustration of the working of such a sj^stem. If a white corn, WWCCPP, is crossed with a white corn, wwccpp, the Fi is of the genetic constitution, WwCcPp, and is white on account of the action of W. The F2, however, shows some purple grains as will become apparent from a study of the accompanying checkerboard, Fig. 65. In F2 such a hybrid segregates in the ratio 55 white : 9 purple, and in F^ the families show the segregation ratios indicated in the proper squares of the checker- THE NATURE AND EXPRESSION OF MENDELIAN FACTORS 143 board. As in the previous instance these ratios are merely modi- fications of the typical Mendelian dihybrid and trihybrid ratios due to the fact that many of the classes are white and hence are merged into one. It should be apparent from the discussion in this chapter that many complex relations exist as respects the nature and expression of factors. Only some of the best established and most conspicuous cases have been discussed and some of these in rather incomplete fashion, but the material presented is sufficient to establish several facts concerning factors, namely that some factors have very minute, others very far reaching effects, that factors may affect many characters in the individual, that factors may vary in their expression in individuals, that sometimes this variability in factor expression is dependent upon definite environmental conditions and sometimes on obscure or unknown causes, and that at times different factors may have similar somatic expressions. It is difficult to treat such various matters in any systematic fashion, con- sequently this chapter must be regarded merely as an introduction to the general topic of factor interactions. CHAPTER VIII ALLELOMORPHIC RELATIONSHIPS IN MENDELISM The present chapter is designed to deal with those relationships in which a single locus in the hereditary system is involved. Mendel worked with seven pairs of contrasted characters and he ob- served that in all of these one member of the pair controlled the expres- sion of the character when the individual was heterozygous. When tall peas are crossed with dwarf the hybrid is tall, in fact slightly taller even than the tall parent. Similarly yellow cotyledons are dominant over green, and smooth over wrinkled seed. The same is true for the other four pairs of characters. So important did this fact of dominance appear to investigators that for some time after the rediscovery of Mendelism reference was very- generally made to the law of dominance, and great significance was attached to any failure to observe dominance in genetic investigations. But subsequent investigations have shown that domi- nance, far from being a general rule, is merely a special condition met with in certain cases of inheritance. That it is by no means universal must be conceded. How far it obtains and what other conditions are met with in its absence, we shall endeavor to show in what follows. Dominance is a relation existing between a factor and its allelo- morph such that in plants heterozygous for the factor in question the character expression is the same or approximately the same as that when the factor is homozygous. Dominance, therefore, applies only to rela- tions existing between a pair of factors. That two contrasted characters show an intermediate condition is no evidence in itself that dominance is lacking. It must further be demonstrated that this condition is due to the fact that the character expression of a genotype Aa lies between that oi AA and aa. Otherwise the intermediate expression of the hybrid character may be the expression merely of the action of several pairs of factors each displaying dominance for one member of each pair, but together giving an intermediate expression. The Extent of Dominance. — Off hand it would appear that com- plete dominance is a very common phenomenon in genetic investigations. The seven pairs of contrasted characters in peas could hardly have dis- played it in all the pairs unless it were a condition of wide occurrence and considerable significance. Otherwise we should have to consider this a remarkable case of coincidence. Likewise the oft-cited investi- 144 ALLELOMORPHIC RELATIONSHIPS IN MENDELISM 145 gations with Drosophila indicate that usually a normal allelomorph is dom- inant to a mutant factor, and in facti often to the eye completely dominant. More precise investigations indicate, however, that al- though for all practical purposes dominance often is so complete as to closely approximate the expres- sion of the homozygous character due to the duplex ' condition of the dominant factor, still the completeness of dominance is often more apparent than real. Darbishire has attacked this problem in the case of the cross smooth as con- trasted with wrinkled peas. Mendel's experiments showed that smooth or round shape is dominant over the wrinkled shape in peas and as in other cases the dominance appears to the eye complete. Darbi- shire investigated the cause of the difference between round and wrinkled peas and found it associated with a difference in starch content. Thus during the development of the seed in those races possessing round seeds the sugar is almost wholly converted into starch so that when the seed is ripe and drying it retains water rather 10 146 GENETICS IN RELATION TO AGRICULTURE firmly and shrinks uniformly to form a round seed. Like the seeds of round races those of wrinkled peas are also round at the height of development, but in peas of such varieties the sugar is very incompletely transformed into starch. Consequently in ripening and drying they give up more water proportionally, than round races and do not shrink uniformly. As a result they become very much wrinkled at maturity. This difference in the starch grains of the wrinkled pea is not only a matter of less complete trans- formation of sugar into starch, but is also associated with less perfect production of starch grains as shown in Fig. 66. Thus in the round races the starch grains are numerous and are large and entire. They show practically no subdivision of the grains. But in the wrinkled peas the grains are not only less numerous, but they show fissures which give them an appearance like that of the compound starch grains of some species of plants. This appearance is probably due to the fact that actual breaking down of starch grains occurs in wrinkled peas during ripening so that the grains remaining are in a partial stage of disintegration. In the hybrid between a round and a wrinkled pea, however, the condition of the starch grains is intermediate between that of the two parents. The grains are intermediate not only in number and shape but also in the degree of disintegration they display. In the contrasted pair of characters, round vs. wrinkled seed in peas, the dominance of round is, therefore, merely a superficial character expression. Actually the basic phenomena involved, i.e., the transformation of sugar into starch, show an intermediate condition in the hybrid. The superficial character ex- pression of this intermediate condition happens to be the same as that of the strict parental round condition, so that dominance here is merely dependent on superficial resemblance. We may well hesitate, therefore, in our judgment as to the completeness of dominance in any case until it has been examined with considerable care. Sometimes the application of more precise character measurements will suffice to detect a difference between the homozygous and hetero- zygous character expression. This is shown for the case of miniature vs. long wings in Drosophila. In miniature- winged flies the wings reach about to the tip of the abdomen, whereas in the long-winged flies they extend considerably beyond the abdomen. The long-winged condition is dominant, to the eye completely, and there is absolutely no difficulty in segregating the long- winged flies of an F 2 population from those which have miniature wings. Nevertheless Lutz has shown that when biomet- rical methods are employed the length of wings of heterozygous flies com- pared with the length of legs is shorter than that for flies homozygous for the long-winged factor. The difference in character expression in this case is slight but it can be demonstrated by the employment of precise methods of measurement. ALLELOMORPHIC RELATIONSHIPS IN MENDELISM 147 Intermediate Expression in the Hybrid. — From those cases in which dominance is nearly or quite complete we may next pass to those in which the character expression of Aa is intermediate to that of AA and aa. There are numberless instances of this kind, and they are of interest because the heterozygous class may be distinguished in F2, so that the typical ratio obtained is lA:2Aa: la, instead of 3^ :1a as in cases where dominance occurs. For a concrete example we may turn to Baur's case in the snap- dragon. Baur and Miss Wheldale have independently conducted very extensive investigations of Mendelian inheritance in Antirrhinum. For most cases one member of a pair of contrasted characters is dominant, but when ivory is crossed with red the Fi is intermediate in color, it is pale red or pink. When F2 is grown it is found to consist of 1 red: 2 pink: 1 ivory. In one case among 97 plants, Baur obtained 22 red, RR X rr Red Ivory Rr Pink 1 RR 2 Rr 1 7T Red Pink Ivory RR 1 RR 2Rr 1 rr rr Red Red Pink Ivory Ivory Fig. 67. — Results of crossing snapdragons with red and ivory colored flowers. 52 pink, and 23 ivory, a satisfactory agreement with Mendelian ex- pectations. The actual proof for this case comes out in growing F3. When this is done it is found that the red plants and the ivory plants give progeny which are entirely red and ivory, respectively. The pink plants on the other hand are all heterozygous and they give in 7^3 and in all succeeding generations plants in the proportion of 1 red: 2 pink:l ivory. The case is very evidently one in which a single factor difference is concerned. If the factor responsible for the production of red in Antirrhinum be designated by R, then we may designate its allelomorph present in the ivory race by r. The case then works out according to the diagram in Fig. 67. In the Four o'clock, Mirdbilis jalapa, it appears to be the rule that heterozygous plants present visible differences from plants homozygous for color factors. For this reason in breeding experiments this plant gives a rather remarkable diversity of colors with relatively few factors involved. Thus we may start with the primary assumption that in one series of colors we have involved two pairs of factors as follows: Y = factor for yellow colored sap. R = factor which turns yellow sap red. 148 GENETICS IN RELATION TO AGRICULTURE The various homozygous combinations of these two factors give four primary races which breed true as follows: YYRR = crimson. YYrr = yellow. yyRR = white. yyrr = white. By hybridizing these races four heterozygous forms may be produced which are of the colors given below: YYRR (crimson) X YYrr (yellow) gives YYRr = orange red. YYRR (crimson) X yyRR (white) gives YyRR = magenta. YYRR (crimson) X yyrr (white) gives YyRr = magenta-rose. YYrr (yellow) X yyrr (white) gives Yyrr = pale yellow. cT YR Yr yR yr 9 YR YYRR Crimson YYRr Orange red YyRR Magenta YyRr Magenta rose Yr YYRr Orange red YYrr Yellow YyRr Magenta rose Yyrr Pale yellow yR YyRR Magenta YyRr Magenta rose yyRR White yyRr White yr YyRr Magenta rose Yyrr Pale yellow yyRr White White . Fig. 68. — Checkerboard analysis of the progeny of a magenta-rose Mirabilis of the genetic constitution YyRr. We thus have seven distinct color classes as a result of various com- binations of two pairs of color factors. Moreover, this species gives a very good example of the diversity which may be obtained in an F^ population. Thus Miss Marryat has shown that when magenta-rose, YyRr, is selfed, the progeny fulfil the conditions indicated by the accompanying checkerboard analysis in Fig. 68. Table XXIX. — F2 Phenotypes and F3 Phenotypic Ratios Derived prom THE Original Cross, Crimson, YYRR X White, yyrr Color of parent Number of plants selfed Number of offspring Color of offspring Yellow 2 2 3 4 3 5 26 23 61 64 46 70 All yellow. All crimson. Crimson Orange red 17 crimson : 31 orange red : 15 white. 18 crimson : 32 magenta : 14 white. 9 yellow : 25 pale yellow : 12 white. 5 crimson : 9 magenta : 6 orange red : 19 magenta-rose : 3 yellow : 7 pale yellow : 21 white. Magenta Pale yellow Magenta-rose ALLELOMORPHIC RELATIONSHIPS IN MEN DELI SM 149 When the Fa was grown from such an F2 population Miss Marryat obtained excellent agreement with this analysis as is shown by the data in Table XXIX. Variable Character Expression in the Hybrid.^ — ^Sometimes the character expression in Fi while intermediate displays a range of varia- tion extending almost from one parent to the other. This is shown rather strikingly in the case of bar eyes in Drosophila (Fig. 69) . The bar eye factor is a sex-linked mutant factor which is responsible for the pro- duction of flies with long narrow eyes instead of the round eyes normal for the species. When a female with bar eyes is crossed to a normal male the F: all have bar eyes. In the males especially the eyes are Fig. 69. -Normal (a, o') and bar eye (b, b') of Drosophila; shown in side view and as seen from above. (After Morgan.) just as narrow as in homozygous races, but among the females some may be found which have eyes nearly as narrow as those characteristic of homozygous bar eye flies and others which have eyes nearly as round as those characteristic of the normal fly. Most of them, however, have eyes which display an intermediate effect of the factor. This case readily admits of explanation, if the genetic phenomena involved are considered. Since the factor for bar eyes is sex-linked we may represent the bar-eyed female as {B'X){B'X), following Morgan in employing the primed symbol to indicate a dominant mutant factor. The male with normal eyes is then (b'X) Y. When a bar-eyed female is mated to a normal male, bar-eyed females and males are obtained in Fi as shown in the diagram in Fig. 70. The Fi bar-eyed male obtains his only X-chromosome from the female and this chromosome contains the factor for bar eyes. He has exactly 150 GENETICS IN RELATION TO AGRICULTURE the same genetic constitution, therefore, as a male of a pure bar-eyed race, and it is to be expected that he will display the character to the same extent as a male from a pure race. On the other hand the female has one X-chromosome which bears the normal recessive allelomorph of the bar-eye factor. This factor may be considered as exerting a competitive influence against the bar-eye factor of the other X-chromo- Pi Gametes Bar-eyed 9 {B'X){B'X) (B'X) X Normal cf' {b'X)Y (b'X) Fi (B'X) ih'X)-^ ^(5'Z) Y Bar-eyed 9 Bar-eyed cT Fig. 70. — Results of mating bar-eyed 9 with normal-eyed cf Drosophila. some, so that the character expression in a sense depends upon a variable equilibrium reached between the two factors. Since they appear to be nearly equal in potency it is possible apparently for this equilibrium to be thrown so much to one side or the other that at times the character expression approaches that of the typical bar-eyed strains and at times that of the normal round-eyed flies. Fig. 71. — Longitudinal sections of corn grains showing differences in character of starch; left, floury; right, flinty. An interesting case which throws considerable light on the competi- tive action of factors in determining character expression has been reported by Hayes and East in maize. Flint races of maize are char- acterized by the production of a very small amount of soft starch in the center of the seed and a large amount of hard corneous starch sur- rounding it. Flour corns on the other hand produce grains the endo- sperm of which is almost wholly made up of soft starch with occasionally a very thin layer of corneous starch at the exterior of the endosperm. These differences are shown in Fig. 71. ALLELOMORPHIC RELATIONSHIPS IN MEN DELI SM 151 When a floury corn is pollinated by a flinty corn the grains which result show no effect of the flinty pollination, they are floury grains of the same character as those of a pure floury race. Similarly when a flinty corn is pollinated by a floury corn, the grains are flinty. Again they are of the same character as the maternal parent. The maternal type of grains is always produced in such reciprocal crosses. Following up this experiment, when Fi corneous grains of the cross corneous 9 X floury cf are grown and selfed, the ears produced show distinct segre- gation into flinty and floury corn in the ratio 1 flinty : 1 floury. Fi floury grains from floury 9 X flinty cf when grown and selfed likewise pro- duce ears showing distinct segregation into 1 flinty : 1 floury. Evidently the Fi grains although different phenotypically display the same genetic phenomena. Cytological research has shown that in the fertilization of maize and other plants there is a double fertilization, one fertihzation giving rise to the embryo and the other to the endosperm. In the case of the embryo, an egg nucleus unites with a nucleus from the pollen grain and from this fusion the embryo develops. In the fertilization which gives rise to the endosperm two nuclei from the female unite with one from the male, so that the cells of the endosperm contain 3a; chromo- somes rather than the duplex number characteristic of the cells of the embryo. If the flinty factor be represented by F, and the contrasted factor for floury by /, the zygote of a flinty corn is FF, but the endosperm connected with it is FFF. Correspondingly for the floury race the zygote is //, and its endosperm ///. In the fertilization of flinty by floury corn, the egg nucleus proper, the genetic constitution of which is F, is fertil- ized by an / pollen grain, giving a hybrid zygote of the constitution Fj. The endosperm which surrounds this embryo, however, arises from the fusion of the two endosperm nuclei, FF, with a single nucleus from the pollen grain, giving a zygote of the constitution FFf. This endosperm is flinty because two doses of F are apparently dominant to one dose of /. On the other hand, when floury corn is pollinated by flinty, the embryo has the same genetic constitution, namely FJ, but the endosperm sur- rounding it arose by union of two endosperm nuclei // with a pollen nucleus bearing the factor F. It, therefore, has the genetic constitution jJF and it is floury because the two doses of / determine the phenotypic expression to the exclusion of the single dose of F. In F2 the hybrid flinty grains from the cross flinty 9 X floury cT give exactly the same results as the hybrid floury grains from the cross floury 9 X flinty 6^. Here the ratio is 1 flinty: 1 floury in each case, and half the members of each class are heterozygous and will reproduce the same ratios in the succeeding generation. It would be difficult to conceive of a more beautiful illustration of 152 GENETICS IN RELATION TO AGRICULTURE the quantitative relations obtaining in the determination of dominance. Apparently the relations are about the same as those shown in the ease of bar eye in Drosophila, for conceivably, if such a thing could be obtained, an endosperm arising from an Ff cell might show the same variation between flinty and floury that is shown in the bar-eye character of flies of the genetic constitution (B'X) (h'X). Mosaic Expression of the Hybrid Character. — ^Another type of hybrid condition is that in which the Aa individuals are a mosaic of the char- acters of the two parents. This condition is very strikingly illustrated in Blue Andalusian fowls. Andalusian fowls are of three types: black, splashed white, and the so-called blue. Of these types the black and splashed white breed true, but the blue is a hybrid and constantly segregates in the ratio 1 black : 2 blue : 1 splashed white. When black and splashed white are mated, the progeny are all blue. The Blue Andalusian fowl of the Poultry Standard of Perfection is, there- fore, a heterozygous form and for that reason all attempts to establish it as a pure breeding race have failed. The case, however, is of interest here because the Blue Andalusian is a peculiar mosaic of the characters exhibited by the black and splashed white. Its "blue" color is simply due to a fine but uneven sprinkling of black pigment through the feathers ; and on some portions as for instance the feathers of the breast, the black is present as a. distinct edging or lacing of the feathers. Similar mosiac hybrids which represent a simple heterozygous con- dition have been reported by Nabours in grouse locusts of the genus Parattetix. Nabours found nine distinct races which bred true for particular color patterns. Hybrids, however, between any two of these species display the entire color pattern of both parents, the color patterns being merely superimposed one upon another and in such a manner that the entire pigmentation of both parents is present in the hybrid and is distributed in the same fashion. If then two races of Parattetix A and B be crossed, the hybrid AB will be a mosiac of the two parents, and it is possible by simple inspection of such a hybrid form to determine what races entered into it. Such a hybrid will give a population consisting of 1A:2AB:1B, thus demonstrating that the case rests on a simple factor basis and that the mosaic pattern is simply an expression of a heterozygous condition in which both A and a, if we designate them thus, work out their full possibility in the development of the hybrid. In certain cases which did not appear to conform to this simple interpreta- tion, a microscopic examination was resorted to. This examination dem- onstrated that the lack of agreement was apparent rather than real. Thus in Fig. 72 the superficial characters of the hybrid (BI) between P. leuconotus (BB) and P. nigronotatus (II) are for the most part those of P. leuconotus except for the broad black band across the pronotum which is ALLELOMORPHIC RELATIONSHIPS IN MENDELISM 153 clearly derived fromP. nigronotatus. In the posterior part of the pronotum particularly the characters of P. leuconotus, appear to be dominant but the microscopic study showed clearly that this was due to differences in distribution in the two parents, and that the characters of P. nigronotatus, although obscured were as much present as those of leuconotus. Fig. 72. — Three types of Paratettix, BB, CC, II, and two of the hybrids between them. (After Nabours.) The Presence and Absence Hypothesis. — The foregoing accounts of the relations existing in the expression of the hybrid characters as compared with the two parental characters serves as an adequate introduction for a brief consideration of the presence and absence hypothesis. Accord- ing to the presence and absence hypothesis as advanced by Bateson and Punnett, the only relations which can exist with respect to a certain fac- tor depend on its presence or absence from the hereditary material. Thus if we consider the factor R for round shape in peas, and its allelo- morph r for wrinkled shape, according to the presence and absence hypothesis the r of the genetic formula of the wrinkled pea is not itself a factor as we have assumed throughout the discussion in this text, but merely represents the absence of the factor R. The wrinkled character, therefore, is merely an expression of the action of the set of genetic fac- tors in peas when the factor R has been taken away from the system. In this text we have throughout assumed that the recessive symbols stand for factors just as truly as do the dominant ones, and we have regarded the difference between a recessive factor and its corresponding dominant allelomorph as dependent upon some change in a dominant factor sometimes profound and sometimes less profound so that all 154 GENETICS IN RELATION TO AGRICULTURE variations from complete dominance to a strict intermediacy may be ob- tained among hybrids. For cases of complete dominance, the presence and absence idea satisJ&es conditions very satisfactorily as far as formal relations are concerned, and intermediacy and even other conditions of the hybrid expression may be assumed to depend upon the quantitative difference in the amount of the factor present in the hybrid race as contrasted with the parent races. Difficulties, however, begin to arise when attempts are made to explain the origin of dominant mutations in terms of this hypothesis, for in such cases it is almost necessary to assume that a factor has been added to the hereditary material. It is usually considered easy enough to account for a recessive mutation as due to the dropping out of a factor from the hereditary material, but when a factor is added to that material, we must ask from whence it came, what its nature, etc. If we regard mutations as simply due to changes in a fac- tor this difficulty vanishes for then dominance or recessiveness of the mutations depends merely on the relations between the mutated factor and its unchanged condition and there is no particular reason for as- suming that all mutations should be of the nature of "loss" mutations, i.e., mutations depending upon the loss of a factor from the hereditary material and resulting in the absence of some dominant character in the individuals concerned. There is no difficulty therefore, in account- ing for the four or five dominant mutations which have been observed in Drosophila, if we regard mutation as a change in a locus, for these par- ticular mutations simply happened to involve changes of such a type that the mutated locus was dominant to the unmutated condition. Obviously, also, such a view conforms more closely with the facts ob- served in cases of the competitive action of factors such as is seen in bar eyes in Drosophila or in the factors for flinty and floury endosperm in maize. But there are more serious objections than these which can be raised against the presence and absence hypothesis. In Drosophila, for in- stance, a number of cases of return mutations have been observed, many of them in cultures so controlled that the possibility of explaining them by chance contamination is practically precluded. Thus in stock so controlled by the presence of other factors that it would practically have been impossible to have a contamination go unnoticed on account of the introduction of other factors, the bar-eyed race of Drosophila has been known to produce normal-eyed mutants (May) and eosin-eyed flies have been observed to give white-eyed flies on several occasions; while on the other hand eosin, although dominant to white, originally arose as a mutant in a stock of white-eyed flies. If we assume that the change from eosin to white involves a relatively unessential change in the W factor in Drosophila, in chemical terms perhaps a slight rearrangement in ALLELOMORPH IC RELATIONSHIPS IN MEN DELI SM 155 the. molecule or a change in an end radical, then it is not difficult to imagine how a reverse mutation might arise. Reverse mutations, there- fore, support the idea that the recessive member of an allelomorphic system is just as truly a factor as the dominant member. Never- theless these considerations do not in themselves confute the argument of presence and absence, although they tend to throw the weight of evidence strongly against it. It is, however, perhaps not amiss to point out that much of the weight of authority of the presence and absence hypothesis depends on the fact that it was advanced at the psycho- logical moment, and that, as Morgan points out, in the light of our present knowledge of the relation between factors and characters it assumes a knowledge far beyond that which we have at present attained. But the really serious objections to the hypothesis are those based on the evidence furnished by multiple allelomorphism. Since the foregoing was written Bridges has published results of his investigation of a case of loss or inactivation of a portion of the X-chro- mosome in Drosophila. The deficient section involved the factor for bar eye. As Bridges points out this constitutes the first valid evidence upon the question of presence and absence. According to the presence and absence hypothesis the original appearance of the dominant bar character was due to the loss from the chromosome of an inhibitor, thereby allowing the normal narrowing effect of the remaining complex to assert itself. It should make no difference whether this inhibitor were lost by a special loss involving only the inhibitor or whether it were lost because of being situated in a particular section which became lost. In other words, the chromosome which is deficient for the region carrying the inhibitor should allow the occurrence of the same narrowing effect that is allowed by the simple loss of the inhibitor. In point of fact, the deficiency of the region in which the inhibitor must be hypoth- ecated does not produce an effect like that of the mutation responsible for bar. For, the female carrying one deficient X and one normal X shows no narrowing of the eye shape, and likewise the female carrying one deficient A' and one bar X is no narrower in eye shape than a normal heterozygous bar. Thus, in the only case which has a direct bearing on the presence and absence hypothesis, it is seen that the ex- pedient of the loss of inhibitors to explain the origin of a dominant mutation is of no avail. Multiple Allelomorphism in General. — Multiple allelomorphism is the term applied to those cases which seem to depend on a series of changes in a given factor locus. Cuenot advanced such an explanation for the inheritance of certain color patterns in mice, and Morgan has since described several cases which occur in Drosophila. Since these later cases are simpler and have been worked out in more detail they will be treated first. 156 GENETICS IN RELATION TO AGRICULTURE Multiple Allelomorphism in Drosophila. — A typical case is that centering around the locus for eye color in Drosophila which we have called W. This locus is situated in the X-chromosome at a distance of one unit from the locus Y for body color. The first mutations in Dro- sophila involved a change in W such that white eyes were produced, a mutation recessive to the normal red-eyed condition. This factor is called w and its inheritance has been dealt with in previous chapters. Later some flies arose in a white-eyed culture which had eosin eyes. When a white cf is mated to an eosin 9 the Fi is eosin ^ and F2 consists of 3 eosin: 1 white. When a red-eyed 9 is mated to an eosin-eyed cf, Fi is red, and F^ segregates in the ratio 3 red:l eosin. The facts are explainable on the assumption that the factor W has been changed in a different fashion to produce the factor for eosin which we designated as w". On this basis the analysis of the genetic constitutions of these different races is as follows : {WX){WX) = red 9 {WX)Y = red & {iv''X){w'X) = eosin 9 {:ufX)Y = eosin cf {wX){wX) = white 9 {wX)Y = white d^. A change in the same locus has occurred in the mutation to white and to eosin, but the change has been different in each case. Later four other changes in this locus occurred giving eye colors which have been named cherry, tinged, blood and buff, and these fulfil the same conditions as those pointed out for eosin. The factors are designated w% w^, w^ and w'"^ respectively. These seven factors therefore display a particular type of behavior depending upon the fact that they occupy the same locus in the X-chromosome. They form together a system of septuple allelomorphs. In Drosophila there are at least three other such systems of multiple allelomorphs. One of these centers around the Y locus in the X-chromo- some which may change to y giving a yellow-bodied fly in place of the normal gray body or may change to y' when a spot-bodied fly is produced. Another system of triple allelomorphs for eye color is located in the third chromosome; it consists of the factors for pink and peach eye color, and the normal allelomorph of these which is concerned in the production of red eyes. A fourth such series of allelomorphs is that of the factors for ebony and sooty body color and their normal allelo- morph concerned in the production of gray body color. This series is also located in the third chromosome. Assuming that more than two factors may occupy identical loci in homologous chromosomes there are several simple relations which must be fulfilled in order to establish the case experimentally. The 1 The Fi 9s actually have an intermediate eye-color, "white-eosin compound". ALLELOMORPHIC RELATIONSHIPS IN MENDELISM 157 linkage values of such a series of allelomorphs when tested with other members of the group to which they belong should be identical. The factor for yellow body color is located at the locus 0.0 in the X-chromo- FiG. 73. — Forms and hybrids of Paratettix. AA, texanus; BB, leucorotus; CC, leuco- thorax; II, nigra notalus. (After Nabours.) somes, and displays definite linkage values when tested with any other factor belonging in this chromosome. The factor for spot gives exactly the same values with all factors with which it has been tested. The factors for eosin and white eye color both give one unit of crossing over 158 GENETICS IN RELATION TO AGRICULTURE with the factor for yellow body color and they give identical linkage values with the other factors in this group. Since the factors occupy identical loci in the homologus chromosomes not more than two can occur in the same individual at the same time. This fact was demon- strated in the breeding tests applied above. Other cases of multiple allelomorphism are known to occur in a large variety of species. In the silkworm there is apparently a series of Expectation Actual Numbers 8 « 3 5 t B 1 6 4 • BC 3 3 C 21 C Expectation Actual Numbers 28.25 29 56.5 53 28.25 31 11 B 3 21 B 1 BC ? c Expectation Actual Numbers 7;5 10 1 15 17 7.5 3 9 A 1 1 • AB B 1 Expectation Actual Numbers 107 977 AB B 1 Expectation Actual Numbers 20 21 A 60 59 1 AB 1 B I 231.25 251 B BC 462.5 452 I BC 231.35 222 C I 65 66 130 136 BC 65 1 5 AB AC AC 111.75 117 AC 6 BC 6 AC 3 AD 37.25 32 F2 Fi A— texanus B = leuconotus AB 1 c5 from the field AC 2 o 8 from the field not virgin Parents C = leucothorax D= punciofemorata Fig. 74. -Chart showing results of a continued series of pedigree experiments with Paratettix involving types A, B, and C. (After Nahours.) multiple allelomorphs concerned in the production of larval patterns. As Tanaka has shown there are four larval patterns, moricaud, striped, normal, and plain. Each of these colors is allelomorphic to the other three and moreover they all apparently display the same linkage values with the pair of factors for yellow and white cocoon color. Like the multiple systems in Drosophila they give no new types by recombination when crossed. Multiple Allelomorphs in the Grouse Locust. — A very striking series of ALLELOMORPHIC RELATIONSHIPS IN MEN DELI SM 159 multiple allelomorphs is that concerned with color pattern in Parattetix. Nabours has investigated the inheritance of pattern in fourteen races of this insect, the grouse locust. Some of these are shown in Fig. 73 and also the hybrids between them. It was pointed out in a previous sec- tion in this chapter that these races when hybridized give intermediate forms in Fi, intermediate in the sense that they display the type patterns of both hybrids superimposed one upon the other. In F^ they segregate into three types, the two parent types, and the hybrid form in the, ratio 1:2:1. Nabours has prepared a chart from the data of an extensive breeding experiment with some of these forms. It illustrates so admirably the type of behavior displayed by multiple allelomorphs that it is given in full in Fig. 74. In these experiments separation of B from AB and C from AC has not been attempted because the type A exerts very little influence on the color pattern of the hybrid. In this chart expected results are indicated wherever the ratio of types actually observed is of significance. The observed results show excellent agreement with expectations. The multiple allelomorphs in Parattetix appear to affect the entire color pattern of the body and to cause different colors to develop in different parts of the body. This, however, is merely another instance of the manifold effects of single factors, and furnishes no sound argu- ment against the conception of multiple allelomorphs. Furthermore, Nabours has discovered at least one modifying factor which can exist only with, and in addition to, any of the fourteen multiple allelomorphs or their hybrids. Multiple Allelomorphs in Maize. — In maize there is apparently a remarkable series of multiple allelomorphs concerned in the development of red color in the husks, silk, pericarp, and cob. Practically all com- binations of these are known in various different varieties of maize, so that it is possible to have varieties with red grain, silk, cob, and husk; red grain, white silk, white cob, and white husk; or any other com- bination whatsoever. When, however, such types are crossed the Fi displays a superimposed set of characters, red being dominant; and in Fi but three forms appear in the ratio 1:2:1 as with Nabours' locusts, namely the two parental types and, if it is different from either of them, the hybrid form. This indicates that the Fi hybrids form gametes bearing factors determining only the conditions represented in the parents. This fact Emerson subjected to direct test by crossing Fi hybrids back to varieties lacking the red color in all these parts. In one case an Fi plant produced ears which had red cobs and variegated red grains. When such a plant was crossed back to a race having white cobs and grains, the next generation consisted only of plants which bore 160 GENETICS IN RELATION TO AGRICULTURE ears with white cobs and variegated grains and ears with red cobs and white grains. None were produced which bore ears having the F\ combination, red cob and variegated grains, and on the other hand none were produced showing the reverse recombination, white ears and white grains. This series of multiple allelomorphs is perhaps the most striking one known and displays just as unique relations as does that series in Parat- tetix. For considering only red vs. white alone in these characters there are sixteen possible combinations which would give pure breeding races. Besides this, however, the red, particularly of the pericarp, may be modified in many different ways with respect to shade and distribution, apparently without altering the relations of the factors involved to the allelomorphic system, so that the number of possible combinations is considerably greater. Emerson has studied the in- heritance of a large number of these types and so far they all may be consistently explained on the hypothesis of multiple allelomorphs but the data are not as yet extensive enough to establish this interpretation beyond any doubt. The general nature of multiple allelomorphism is attested to by its occurrence in widely separated species of animals and plants. Its occurrence in Drosophila, the silkworm, Parattetix, and maize has been noted above. Besides these Morgan has pointed out that cases are known in rabbits and mice among animals, and in Aquilegia, Lychnis, and the bean among plants. In rabbits the factors concerned are three, those for self-color, Himalayan pattern, and albinism. In the mouse apparently four factors make up a similar system, namely those for yellow, black, gray, and gray with white belly. In Aquilegia the system has to do with leaf color and three factors are involved, those for green, variegated, and yellow leaf color. Shull's case in Lychnis has to do with sex-determining factors. In the bean the case is somewhat like that in corn but the series is less extensive. The system there as re- ported by Emerson is green leaves, green pods; green leaves, yellow pods; yellow leaves, yellow pods. Morgan has brought together the arguments in favor of multiple allelomorphism and the following discussion is based for the most part upon his presentation. This discussion will serve in a sense as a summary of the material dealing with multiple allelomorphism. 1. Systems of multiple allelomorphs appear always to affect the same character. This fact is readily apparent from a consideration of the cases which have been cited above. Beyond this the cases often give a series of diminishing intensities with respect to the character affected as for example, black, Himalayan, and white in rabbits. On this basis, Pun- nett has sought to disprove the validity of the hypothesis of multiple ALLELOMORPH I C RELATIONSHIPS IN MEN DELI SM 161 allelomorphs as applied to the case in rabbits, for although the homo- zygous forms give such a series of diminishing intensity of melanic pig- ment, nevertheless the heterozygous forms give inconsistences. Black by agouti gives agouti-black, but black by yellow gives full black, in spite of the fact that yellow is regarded as a lower intensity of pigmenta- tion than agouti. The argument does not appear to be valid, however, for specific relations may still exist among the factors of a system of multiple allelo- morphs. Bridges has pointed this out in the case of the eye color series red, white, cherry, eosin, tinged, blood and buff eye-color in Drosophila. He has discovered a number of factors which modify eosin, one in partic- ular called whiting changes eosin to pure white, but does not produce any visible effect on the other members of the series. The conception of diminishing intensity as applied to multiple allelomorphs is clearly not fundamental to the hypothesis. 2. The behavior in inheritance is different from that which would be expected in case different loci in the hereditary system were involved. When different loci are involved, each of two different mutant types will contain besides its own mutant factor the normal allelomorph of the mutant factor of the contrasted type. Consequently on crossing they will unite the series of factors present in the original type and give a character expression corresponding to that of the original form. Such is normally the case in undoubted instances of mutations affecting differ- ent loci, but in the case of multiple allelomorphs one or the other of the mutant types or an intermediate is produced in F\. When identical loci are concerned in two mutations, the hybrid between them will not reconstitute the original system, but will contain only the two mutant factors at that locus. The character expression of the hybrid therefore will depend on the interrelations existing between the mutant factors and the rest of the hereditary system rather than on the reuniting of the normal allelomorphs of the mutant factors. 3. There are difficulties in explaining the origin of some of the forms on the basis of complete linkage between factors, which disappear on the adoption of the hypothesis of multiple allelomorphism. The difficulty may be illustrated by a specific case, that of the series red, white, cherry, eosin, tinged, blood and buff eye-color in Drosophila. Considering two specific instances, cherry and white, both of which arose from red immedi- ately, it must follow on the basis of complete linkage that one differs from red by one factor and the other by two factors. If red be {CE){CE), then cherry, which is recessive to red would be (cE){cE), and white, which is' recessive to both red and cherry would be (ce)(ce). This involves the assumption that white arose as a result of simultaneous mutations in two completely linked factors affecting the same character, a practically 11 162 GENETICS IN RELATION TO AGRICULTURE inconceivable thing, if viewed from a purely mathematical standpoint, unless a special biological mechanism exists which favors such mutations. The same difficulties are met with in the case of other systems of multiple allelomorphs the origin of which have been observed in pedigree cultures, consequently the situation in the above system is not unique. 4. If a curve of linkage values be plotted in Drosophila for a consider- able number of known factors it will be found that the frequencies of different values correspond with one another until those displaying multiple allelomorphism (or complete linkage) are met with and these are far in excess of the number normally to be expected from purely mathematical considerations. They are not, therefore, merely the ex- tremes of ordinary cases of linkage. 5. There are no very good reasons why only one sort of change should be possible in a given locus in the hereditary material. It is true the presence and absence hypothesis does hold that the only difference with respect to a given factor is its presence in the hereditary material or its absence from it, but there are many reasons why this view at present appears untenable. A factor in the hereditary material may well be regarded as a complex chemical substance of some kind which maintains essential relations with the other factors in the system such that to lose it entirely might well disorganize the entire system. But such a com- plex chemical substance might well change in many relatively slight ways which would modify the particular character in which it is concerned in various directions depending upon the specific manner in which the factor has been altered. When these arguments are considered and the type of cases to which it is applied are taken into account, it is apparent that the theory of multiple allelomorphism is a useful analytic tool in the solution of a certain class of pecuUar Mendehan phenomena. Although some of the above cases may prove to be instances of extremely close linkage, never- theless for most of them the case is firmly established experimentally, and deserves careful consideration from that standpoint. CHAPTER IX TYPES OF FACTOR INTERACTIONS In the present chapter it is proposed to ilhistrate some of the various types of relations which exist between different pairs of factors so that the student may come to appreciate some of the more complex features of factor interaction. The System of Aleurone Color Factors in Maize. — It has already been pointed out that some particular cases of Mendelian phenomena depend for their explanation on the presence of different factors which g;ive similar results. These are to be considered as cases of Mendelian factor interactions as specifically as those more fully discussed in this chapter. Such a condition may be illustrated very satisfactorily by the four-factor system for aleurone color which is known to exist in maize. The factors involved in this system and their actions are as follows: C — a factor for chromogen base. C is necessary for the production of any aleurone coloration in maize. Its allelomorph c constantly gives white grains. R — a factor which when present with C gives a red aleurone color. Its allelomorph, r, constantly gives white. P — a factor which when present with C and R gives purple aleurone color. W — a dominant factor for white aleurone color. When it is present the grains will be white no matter what other factors may be present. With this series of factors, the following homozygous races are ob- tainable and have the phenotypic expression here indicated: 1. WWCCRRPP— white. 9. wwCCRRPP— purple. 2. WWCCRRpi}— white. 10. wwCCRRpp— red. 3. WWCCrrPP— white. 11. wwCCrrPP— white. 4. WWCCrrpp— white. 12. WWCCrrpp— white. 5. WWccRRPP—white. 13. wwccRRPP— white. 6. WWccRRpp — white. 14. wwccRRpp — white. 7. WWccrrPP — white. 15. wwccrrPP — white. 8. WWccrrpp — white. 16. wwccrrpp — white. Of the sixteen pure breeding forms, fourteen are whites, and although indistinguishable phenotypically these whites may be separated by proper breeding tests. Some of the relations existing between the different genotypes have already been dealt with, but the student may 163 164 GENETICS IN RELATION TO AGRICULTURE be interested in tracing out others. Thus the presence of a factor for dominant white may be demonstrated by crossing with a purple race, in which case the grains will be white, if such a factor be present. More- over, several of the whites when crossed give colored forms in Fi, thus 11 X 13, 11 X 14, and 12 X 13 give purple, and 12 X 14 gives red. It has already been shown how in case of the presence of the factor for dominant white, whites when crossed may give a white Fi and white, purple, and red in various proportions in F2. Such is the case for ex- FiG. 75. — Comb types in poultry. Single, a; pea, b; rose, c; walnut, d; and breda, e. {After Morgan.) ample in the cross 1X16 which will give Fi white and F2 in the ratio 220 white: 27 purple: 9 red. The complex relations here existing between only three phenotypes is a very good example of the sort of problems which must be solved by experimental genetics. Comb -characters in Fowls. — A variety of comb-characters are found in the domestic breeds of poultry and Bateson has made these the sub- ject of an extensive Mendelian investigation involving the rearing of over 12,000 individuals. The comb types involved are shown in Fig. 75. In this series of characters both rose and pea comb were found to be dominant to single comb, and give in F2 simple 3 : 1 ratios. These relations obviously indicate that there is a single factor difference be- TYPES OF FACTOR INTERACTIONS 165 tween rose and single and pea and single, but that the pair of factors involved must be different in each case. This is again shown in crosses between walnut and rose or pea-comb fowls for such crosses give walnut in Fi and 3 : 1 segregation in F2. Accordingly taking walnut as a domi- nant type, rose comb may be conceived of as differing from it by the factor r, and pea comb by the factor p. Walnut comb would then contain the dominant allelomorphs RRPF, and rose comb would be of the genetic constitutioi;^ rrPP, and pea comb, RRpp. Since single comb again differs by one factor from both rose and pea comb it must be of the genetic constitution rrpp. This analysis explains the experi- mental results which we have thus far outlined, but the critical test of RP Rp rP rp RP RRPP Walnut RRPp Walnut RrPP Walnut RrPp Walnut Rp RRPp Walnut RRpp Pea RrPp Walnut Rrpp Pea rP RrPP Walnut RrPp Walnut rrPP Rose rrPp Rose rp RrPp Walnut Rrpp Pea rrPp Rose rrpp Single Fig. 76. — Checkerboard analysis of theoretical expectations in F2 from a cross between rose-comb fowl (rrPP) and pea-comb fowl (RRpp). the hypothesis lies in the cross rose X pea. This should give walnut, RrPp, in Fi and in F2 all four types in the proportions 9 walnut: 3 rose: 3 pea:l single as shown in the checkerboard in Fig. 76. In one series of such experiments Bateson obtained the results shown in Table XXX. Table XXX. — Inheritance of Comb Type in Fowls. Cross Walnut Pea | Rose | Single RrPp X RrPp Observed. . . Expected. . . Ratio 279 312 9 132 104 3 99 104 3 45 35 1 RrPp X rrpp j Observed. . . Expected. . . Ratio 664 687 1 705 687 1 664 687 1 716 687 1 In this table the results of the back cross of Fi walnut to single are also given. A comparison of the values given with those expected on 166 GENETICS IN RELATION TO AGRICULTURE the basis of independent segregation of the factors indicates a fairly close correspondence between the two. It may be of some significance, however, that walnut and rose are the deficient classes in both cases. From the standpoint of factor interaction this case is of interest because it shows clearly that the character expression of a given set of factors cannot be predicted with certainty from the known character expression of some of these factors. It would have been impossible to predict from the character expressions involved that rose X pea would give walnut-comb fowls or that by recombination of the two recessive factors involved a single-comb fowl would result, for these two new comb types are totally different from the rose and pea types from which they can be derived. The obtaining of new characters of this kind by factor recombination is by no means an unusual thing in genetic experiments, and is sufficient justification in breeding work for testing factor combinations to determine what sort of character ex- pression may result from them. CRVH CrVH cRVH crVH CRVH CCRRVVHH Violet hairy CCRrVVHU Violet hairy CcRRVVHH Violet hairy CcRrVVHH Violet hairy CrVH CCRrVVHH Violet hairy CCrrVVHH Cream glabrous CcRrVVHH Violet hairy CcrrVVHH Cream glabrous cRVH CcRRVVHH Violet hairy CcRrVVHH Violet hairy ccRRVVHH White glabrous ccRrVVHH White glabrous crVH CcRrVVHH Violet hairy CcrrVVHH Cream glabrous ccRrVVHH White glabrous ccrrVVHH White glabrous Fig. 77. checkerboard analysis of a cross between two varieties of stocks, white glabrous (ccRRVVHH) and cream glabrous (CCrrVVHH). Miss Saunders' Factor System in Stocks. — A more complicated case of factor interaction as related to character expression has been in- vestigated by Miss Saunders in stocks (Matthiola) and has been inter- preted in somewhat the following fashion with respect to the factors and factor relations therein concerned. C — a factor for chromogen base which by itself gives a cream-colored flower. Its allelomorph, c, gives white flowers. R — a, factor for red coloration, epistatic to C. V — a factor for violet coloration epistatic to R. H — a factor for the production of hairs on the leaves, active only in the presence of C and R. The complicated relations existing between these factors are well TYPES OF FACTOR INTERACTIONS 167 illustrated by the cross white glabrous {ccRRVVHH) X cream glabrous (CCrrVVHH). This gives in T^i violet hairy plants (CcRrVVHH) which segregate in F2 according to the analysis given in the accompanying checkerboard (Fig. 77). The proportions are 9 violet hairy: 3 cream glabrous: 4 white glabrous. The peculiar feature of these relations is the fact that the factor H for hairiness can only act in the presence of C and R. In fact as far as the above experiment goes, the hairy condition might be considered as merely an extra effect of the interaction of C and R. However, glabrous violet plants are known and in these the factor h for the glabrous con- dition must be present. When a violet gla])rous {CCRRVVhh) plant is crossed with white glabrous {ccRRVVHH) the Fi again is violet hairy {CcRRVVHh), this time because the factor for hairiness is brought in by the white plant, and in F2 the segregation is as indicated in the CRVH CRVh cRVH cRVh CRVH CCRRVVHH Violet hairy CCRRVVHh Violet hairy CcRRVVHH Violet hairy CcRRVVHh Violet hairy CRVh CCRRVVHh Violet hairy CCRRVVhh Violet glabrous CcRRVVHh Violet hairy CcRRVVhh Violet glabrous cRVH CcRRVVHH Violet hairy CcRRVVHh Violet hairy ccRRVVHH White glabrous ccRRVVHh White glabrous cRVh CcRRVVHh Violet hairy CcRRVVhh Violet glabrous ccRRVVHh White glabrous ccRRVVhh White glabrous Fig. 78.- -Fi checkerboard analysis of a cross between violet glabrous {CCRRVVhh) and white glabrous {ccRRVVHH) stocks. checkerboard in Fig. 78. The phenotypic ratio obtained this time is 9 violet hairy: 3 violet glabrous: 4 white glabrous. This analysis not only adequately accounts for the phenomena as given above, but it also accounts for the F3 results and the various types of results that are obtained by mating other genotypes. In addition Miss Saunders found that when purple or white incana were mated to cream of the type above, the entire series of forms recorded for the previous white X cream mating were obtained and in addition cream hairy and cream glabrous. This at first sight appears to contradict the hypothesis that no cream or white hairy forms are possible. But closer examination has revealed the fact that white incana, which is itself hairy is in reality a colored form, i.e., possesses the factors C and R. This is shown by the fact that a slight tinge develops in flowers of this variety on fading, and in the F2 from a cross of this form with cream 168 GENETICS IN RELATION TO AGRICULTURE glabrous, those whites which tinge on fading are hairy and those which show no sign of coloration on fading are glabrous. The apparent diffi- culty is therefore merely due to the fact that some plants which possess C and R are still white on account of the action of other factors. Altenburg and MuUer's Truncate -winged Drosophila. — An even more complicated case of factor interaction is that concerned in the production of truncate wings in Drosophila (Fig. 79). The factors here involved appear to be the following: t — a factor for truncate wings. It is a recessive factor located in the second chromosome, and without this factor the truncate wing character cannot appear. ^1 — a factor which intensifies the expression of the truncate wing character, but which is not absolutely essential. This factor is located in the first chromosome. tz — another factor which intensifies the expression of the truncate wing, but is not absolutely essential to it. B' — -the dominant factor for bar eyes which in ad- dition acts as an intensifier of truncate. This is a first chromosome factor. line drawing of a h — a factor for black body color located in the second truncate- winged chromosome. This factor has such an influence that ter Morgan.) flies of the constitution {bT){bt) or even {Bt){hT) may display the truncate wing character. The truncate wing character was particularly baffling on account of the extraordinary relations which it displayed both in hybridization and in selected strains. In hybridization instead of a 3:1 ratio of long to truncate wing the ratio was about 7 : 1 and in selected strains even after 100 generations of selection there were still about 5 per cent, of long winged flies. That these long winged flies were different genetically from the truncate winged flies was shown by breeding tests for in such tests they did not produce as high a percentage of truncate winged flies as did those which had truncate wings. By means of linkage relations, however, it was possible to determine the factors concerned, and their specific effects. Particularly noteworthy is the fact that the factor B' for bar eyes acts as an intensifier for truncate, thus providing an analo- gous case to that in stocks where the color factors are necessary for the action of the factor for hairiness. No less interesting is the affect of h, for it was found that this factor, whether homozygous or heterozygous, changed the dominance relations in the allelomorphic pair Tt, so that the truncate wing character is expressed in such individuals when hetero- zygous for t. Furthermore since truncate appears more readily in the female than in the male it would appear that the sex factors also act as intensifiers. TYPES OF FACTOR INTERACTIONS 169 The important point involved in this case, however, is the ingenious way in which the investigators made use of the hnkage relations and the known fact that crossing-over does not occur in the male in order to study these factors, particularly with reference to their constancy, since they are variable in phenotypic expression. They took a truncate male which 9{bT){pT,)X (bT)ipT,)X {bT)ibT){pT,)XX black pink 9 long {bT)iPt,)X (bT){bT){Ph){pT,)XX black red 9 long {Bl){pT,)X {Bt){bT)iPt,)ipT,)XX Gray pink 9 long or truncate {Bt)iPl^)X {Bl){bT)iPk)ipT,)XX Red gray 9 long or truncate ibT)(pT,)X {bT){bT){pT,){pT,)XX Black pink cf long {bT){Pt,)Y {bT){bT){Ph){pT,)XY Black red cT long {Bt)ipT,)Y {BT)(JbT){pT,){pT,)XY Gray pink cT long or truncate (Bt){Ph)Y {Bt)ibT){PT,)ipT,)XY Gray red cf long or truncate Fig. 80. — Checkerboard analy.sis of Fi generation obtained by mating an Fi male Droso- phila of the constitution {hT){Bt)(pTi){Pt:i)XY with a pink l)laok long female. contained the truncate factor and also the truncate intensifier of the third chromosome and mated it to a long-winged black-bodied female with pink eyes. The genetic constitution of the truncate male with respect to the factors involved was {Bt) (Bt) (Ph) {Ph)X Y, and the contrasted black female was {bT){bT){pT3)ij)T3)XX. A male from such a cross is of the 170 GENETICS IN RELATION TO AGRICULTURE genetic constitution {hT){Bt){pT3)(Pt3)XY, and since no crossing-over occurs in the male it produces the following series of gametes: {bT){pT,)X {bT){pT,)Y {bT){Pt,)X {bT){Ph)Y {Bh){pT,)X {Bh){pT^)Y {Bh){Ph)X {Bh){pT,)Y When, therefore, such an Fi male is mated back to a black long pink female the results are as recorded in the checkerboard in Fig. 80. Of the male flies only the gray reds bear both the factors t and ^3. Such flies are long or truncate winged, but they should behave in the same fash- ion in further breeding tests unless the factors themselves are variable. Actually it was found that continued breeding back of these gray red males to black pink females gives approximately the same proportions of truncate to long in every generation. This method of taking advan- tage of the linkage relations and using the pink factor so that a given genotype could be determined without fail has in this series of experiments been the means of analyzing a case which otherwise would have baffled investigation, for the results clearly point to the fact that the genotypic differences which exist between the long and truncate flies of a selected culture are due to the fact that the lower vitality of truncated flies homo- zygous for the three factors directly concerned in the expression of this character favors the survival of heterozygous individuals, and it is, there- fore, practically impossible to secure a strain of truncate winged flies which will breed true. The Factor Explanation of Reversion. — Many phenomena included under the term reversion can be explained satisfactorily as instances of complex factor interaction. Reversion in general is a term applied to sudden return to an ancient, generally wild form, whether by hybridiza- tion or from other causes. The Mendelian explanation of reversion is most simply illustrated in Drosophila, for in Drosophila the relation of any particular form to the wild type is known accurately. Thus for example a form of Drosophila with miniature wings arose as a mutation directly from the long-wing type. Likewise several other wing characters have arisen from the long- wing type by a single mutation, among them vestigial wings. When now a vestigial-winged female is mated to a miniature male, the progeny all have long wings. This phenomenon may be explained by the fact that in a vestigial fly, a mutation has occurred in the locus V, which changed it to V without affecting the normal allelomorph of the miniature factor. Similarly the miniature fly bears the normal allelomorph of the vestigial factor, so that when the two are mated the original series of factors of the long-winged type is reunited and consequently the characters of the original wild form are reproduced. This is the principle on which rever- TYPES OF FACTOR INTERACTIONS 171 sion in hybridization depends, and other cases differ from this one only in the number of factor differences involved. Among the most notable cases of reversion arc those which Darwin describes in pigeons and fowls. Darwin regarded these throw-backs to wild types which he obtained by crossing various breeds of pigeons as important evidence of phyletic origin, and largely on the basis of this evidence concluded that the many varied modern breeds of pigeons are monophyletic in origin, that they are all derived from a single wild species. This species is the Wild Rock Pigeon, Columha livia, and in the wild it has an extended range over Europe, Abyssinia, India, and Japan. Even in the wild state it is variable, but under domestication breeds have been developed which show truly remarkable differences, and Darwin has described and illustrated these with great care. The hybridization experiments which Darwin conducted with domes- ticated breeds of pigeons were undertaken for the purpose of establishing relationship to the Wild Rock Pigeon. The phenomenon of throwing blue in pigeons is an exceedingly common one, but Darwin conducted experiments with breeds which had been bred for many generations and rarely, if ever, gave blue birds. Cole has summarized the results of one of his experiments about as in Fig. 81 : Barb cf X Spot 9 (Self black) (White with red spot on forehead and red tail). Fi Barb-spot (Black or brown with some white splashes). Barb C -a ^ H bO _c . "C .2 sc C bC ■"' c3 o > a +^ bD . t5 Sd ^ >> o o > ° o X !r! <^ rH a; OJ c § o bc cj ts .is T3 OJ & c3 -^ 1^ O bC <^ _ OS Q^ C >r^ 03 ^ .S >o CO o 'O t^ rt 1 ■* -H o» IN 00 c<3 CO •* ■* _r 00 lo o + + 00 1^ O O O O ro o o» "o ira 00 iCi t-H ^ -t< _- 0) lO o " + + ■* i-H o» "5 O "-i rt o lo w lO e<3 "* . . eij • • • rH Tjt T)< " (N 10 O + + ro ■* o» "O lo 00 tH 03 M t^ C^ --1 1— t Tj4 CO cra 4. + + 00 00 «D in o "-1 00 t^ t- IN O ^ O . .(.... 00 CO N >n o + + 00 •<«< « "O o >-< t^ >0 « t^ >0 IN 03 • ■ to • • • CO CO - .-1 ■* o + + lO 00 08 >0 O C5 t^ TjH O t^ •'5 O 00 • . ^ . . . CO CO _r 1-H TjH o " + + CD O e- O "5 -M lO IN e- lO l> IN t^ . . o • ■ • CO 00 _r "< •* o ^ + + IN ^ »H O O 00 lo >-i 0 lO CD CO 03 «H t^ C^ i-l lO . . ,o • • • CO IN O ■* O + + Ci 00 ^ lO uo CD o t~ ■* t^ t^ o •* . . ^ . . . 00 IN O CO O + + CO i-i »H lo o CO t» lO «* t^ O CO n . . w • • • IN IN- O ■* O + + O IN iH O "O .-1 >0 0> t- O t^ CO (M . . ^ . . . (N rH 1-1 CO O 1 + ^ lO O O O Oi in o US o o IN • tH • • • IN N >-l CO O + + tl n a 1 bi) 5P a be c ■- ^ are ffsp ring offs off nts a o a^ "S K ^ ^ m O " 03 o o to o a 0) a> "S S - c T3 "O "S °^ t- S c OS 03 ° fc, M .2 1- 1-. !h '^ .3 c c 1 5 % 1 3 i , a i "a 3 C > a-^ a; .3' -d — ' bfl ■— 0) bC a - to bX) a; .5 1^ bc 1:5 03 CO ^ ^ O O 3 S o e t3 O "^^ TS -^ >> o -^ +^ c3 O O rl3 o ^ ^ - 3 ^ ^ « fli to oj o o OJ OJ '2 - ^ g =3 g bc -a g ^ 0 IN -«i . . t. . . . IN C^ rt 05 O II II 00 03 <0 "O O 00 ■* CO O t^ 'O -H CO . . o ■• • IN C^) - r^ CO O II II •* CO I- o o o •i»< C^ CO o >o •* IN C^ N _r -H CO O 1 1 1 1 O "O * O O 00 CO rH 00 O «5 O . . o» • • ■ *"* IN IN rH CO O II II 00 rH W O "(5 lO rH O lO O C<1 rH O . . ^ . . . IN Oq rH CO O 1 1 1 1 rH IN rt O "O lO rH OS 0> lO t^ O 03 (N rH - O IN O 1 1 1 1 1(5 O to O "5 03 O 00 « WO oo IN rH - IN O II 1 -H CO O O lO •* O t^ 00 I^ rH w N rH - (NO 1 I "^ • 1 CO CO « O O CO 00 1(5 lO in rH CO rH rH - (NO 1 1 1 05 rH tH O O 00 t^ ■* O "5 rH >c rt rH (NO 1 1 1 OS 00 «> O lO IN CO IN « >0 W O 'if . ■ CO ■ ■ ■ rH rH O IN O II +1 CO CO U9 O O O K5 rH 0> O IN CO r^ rH (NO 1 1 1 rH t^ e« O O M ■*! O « lO O o IN ■ iH • • • rH rH O N O 11 +11 CO O US "5 O ■* O lO (N O ^~* rH rH O IN II +1 . . 5 n • • a a ^1 . big •C.Sto a t o aren ffspr ring offs snts- a o a o .« (- ade of ade of of offs grade rade o ion, pa c c a c a M b£ a; -e .o 03 Mean Mean >J„r.,W 5.2' 1 J 192 GENETICS IN RELATION TO AGRICULTURE and intergrading with the parental forms, escaped notice as mutations but were selected for continuing the experiment? Such an assumption would render intelligible the efficacy of return selection which would be difficult of interpretation on even a multiple factor theory of heredity. That such a system may exist in qualitative characters has been shown by Bridges for the relation between eosin eye color and its modifiers in Drosophila. One modifier called dark intensifies the eosin character. The other six modifiers are all diluters. Thus cream a changes eosin to pale yellow or cream color, cream h has a similar effect, but not so marked. Whiting changes the eosin color to white, so that eosin-whiting flies are indistinguishable from white-eyed flies in color. In these cases there is no question as to the operation of a multiple system of factors, for the specific factors have arisen singly by mutation and their linkage relations establish completely their identities. Nevertheless taken together they would give in a qualitative character a remarkably close imitation of the behavior of Castle's hooded rats. If, however, we assume with Castle that factors like characters are variable and that allelomorphic contamination occurs, then we may offer an explanation based on a consideration of a single allelomorphic system. For such an explanation the hooded pattern may in general be represented by h, and its dominant allelomorph, the fully colored condition, by H. Self-color is dominant to hooded, but the hooded condition varies greatly in the amount of pigmentation present in the coat. These variations appear to be correlated with definite factor variations, consequently wc may designate the factors determining the various degrees of pigmen- tation in the hooded pattern by hi, h^, hs, hi, . . . hn. This series runs from individuals which show practically no color to those which display almost a self-colored coat. If we assume that the character expression of an animal of the genetic constitution hihw be intermediate between that of an animal of the genetic contitution hihi, a very light type, and one of the constitution hiohio, a very dark type, then we may point out what would occur if selection were carried out in the progeny of such an individual. In the fii'St place the genetic constitution hihio of such an animal represents merely the values of the gametes that united to form the zygote. They are assumed to interact immediately, so that perhaps, in addition to the factors hi and hio, such a zygote will produce gametes bearing for the most part the factors h^ and he, representing a sort of equilibrium for the interaction of the factors hi and hio- There would, therefore, be in the progeny of such an individual some individuals of the genetic constitution hih^ which would be lighter than the parents, and some of the genetic constitution h^hio which would be considerably darker than the parents. If other products of this reaction, such as hs, h^, /17, hg, etc., were also produced, and like the original reacting factors hi and hjo FACTOR RELATIONS IN QUANTITATIVE INHERITANCE 193 occurred in relatively infrequent numbers in the gametes, then other combinations would result. A graded series running from light to dark would then be produced, but since the mid-products, h^ and ha, would be by far most numerous, in small progenies most of the individuals would display a mid-condition of pigment development. On such an explana- tion any particular type of hooded pattern would be allelomorphic to the wholly pigmented condition or to the Irish condition, if these latter two with hooded be members of a system of triple allelomorphs. Also, the variability of the factor due to interaction with whichever other member of the allelomorphic pair it happened to be in contact would account for the variability in the expression of the hooded pattern following hybridization. The progress of selection in such a form on the basis of a single factor as determining not only the hooded pattern, but also the extent of pigmentation in the hooded condition requires us to assume an instability in the factor even when in the pure condition. We would, therefore, assume that, in an animal of the constitution h^h^ with respect to the factor for hooded pattern, the instability of the factor leads at times to the production of gametes by such an animal bearing the factor h4 on the one hand, or h^ on the other hand. If such gametes were produced relatively infrequently, they would almost invariably mate with gametes of the genetic constitution h^. The resulting progeny would have the genetic constitutions /14/15 and hohe and they would be slightly lighter and slightly darker respectively than the bulk of the animals of the genetic constitution /15/15. Selection of such individuals would rapidly lead to the production of races of the genetic constitutions hJiA and h^he. Individuals of the genetic constitution ^4/14 on account of the variability in the factor itself would produce some gametes bearing the factor hs or the factor h^, and by continuing the same process of selection a still lighter race of the genetic constitution hshs might soon be estab- lished. Assuming, therefore, that factor variability of this type occurs it is not difficult to see how a continuous process of selection such as Castle has employed should finally result in the establishment of new races differing markedly in their character expressions and possessing a different but related genotype to that of the original type from which selection has been made. Moreover, such an hypothesis accounts for the observed fact that return selection is just as efficient but no more so than the original selection in changing the mean of the races, a fact which presents some difficulties for a strict multiple factor interpretation. It should be stated that this hypothesis of factor variability does no violence to our conception of the nature of factors, except with respect to a rather ill-established belief in factor constancy. The continuous change in a factor such as we have outlined above reminds us very forcibly 13 194 GENETICS IN RELATION TO AGRICULTURE of the behavior of certain chemical systems. It is a well-known fact for instance that in some systems an equilibrium is reached when a certain proportion of two substances are present in a chemical system. Thus a system consisting of A and B, two compounds mutually convertible into each other, may reach an equilibrium when say 2 parts of A and 1 of B are present in the system. If now a certain proportion of A is removed from the system, enough of B will be converted into A to rees- tablish the old equilibrium of 2A : IB. It is not difficult to see, there- fore, that continuous removal of A from such a system would finally result in the conversion of all of B into A. Assuming, therefore, that our original system consisted merely of an unstable chemical compound, it might be possible by continuously removing a certain product of its instability to gradually alter the system in a given direction, much as we have outlined the case for alteration of the hooded pattern by continuous selection in rats. Since such changes are usually reversible, the efficacy of return selection is adequately accounted for. Nevertheless, although it must be admitted that an interpretation such as we have given above may account for all the known facts of quantitative inheritance, and as the student can readily see it may be employed to interpret the entire set of eight conditions which East has outlined, we advocate the strict multiple factor hypothesis of size inheritance for the following reasons: 1. It is definitely known that large numbers of loci may be concerned in the expression of a certain character. Morgan has stated that over twenty-five factors are known to be concerned with eye color in Droso- phila, and similarly a large number of factors affect body color and wing characters. The assumption of large numbers of factors as concerned with a single character does not, therefore, do violence to modern con- ceptions of factor and character relationships. 2. Size is a complex character depending on the cooperation and coordination of many organs, tissues, and physiological processes. Some factors may, therefore, affect one organ, some another, so from this viewpoint a large number of factor differences might be expected to be present in cases of quantitative inheritance. 3. Although factor constancy cannot yet be considered a universally established fact, those definite investigations which have been reported indicate that factors possess on the whole a high degree of stability. More definite work is needed along this line; provisionally it appears wise to consider factors for all practical purposes as constant.^ 4. Simple factor differences are known to give size differences, ^ That factors are relatively stable entities is being evidenced more clearly all the time. Witness the definite arguments advanced by Bridges and Muller respectively in their recent papers on "Deficiency" and "An Oenothera-like case in Drosophila." FACTOR RELATIONS IN QUANTITATIVE INHERITANCE 195 depending many times upon some definite character change in the race, for example, bush and Cupid sweet peas. It appears reasonable to refer more complex size differences merely to differences in several such definite characters. 5. Factor systems affecting a single character have been worked out definitely, which in the complexity of the interrelations they display, rival those interpretations which have been postulated for cases of quantitative inheritance. The trend of investigation seems to establish more firmly all the time the probable validity of the multiple factor interpretation of quantitative inheritance. CHAPTER XI INHERITANCE OF SEX AND RELATED PHENOMENA In the description of the chromosome relations obtaining in the distri- bution of hereditary units, we have had occasion to show how sex in one form, Drosophila a7npelophila, depends upon differences in the chromo- some constitution. In this species three pairs of chromosomes have equal members in both sexes, but the remaining pair in the female consists of two equivalent X-chromosomes, in the male of one X-chromosome like those in the female paired to an unequal F-chromosome. The distri- bution of sex-linked factors finds a logical explanation in their location in the X-chromosomes, and in Drosophila more than fifty sex-linked factors have been studied. But thus far the F-chromosome has not been demonstrated to carry any of those factors which are known to be located in the X-chromosome. When the chromosome relations obtaining in the inheritance of sex in Drosophila are outlined they are found to be as follows: XX X XF XX Morgan has called this the XF type of sex inheritance. This type of sex inheritance is characterized by the fact that females are homozy- gous for the sex determiners and males are heterozygous for them. Accordingly females produce but one kind of egg with respect to the sex determiners borne by them, but the males produce two kinds of sperm in approximately equal numbers. These two kinds of sperm have been called female-producing and male-producing sperm, because normally when a female-producing sperm fertilizes an egg a female is produced and when a male-producing sperm fertilizes an egg a male is produced. The production of male and female producing sperm in approximately equal numbers and random mating with the egg cells accounts for the approximate equality of the sexes in each generation. The XF type of sex inheritance is characteristic of a large number of forms. Apparently all mammals, including man, belong to this type, a number of insects, and the plants Bryonia and Lychnis. The evidence in some cases is based on the results of sex-linked experiments, in some 196 1 INHERITANCE OF SEX AND RELATED PHENOMENA 197 cases on favorable cytological evidence, but in only a few cases has satisfactory evidence been secured from both sources. In a previous chapter we have shown in detail how a sex-linked char- acter in Drosophila is inherited. By referring to the list of factors in Drosophila it may be seen that in this insect about fifty factors are known to belong to the first chromosome, and, therefore, to display the sex- linked type of inheritance. Although cases of sex-linked inheritance are known in' other animals, in none do we have as complete a body of knowledge as in Drosophila. Nevertheless, there is a sufficiency of other cases to lend strong support to the evidence derived from the Dro- sophila investigations. In man particularly several sex-linked factors are known, and the evidence in support of this analysis is fairly satisfactory. A typical case in man is that of color-blindness, which is much more common in males than in females. The factor for color-blindness may be called b and its normal allelomorph B. A normal-visioned woman is then of the genetic constitution (BX)(BX), and a normal man is (BX)Y. The corresponding abnormal forms are for women (bX) (bX) and for men (bX) Y. Since the factor for color-blindness is recessive, a woman of the genetic constitution (BX)(bX) will have normal color vision. In this we see the reason for the greater number of men that are color-blind. A man with a simplex dose of the factor is color-blind, because the F-chromosome as in Drosophila carries no demonstrable factors. In the simplex woman, (BX)(bX), on the other hand, the dominant allelomorph determines the type of color vision, so that a normal woman is produced. Simplex women are just as common as simplex men, the greater number of men displaying the color blind character is simply due to the different chromo- some constitutions of the two sexes. The relations which exist in the inheritance of color-blindness are exactly the same as those which exist in the inheritance of white eye color in Drosophila. A normal woman {BX){BX) mated to a color-blind man {bX) Y produces in Fi normal daughters of the genetic constitution {BX) {bX) and normal sons of the genetic constitution {BX) Y. These Fi normal sons are of exactly the same genetic constitution as all other nor- mal men and, therefore, although they had a color-blind father, they can never transmit the defect. The normal Fi women of the genetic constitution {BX){bX)^ however, when mated to normal men produce daughters of the formulae {BX){BX) and {BX){bX), all of which are, therefore, normal, and sons in equal numbers of the constitution {BX) Y, normal, and {bX)Y, color-blind. A simplex woman, therefore, although she does not herself exhibit the color-blind character, when mated to a normal man, transmits that character to none of her daughters, but to half of her sons. A color-blind woman can be produced by the rare 198 GENETICS IN RELATION TO AGRICULTURE mating, simplex woman {BX){hX) by color-blind man Q}X)Y, or by the still less frequent mating of color-blind woman (6X) (bX) by color-blind man (bX) Y, in which latter case all the offspring whether sons or daughters are color-blind. A considerable list of other sex-linked factors demon- strate beyond question that the inheritance of sex and the distribution of sex-linked factors in man is strictly analogous to that which we have found to obtain in Drosophila. Non-disjunction in Drosophila. — Of particular interest from the standpoint of the inheritance of sex and of the relation between factors and the chromosomes are the results which Bridges has obtained from his extensive in- vestigations of non-disjunction in Drosophila. The investigations on non-disjunction had their origin in certain ''exceptions" which ap- peared from time to time in cultures of Drosophila. Ordinarily in the case of sex- linked characters when a female with the recessive character is mated to a male with the dominant character all the females in Fi exhibit the dominant sex-linked character and all the males the recessive character. The reason for this fact has been explained already, but it will be clearly apparent from a consideration of Fig. 89, which is a diagram of the results of crosses between vermilion females and red males. The vermilion factor V is borne by the sex chromosomes, and since the males from crosses between vermilion females and red males receive their only X-chromosome from the mother they should all be vermilion-eyed. The females from such a cross receive from the father an X-chromosome bearing the dominant allelomorph of v, consequently they should all be red-eyed. In the great majority of cases, this is the result actually obtained from such matings, but occasionally, about once in 1700 individuals, an exception, a vermilion female or a red male, is produced. The in- vestigation of the "exceptional" females from sych matings has pro- vided unique evidence in support of the chromosome theory of heredity and in regard to the relations existing between the sex chromosomes and sex differentiation. The production of exceptional individuals from matings such as we have considered above apparently results from occasional aberrant re- duction divisions in the female such that the two X-chromosomes fail Fig. 89. — The relations of the sex chromsomes to sex produc- tion and to the inheritance of the recessive sex-linked char- acter, vermilion eye color, in Drosophila. The straight chro- mosomes are the X-chromo- somes, and the crooked ones the y-chromosomes. (Adapted from Bridges.) INHERITANCE OF SEX AND RELATED PHENOMENA 199 to disjoin from each other. As a result eggs are occasionally produced which contain two X-chromosomes instead of one as is normally the case. In Fig, 90 are illustrated in diagram the consequences of such aberrant reduction divisions in the female. If the X-chromosomes fail to disjoin in the reduction divisions, they may be included in the egg, in which case an egg wjth two X-chromosomes is produced, or they may both be thrown out into the polar body, in which case an egg with no X-chromo- some is produced. This phenomenon Bridges calls primary non-dis- junction. An egg (vX){vX) fertihzed by a F sperm gives a (vX)(vX)Y zygote, and it develops into an exceptional vermilion female. An Fig. 90. — Diagram of the production of exceptional individuals, vermilion females and red males, through primary non-disjunction from matings of vermilion female by red male. (Adapted from Bridges.) egg (one which contains no sex chromosome) fertilized by a (FX) sperm gives a {VX)0 zygote, and it develops into an exceptional red male. Zygotes of the constitution ( FX) (vX) (vX) and YO are, also, possible as a consequence of such non-disjunction but it is certain that they die, consequently nothing definite can be determined as to their characters. The proof that non-disjunction is. the correct interpretation of these exceptional cases in the transmission of sex-linked characters has been established by breeding tests and by actual cytological examination of exceptional individuals. Assuming that homologous chromosomes pair in synapsis, in an XXY exceptional female two types of reduction divisions are possible. If the two X-chromosomes pair, then in reduction they disjoin and one goes^to each pole. The free F-chromosome then passes as often to one pole as to the other, and as a consequence, two kinds of eggs, XF and X, are produced in equal numbers. On the other hand, when the 200 GENETICS IN RELATION TO AGRICULTURE Vermilion XXY Female XY Synapsis 16% XX Synapsis 84'%7 Dies(l) Wild Type Male (2) Wild Type Female (3) Wild Type Female (4) Exception Fertilization by Y Sperm of Wild MaW Vermilion Female ( 6 ) Exception Dies (6 ) Vermilion Male (7) Vermilion Male (8) Fig. 91. — Secondary non-disjunction in the female. Diagram showing the constitu- tion of an exceptional vermilion female, the two types of synapsis, reduction, and the four classes of eggs produced. Each kind of egg may be fertilized by either of the two (X and F) kinds of sperm of the wild male, giving the eight classes of zygotes shown. (After Bridges.) INHERITANCE OF SEX AND RELATED PHENOMENA 201 } ''-chromosome pairs with an X-chromosome, the free X-chromosome then goes as often to one pole as the other and this results in the pro- duction of equal numbers of A", XX, XY, and Y eggs. This set of re- lations is shown in diagram in Fig. 91, which illustrates the phenomena exhibited in the production of gametes by a vermilion non-disjunctional female. From experimental evidence it has been determined that homosynapsis, 'i.e., pairing of the two X-chromosomes, takes place in 84 per cent, of cases in non-disjunctional females and heterosynapsis, pairing of an X"- with a F-chromosome, in 16 per cent, of cases. A non- disjunctional female, therefore, will produce four types of eggs in the following proportions 4 (vX) (vX) : 4 F : 46 (vX) : 46 (yX) F. When a vermilion non-disjunctional female is mated to a red male, the Fi consists of about 46 per cent, each of red females and vermilion males and about 2 per cent, each of further exceptions, vermilion females and red males. Non-disjunctional females are, therefore, characterized by the production of further exceptional offspring to the extent of about 4 per cent. This type of non-disjunction consequent upon the presence of an extra F-chromosome is styled secondary non-disjunction. Two additional types of zygotes are produced as a result of secondary non- disjunction, those of the constitution YY which die, and those of the constitution Xy^Y, which make up half of the males and are not ex- ceptional with respect to their characters but which can transmit non- disjunction to a certain proportion of their offspring. It will also be noted that of the regular daughters half are of the constitution XX F. They possess the power of producing exceptions on account of the presence of the extra F-chromosome, but they can only be distinguished from their normal sisters by breeding tests or less conveniently by cytological examination. It is evident that an Fi population such as this from the mating of a vermilion female to a red male is very different from that which is normally obtained. Bridges has followed out very skilfully many of the consequences of the assumption that these exceptional individuals are actually due to non-disjunction of the sex-chromosomes and consequent production of various types of abnormal chromosome constitution. Thus if we con- sider the exceptions produced by a non-disjunctional female, it is clear that they are a consequence of heterosynapsis in the female. Now when the X-chromosome pairs with a I'-chromosome in synapsis, it very evidently has no opportunity to exchange chromatin material with the free A^-chromosome. Accordingly all the XX eggs and con- sequently all the exceptional daughters from such a female will belong to non-cross-over classes. A consideration of an actual experiment 202 GENETICS IN RELATION TO AGRICULTURE will make this matter clearer. Bridges took non-disjunctional females known from the type of mating involved in their production to be of the genetic constitution (WVFb'X)(w''vfb'X)Y and mated them to bar- eyed males {WVFB'X)Y. Obviously the regular daughters of such a mating will be bar-eyed, because they receive from the father an X- chromosome bearing the dominant factor for bar eyes, but the excep- tional daughters will not be bar-eyed since both their ^-.chromosomes are derived from the mother. The question concerning these excep- tional daughters is as to whether they are invariably of the genetic constitution (WVFb'X)(w''vfb'X)Y or whether they may occasionally be cross-overs, for example (WVfb'X){w^vFb'X)Y or {Wvfb'X)(w'VFb'- X)Y. Since the loci involved in this case are W = 1.1, V = 33.0, and F = 56.5, normal crossing-over should give about 50 per cent, of cross-overs. By testing the exceptional females again with bar males of the above genetic constitution, the distribution of the males into phenotypes serves as an accurate indication of the genetic constitution of the mother. In every case in tests of thirty-seven exceptional daugh- ters, wild type males (WVFb'X)Y and eosin vermilion forked males (w^vfb'X) made up the largest classes. This indicated that the females were all of the genetic constitution (WVFb'X){w''vfb'X)Y, and, there- fore, were non-cross-overs. The above facts are to be taken in conjunction with the fact that crossing- over actually may occur in non-disjunctional females in homo- synapsis. We have pointed out in another place that crossing-over does not occur in males. Now in non-disjunctional females the occur- rence of heterosynapsis might well set up a condition like that which is responsible for non-crossing-over in the male for we would have duplicated the exact type of reductional divisions which occur in the male aside from the presence of an unpaired X-chromosome in the reduction spindle. But as a matter of fact the presence of the F-chromosome does not appear to affect crossing-over between the X-chromosomes in homosynapsis. Thus Bridges has summarized the data for crossing-over in non-dis- junctional XX Y cultures and compared them with the data for crossing- over in normal XX cultures with the results given in Table XXXVI. Far from resulting in no crossing-over the presence of the F-chromosome actually appears to have increased the per cent, of crossing-over be- tween loci in the X-chromosomes. No reason can be readily assigned for this increase in crossing-over, but it is of interest to note that the presence of a F-chromosome does not preclude the occurrence of cross- ing-over. In Fig. 91 it is shown that half of the regular sons of a non-dis- junctional female are of the type XYY instead of XF as normally. The hereditary behavior of such males as determined by experiment is shown INHERITANCE OF SEX AND RELATED PHENOMENA 203 Table XXXVI. — A Comparison of Cross-over Values from Normal and Non- DxsjUNCTiONAL CULTURES IN Drosophila {Data from Bridges) XX cultures XX y cultures Increase Loci Total Cross-over value Total Cross-over value increase WT 2,600 15,177 6,262 1,699 2,600 6,262 24.4 29.5 43.1 43.6 5.6 22.4 2,436 12,817 3,651 257 2,436 3,651 26.0 33.7 49.8 53.0 5.9 26.0 1.6 4.2 6.7 9.4 0.3 4.4 6.6 WV 14.2 WF 15 5 WB' 21.6 TV 4.6 VF 19.6 Vermilion XYY Male XY Synapsis 67% Reduction! YY Synapsis 33% Sperm Offspring! Sable Male (1) Wild Type Female (2) Wild Type Female (3) Sable Male (4) Fig. 92. — Diagram of secondary non disjunction in the male. Four kinds of sperm are produced, but none of these lead to the production of phenotypic exceptions in Fi. (.After Bridges.) 204 GENETICS IN RELATION TO AGRICULTURE in diagram in Fig. 92. There are two possible types of synapsis in non-disjunctional males, the ordinary type of heterosynapsis in the male in which Y is paired with X, in which case one F is free, or the YY type of homosynapsis in which the X-chromosome is free. Obviously, if these two forms of synapsis take place according to the laws of chance homosynapsis will occur twice as often as heterosynapsis. Assuming this to be true the gametic series of a non-disjunctional vermilion male will be as follows: 2(vX)Y:2Y:l{vX):lYY. When such males are mated to sable females, all the males in Fi are sable and all the females are of the wild type. No exceptions, therefore, are produced in Fi, but two-thirds of the daughters are non-disjunctional and should give exceptions in F2. Bridges showed that among fifty- four females only fifteen gave no exceptions in F2. Consequently 72 per cent, of the females must have been non-disjunctional, and this may be regarded as an insignificant deviation from the expected value of 67 per cent. We cannot go into detail concerning any other of the numerous points which have been investigated with respect to non-disjunction and its attendant phenomena. That non-disjunction is not due to the pres- ence of a sex-linked factor was proven by two lines of experimental evidence. In the first place such a factor should have shown linkage relations with the sex-linked factors and consequent crossing-over in definite percentages with different loci. An extensive series of matings showed, however, that non-disjunction was entirely independent of linkage relations. The other line of evidence related to attempts to establish pure stock of non-disjunction. These attempts failed com- pletely, a fact readily explainable on the basis of non-disjunction, but reconciled with considerable difficulty to the factor idea. If this were not sufficient evidence, the results of cytological examination are cer- tainly conclusive. Examination of a number of exceptional females showed them to be of the chromosome constitution XX Y, and examina- tion of regular females from non-disjunctional mothers demonstrated that about half of them were XX Y, as was to be expected from theory. In brief the entire series of investigations give unique support to the chromosome theory of heredity, for throughout in this exceptional behavior of the hereditary mechanism, the factor distribution exactly parallels the unusual history of the X-chromosomes. From the standpoint of the inheritance of sex the investigations on non-disjunction throw interesting sidelights on the relations of chromo- some constitution to sex. Thus females may be of the constitutions XX or XX Y or even XX YY. Evidently, therefore, the presence of the INHERITANCE OF SEX AND RELATED PHENOMENA 205 extra }^-chromosomc has no influence on the determination of. sex, although it does give rise to unusual relations in the production of gametes. Zygotes of the constitution XXX would presumably be females, but they die and consequently nothing can be determined as to their behavior. Males can be either normal XY or exceptional XYY and XO. The last, although normal males in appearance, are always sterile. The y-chroniosome, therefore, must play some definite, positive role in gametogenesis, although we are at present unable to state just what its function is. Along with the preceding cases of female constitutions, these different types of males indicate that the determina- tion of sex depends upon the number of X-chromosomes present. If two be present, a female is produced and the presence of one or two super- numerary F-chromosomes does not alter this fact. If only one A^'-chromo- some is present a male is produced, and it is immaterial whether no Y is present or whether one or two such chromosomes are present. Throughout, the inert nature of the F-chromosome is emphasized, the only evidence we have of its positive action being the sterility of XO males. It is important also to note that the derivation of the chromosomes, whether from the female or from the male, does not influence the sex of the oiTspring. Ordinarily a male is produced when a gamete from the female bearing an A-chromosome is fertilized by a gamete from a male which bears a F-chromosome. In non-disjunctional strains, however, some males are produced from the union of a F-bearing egg with an A-bearing sperm, exactly the reverse of the usual procedure. Also in such strains some females are produced by the union of an egg containing two A'-chromosomes with a F-bearing, or ordinarily male-producing, sperm. Non-disjunction, therefore, establishes firmly the intimate relation between chromosome constitution and sex determination. The WZ Type of Sex-inheritance. — A method of sex-inheritance exactly the reverse of the AF type is that which Morgan has styled the WZ type of sex-inheritance. In this type of sex inheritance the females are heterozygous for a sex-determiner and the males homozygous. If we diagram the relations which exist here, they will be as follows : WZ X zz I \ I w z ^-z WZ The classical example of this type of sex-inheritance is Abraxas grossulariata, and, as in the AF type, the evidence for the relations obtaining in the inheritance of sex was given by the behavior of a sex- 206 GENETICS IN RELATION TO AGRICULTURE linked character. As it occurs in the wild, the currant moth' is usually of the typical form which is characterized by dark markings on the wings which although highly variable are of characteristic shape and arranged in a definite pattern. This is the form styled grossulariata. Occasionally in nature, however, a female is discovered which is much lighter than Fig. 93. -Diagram illustrating the inheritance of laclicolor type iu Abraxas. A lacticolor female mated to a grossulariata male. {Adapted from Morgan.) the type on account of a reduction both in number and size of the black markings of the wings. This is the form styled lacticolor. It is of in- terest to note that according to Doncaster, save in one doubtful case, only females of the lacticolor type have been discovered in nature. The inheritance of lacticolor type is illustrated diagrammatically in Figs. 93 and 94. In these diagrams the TF-chromosome is represented INHERITANCE OF SEX AND RELATED PHENOMENA 207 as containing no factors, and the Z-chromosomes, as containing either the recessive factor I for lacticolor, or the dominant allelomorph L which con- ditions the development of the grossulariata type. When lacticolor females from nature are mated to grossulariata males, Fi consists of grossulariata males and females of the genetic constitutions W(LZ) Fig. 94. — Diagram illustrating the inheritance of lacticolor type in Abraxas. A grossxdariata female mated to a lacticolor male, the reciprocal cross of that represented In Fig. 93. {Adapted from Morgan.) and (LZ){IZ) respectively. The females are genetically the same as those of a pure race of grossulariata. Abundant experimental evidence demonstrates conclusively that not only are they themselves grossulariata, but they cannot transmit anything but the grossulariata character to their offspring. The Fo consists of grossulariata males half of which are 208 GENETICS IN RELATION TO AGRICULTURE homozygous for grossulariata, therefore {LZ){LZ), and half heterozygous {LZ){IZ). Of the females half are grossulariata W{LZ) and half lacticolor W{IZ). No lacticolor males are produced in this generation, but they may be obtained from matings of heterozygous grossulariata males (LZ) (IZ) with lacticolor females W{IZ). The reciprocal cross requires no special explanation, since it is perfectly clear from the diagram just how the lacticolor factor is transmitted in such cases. Throughout, the whole set of experimental evidence duplicates exactly the relations found to exist for the inheritance of white eye color in Drosophila except that the sex relations are reversed. The cytological relations in Abraxas do not appear to rest upon as firm a basis as those in Drosophila. Apparently there are normally 56 chromosomes in both the male and female, and no pair are obviously unequal in either sex. Apparently then the TF-chromosome in the female is about the same size as the homologous Z-chromosome, but like the Y in Drosophila it is a neutral chromosome, i.e., it carries none of the dominant sex-linked factors. Some additional cytological evidence is provided by examination of lines giving aberrant sex ratios. Doncaster discovered certain strains in which some of the females gave only female offspring, others only a few sons, and still others the normal 1 : 1 ratio. In these strains the males had 56 chromosomes, but the females only 55. As Bridges points ~ out, if 56 is the normal chromosome number for the females of Abraxas, then those females having 55 chromosomes may be regarded as of the ZO type, corresponding to the XO mates in non-disjunctional strains of Drosophila. Such females produce eggs some with 27 and others with 28 chromosomes. If as Doncaster's early observations seemed to show, the odd chromosome ordinarily is included in the polar body, then the eggs would contain mostly 27 chromosomes, and these on fertilization would give 55 chromosome zygotes, presumably females of the ZO type. Later observations of Doncaster's, however, do not confirm the conclusion that 27 chromosome eggs are more frequent than those containing 28 chromosomes. Moreover, although this is perhaps not a very weighty argument, it is not clear why ZO females in Abraxas, if such exist, should not be sterile like their counterparts, the XO males in Drosophila. It is of considerable interest that exceptions in the transmission of the sex-linked character lacticolor occur in Abraxas just as they do in Dro- sophila. The mating grossulariata female by lacticolor male should give only lacticolor females and grossulariata males. However, Doncaster found among 611 females, the offspring of 27 such matings, three gros- sulariata females and two of these were in the same brood. Assuming that the two which were in one brood represented cases of secondary non-disjunction, it would appear that primary non-disjunction in Abraxas INHERITANCE OF SEX AND RELATED PHENOMENA 209 is not certainly more frequent than in Drosophila. There appears at present to be good reason for accepting the explanation of non-disjunction for these exceptional cases, although Doncaster has advanced the sugges- tion that, if the sex-differentiator be assumed to occupy a definite locus in the Z-chromosome, then, if the Z-chromosome divides in such a way that the factor / is separated from the sex factor, exceptions will be produced. This case has not yet been worked out as carefully as has that in Droso- phila, but it presents so many close analogies that the possible interpreta- tion is fairly clear. Of forms showing the WZ type of sex inheritance a number are known. Moths and butterflies appear to exhibit this type universally, and such birds as have been investigated are all of the WZ type. A familiar example is that displayed by the barred pattern factor in poultry. When black hens are mated to barred cocks, Fi consists of barred hens and barred cocks and F-2 of 2 barred cocks : 1 barred hen : 1 black cock. The reciprocal cross barred hen by black cock gives in Fi black hens and barred cocks, and in Fo 1 barred cock : 1 black cock : 1 barred hen : 1 black hen. These are the relations which Pearl and Surface have demonstrated for crosses between the Plymouth Rock fowl and the Cornish Indian Game. The relations are exactly like those in the crosses of grossulariata and ladicolor, to diagram them it is merely necessary to substitute barred for grossulariata and black for laaicolor. A number of other characters in birds display the WZ type of sex- linked inheritance. Red-eye color in canaries behaves like ladicolor in Abraxas when contrasted with black-eye color, but exceptions seem to be unusually numerous. In pigeons a number of factors are known to be sex-linked. Thus in turtledoves normal color is sex-linked when con- trasted with white; and in the domestic pigeon the factor for intense coloration is sex-linked. In the fowl Bateson and Punnett have shown that a factor for silky plumage is sex-linked, and Pearl has demonstrated the existence of a sex-linked factor for high egg production. The latter case because of its economical importance will be given full treatment in another place. Besides these there are many other suspected cases, but they all occur either in moths and butterflies or in birds. Finally it remains to call attention to another analogy between the XY and WZ types of sex-inheritance. It is a fact firmly established by abundant experimentation that no crossing-over takes place in the male of Drosophila. Not enough evidence has yet been obtained in other forms to indicate whether the lack of crossing-over is a general phenom- enon in males that display the XY type of sex-inheritance, but it is highly probable that such is the case. In the silkworm moth which may be assumed to follow the WZ type of sex-inheritance, Tanaka has studied very thoroughly the linkage relations exhibited by a system of 210 GENETICS IN RELATION TO AGRICULTURE quadruple allelomorphs, the factors for striped, moricaud, normal, and plain larval pattern, and a pair of factors for yellow and white cocoon color. Sturtevant has pointed out that the experimental results are explicable if there is no crossing-over in the female, the sex-heterozygote in this case. Typical results are given by crosses involving striped, Pg, and plain, p, larval patterns and yellow, F, and white, y, cocoon colors. Thus striped yellow {Ps Y) {P^ Y) crossed with plain white (py) (py) gives in Fi striped yellow individuals of the constitution (PsY)(py) in both sexes. When Fi males were crossed back to plain white females, there were obtained 2907 individuals of which 865 were striped white and plain yellow, which are the cross-over classes. This gives a value of 29.8 per cent, for cross- ing-over in the male. Similarly striped yellow males of the genetic con- stitution {P,y){pY) crossed back to plain^ whites gave 488 individuals of which 151 were striped yellow and plain white, which are the cross- over classes in this case. The value for crossing-over in this latter case is 30.9 per cent., substantially in agreement with the previous calculation. These results are to be compared with those obtained by crossing back striped yellow females of the genetic constitution {PsY){py) to plain white males. From such crosses 1183 offspring were reared all of which were either striped yellow or plain white, consequently non-cross-overs. In both types of sex-inheritance, therefore, no crossing-over occurs in the sex-heterozygote. However, in plants which have the male and female organs in the same individual, crossing-over takes place both in the formation of pollen grains and ovules. The relations exhibited in sex-determination in some insects are extremely complex and present many differences from the simple types which have been described above. Much painstaking cytological in- vestigation has been done in determining these intricate relations with results which for the most part confirm our general observation as to the essential role played by the chromosomes. One of the simplest cases is that of the honey bee. As is well known there are three forms of the honey bee; the queens, the drones, and the workers. Worker bees are females with their sex organs undeveloped as a result of the kind of food furnished them during the larval state. Queen bees lay fertilized or unfertilized eggs. From the former, worker bees and queen bees develop according as to whether they are provided with royal jelly in the larval stage. Unfertilized eggs on the other hand always give drones. Along with these observations it should be noted that under exceptional con- ditions worker bees lay eggs and these always develop into drones. From a chromosome standpoint, therefore, queen bees and worker bees 1 This cross as reported by Tanaka actually involved moricaud larval pattern not plain larval pattern, but, as previously stated, it has been proved that the factor for moricaud occupies the same locus as the factor for plain. INHERITANCE OF SEX AND RELATED PHENOMENA 211 possess the diploid number of chromosomes and drones the haploid number. By moans of experimental investigations on the sex ratio A. F. Shull has recently shown that sex-determination in the mullein thrips, Anthothrips verbasci, is accomplished by the same method as in the honey bee, i.e., females have the diploid number and males the haploid number of chromosomes. Morgan has worked out in detail the complex type of chromosome relations obtaining in the inheritance of sex in the hickory phylloxeran, iBt Generation XxXx XxXx' stem Mother Female Proaucing Line Stem Mother-Male ProJuciug Line. I'olar Spindles of Stem Mother's Eggs 2nd Generation XxXx Migrant-Female Prodocer XxXx Migrant-Male Producer XXxx' Female Egg Male Eggs XxXj Female 3rd Generation Xx Xx' Male-Type I Male-Type II Female Producing Sperm _^^ \/ \/ Polar Spiudle Xx Sexual Egg Fig. 95. — Diagram to illustrate the chromosomal cycle of Phylloxera carycecaulis {After Morgan.) Phylloxera carycecaulis. The life cycle of this insect with respect to the chromosome cycles is shown in diagram in Fig. 95. There are eight chromosomes in this phylloxeran and of these four appear to be con- nected with the determination of sex. They are the only ones illustrated in the diagram. Beginning with the stem mothers at the top of the diagram, these emerge in the spring from fertilized eggs. They immedi- ately attach themselves to the hickory leaves, thereby causing a gall to be formed around them, and in this gall they lay their eggs. As 212 GENETICS IN RELATION TO AQRICULTURE shown in the diagram, these eggs extrude a polar body, but the division is not reductional for the eggs all have four sex-chromosomes, the same number as the mother. These eggs hatch without fertilization into the winged migrant females. Of these there are two kinds, those which lay large eggs and those which lay small eggs, and, moreover, all which come from the same gall and, therefore, from the same stem mother lay the same kind of eggs. Accordingly the stem mothers are of two kinds with respect to their chromosome content as illustrated in the diagram. The female producing stem mothers are XxXx and the male producing stem mothers are XxXx', and the migrant females have the same chro- mosome content as the stem mother from which they were derived. The migrant female of the type XxXx produces large eggs which throw off a polar body, but do not undergo reduction. The resulting egg de- velops without fertilization into a minute sexual female. The other type of migrant females, however, lays small eggs in which, prior to extrusion of the polar body, the large X's and the small x's conjugate. One of each of these pairs then passes out into the polar body, so that two types of eggs are produced Xx and Xx' and these develop without fertilization into the minute males. In the sexual females a true reduction division takes place so that her single egg is of the chromosome constitution Xx. The males on the other hand produce sperm cells half of which are Xx or Xx' according to the type of male and half of which have none of the sex-chromosomes. Sperms of this latter type degenerate, so that only female producing sperm remain. When these fertilize the sexual egg the resulting. eggs are either XxXx or XxXx', and give rise to the corre- sponding type of stem mother. This completes the complicated life cycle in this form, and illusti'ates again the close dependence of sex- determination on chromosome content. In plants only a few cases of sex-inheritance have been studied and these for the most part inadequately. Two of these, namely Bryonia and Lychnis, appear to display the XY type of sex-inheritance, but in a somewhat modified form. Thus Correns crossed Bryonia alba, which is monoecious, with Bryonia dioica, which is dioecious. The former species as a rule bears male and female blossoms on the same inflorescence, the female above and the male below, whereas the latter species con- stantly bears all male or all female blossoms on the same stem. Correns summarizes his results under four heads as follows: 1. Female plants of Bryonia dioica pollinated by male plants of the same species give approximately equal numbers of male and female plants. 2. Female plants of Bryonia dioica pollinated by Bryonia alba give only female offspring. 3. Bryonia alba pollinated by male plants of Bryonia dioica gives approximately equal numbers of male and female plants. INHERITANCE OF SEX AND RELATED PHENOMENA 213 4. Bryonia alba self-pollinated gives only monoecious plants. If we assume that all the pollen grains and ovules of Bryonia alba are of one kind which is indicated by the fact that it breeds true to the monoecious condition, then there is no escape from the conclusion that female plants of Bryonia dioica produce only one type of ovule but male plants produce two types of pollen grains. Unfortunately as is often the case in interspecific hybrids, the Fi of this cross is sterile and con- sequently the analysis cannot be carried further. Shull, however, has studied the inheritance of sex in Lychnis dioica which is normally dioecious but occasionally produces hermaphroditic plants. Although this case has not yet been fully analyzed, the results thus far indicate clearly that the male is heterozygous with respect to a sex-determiner, and the female homozygous. The results of Shull's investigations may be stated under several definite heads as follows : 1. Females with pollen from males give substantially equal numbers of male and female offspring. 2. Females with pollen from genetic hermaphrodites give equal numbers -of hermaphrodite and female offspring. 3. Females with pollen from somatic hermaphrodites give equal numbers of male and female offspring. 4. Genetic hermaphrodites selfed give equal numbers of hermaph- rodite and female offspring. 5. Genetic hermaphrodites with males give equal numbers of male and female offspring. 6. Females from whatever source are genetically identical. Thus females from the cross female X hermaphrodite transmit the same sex- determiners as females from the cross female X male. 7. In crosses between female and hermaphrodite a. small percentage of mutant males always appears and in crosses between female and male approximately the same percentage of mutant hermaphrodites appears. In the above resume of the experimental evidence on sex-determina- tion in Lychnis, the equality of sexes was only approximate, in fact females usually occurred in excess, and sometimes in considerable excess. Shull has interpreted this evidence to indicate that in Lychnis the hermaphroditic condition results from a modification of the male con- dition, and that this modification is reversible as shown by the evidence in 7, above. Interpreted in terms of the XY type of sex-inheritance then, females are XX; males, XY; and hermaphrodites, XY'; and the change from Y to Y' is reversible. Clearly the results indicate that males and hermaphrodites are heterozygous with respect to the sex-determiner, and females homozygous, although later investigations which have not yet been fully interpreted indicate that some disturbing factors are at work, at least in certain cases. 214 GENETICS IN RELATION TO AGRICULTURE Shull's conclusions are further supported by evidence from the inheritance of a sex-Hnked character in Lychnis, the only sex-linked character thus far known in plants. The character in question is that of narrow rosette leaves as distinguished from the normal broad type of leaf, and there are other associated character differences (Fig. 96) . The narrow-leaved form, called angustifolia, was discovered by Baur as a single male mutant individual, a significant fact when taken in connection with its subsequent behavior. The factors in this case are B for the broad-leaved condition and h for the narrow-leaved condition. Crosses between typica females (BX)(BX), and angustifolia males ibX)Y Fig. 96.^ — Adult rosettes of Lychvis dioica; on the left a plant of the normal form, typica; on the right a plant of the narrow-leaved form, angustifolia. {After Shull.) gave in Fi all broad-leaved plants (BX) (bX) females and (BX) Y males. Heterozygous broad-leaved females (BX)(bX) mated to broad-leaved males (BX)Y gave all broad-leaved females, and approximately equal numbers of broad-leaved and narrow-leaved males. Hermaphrodites were also found to behave the same way with respect to the factor B as did the males, which confirms the hypothetical relation supposed to exist between hermaphrodites and females. The evidence clearly indicates the existence of sex-Hnkage of the kind called for on the assump- tion that Lychnis exhibits the XY type of sex-inheritance. Secondary Sexual Characters. — Secondary sexual characters are those which appear as an invariable or almost invariable accompani- ment of a particular sex in most animal forms. They include many diverse things, such as the antlers in male deer, the horns of the males of some breeds of sheep, the mane of the lion, the power of song of many INHERITANCE OF SEX AND RELATED PHENOMENA 215 birds, and various fantastic, ornamental, and combative characters, usually confined to the male. Much historical interest attaches to secondary sexual characters because of the attention directed to them by Darwin's theory of sexual selection. With that we have no particular concern in the present chapter, but shall only consider the inheritance of them in one form as it is related to the inheritance of sex. In the foregoing discussion no particular reference has been made to sex-factors, because after all so little is known concerning them. In some cases we have found sex accompanied by differences in chromosome content, one sex containing an equal pair of chromosomes which are represented in the opposite sex by an unequal pair, in another case the difference in sex appears to depend upon whether the individual possesses the haploid or diploid number of chromosomes. We have also noted that there are two different types of sex-inheritance, one in which the male is heterozygous and the other in which the female is heterozygous. It is only fair to conclude, therefore, that until more light is thrown upon these matters, the assumption that sex-determination depends upon a sex-factor rests on a rather slender basis. The experimental evidence, it is true, is strictly analogous to certain types of Mendelian inheritance, and an interpretation of the sex-factor may be given which does no violence to our ideas of the complexity of sex-differences. Thus it has been shown by ample evidence that the color of eyes in Drosophila depends upon the cooperation of a number of different factors; we can- not say definitely how many, but mutational changes have indicated that over twenty-five different loci have something to do with the re- actions concerned in pigment production in the eye. Yet in spite of this fact the presence of a single factor may make all the difference between a red eye and a white eye. Similarly the sex-factor may act in conjunction with a whole series of other factors, yet the difference dependent upon its presence in the homozygous or heterozygous con- dition may make all the difference between the two sexes. At least in one form, however, we have even more definite evidence of the presence of a definite sex-factor. Shu 11 has shown in Lychnis that where males are expected, hermaphrodite mutants occasionally appear. If we offer the same explanation for the occurrence of these mutants as we have offered for the occurrence of mutations in Drosophila, a change in a single locus in the hereditary system, then the appearance of these hermaphrodites might be offered as almost conclusive evidence of the sex-determining action of a single sex -factor in this particular case. The evidence here becomes even stronger when we consider the fact that this particular type of change is reversible. Some additional light may be thrown upon this question by the consideration of secondary sexual characters as related to the inheritance of sex, although thus far the 216 GENETICS IN RELATION TO AGRICULTURE evidence has not admitted of an entirely satisfactory interpretation. We shall consider one case, that which Goldschmidt has investigated in Lymantria as an example of the results obtained by investigations of this kind. Goldschmidt's investigations are concerned with Lymantria dispar, the European gypsy moth, and L. japonica, its Japanese form. As may be seen from Fig. 97, Lymantria is strongly sexually dimorphic, the females are much lighter in color and larger than the males ; japonica is somewhat larger than dispar, but otherwise in general agrees with it. Goldschmidt's investigations deal with the production of intersexual forms in crosses Fig. 97. — Typical forms and hybrids of Lymantria; 1 and 2, male and female of L. dispar; 3 and 4, male and female of L. japonica; 5-16, hybrids combining male and female characters. (After Goldschmidt.) between these two species. He has shown that with proper combinations of different races of these two species, intersexes may be produced which occupy all possible intermediate positions in a continuous series in which maleness and femaleness are the two extremes. Thus female intersexes, i.e., individuals which are of the chromosome constitution WZ, may be obtained which range from those that show only a very slight develop- ment of male characters in the feathering of the antennae to those which are so nearly males that they show only a faint trace of their female origin in a few minor characters. On the other hand, male intersexes of the chromosome constitution ZZ may be produced ranging from those which exhibit a few white flecks on the wings up to those which INHERITANCE OF SEX AND RELATED PHENOMENA 217 have gone about three-fourths of the way toward the assumption of the entire set of female characters. Goldschmidt ascribes these results to differences in potency of the sex- factors. The European gypsy moth was found in all races to possess sex- factors of low potency, whereas in the Japanese races the potency was in general higher, but ranged from the lowest to the highest condition. Thus males of a moderately strong Japanese race mated to females of a Japanese race of slightly less potency give in Fi very low-grade female intersexes. When mated to a somewhat less potent Japanese race a higher grade of female intersexualism results, and when mated to the weakest European race nothing but high-grade female intersexes are produced. The highest grade of female intersexualism, the transforma- tion of those individuals which are genetically females entirely into males, results from matings of females of European races of the lowest potency to males of Japanese races of the highest potency. Now if the development of sexual characters depends upon the sex-factors acting in conjunction with other elements in the genotype, the existence of sex- factors or rather of systems of factors might operate in somewhat the following fashion. In the female the sex-factor in a heterozygous condi- tion acts in conjunction with a set of factors some of which are perhaps sex-linked, although the number of chromosomes, 62 in this case, would indicate that perhaps most of them were located in other chromosomes. In a heterozygous condition then a certain sex-factor with those factors with which it acts produces a female with the female set of secondary sexual characters. In the homozygous duplex condition the same sex- factor, presumably acting in conjunction with the same set of factors as in the female, produces a male with the male set of secondary sexual characters. If now there should be variations in the potency of a sex- factor, as Goldschmidt assumes, then a strong sex-factor, or a sex-factor which would interact more effectively in a given genetic environment would have a tendency in the heterozygous condition to throw the reac- tion more in the direction of that formerly conditioned by the existence of the normal sex-factor in the homozygous condition. Such relations would result in the formation of female intersexes, individuals genetically females so far as the chromosome constitution is concerned, but develop- ing male characters in a degree corresponding to the greater potency of the introduced sex-factor as compared with the sex-factor normal for the race in question. In case the introduced sex-factor, along with the factors with which it normally interacts and which must never be dis- regarded, equals in sex-determining power that of the normal sex-factor in the duplex condition, then we might expect to get males of the chromo- some constitution WZ. This appears actually to be the case in certain of the experiments. Similarly a weaker potency of the sex-factor might 218 GENETICS IN RELATION TO AGRICULTURE be conceived to result in the production of male intersexes, i.e., individuals of the chromosome constitution ZZ which display female characters, because the weaker potency simply means a more or less close approach to the potency of the normal factor in the heterozygous condition and a consequent approach of the individual to the characters of the male. Goldschmidt's results are intensely interesting and promise much for an elucidation of the problems connected with sex-determination. We cannot refrain from drawing a comparison between these re- sults and some which have been secured in species crosses in Nicotiana. Thus a definite factor for calycine flower in Nicotiana tahacum causes the flowers to develop a petaloid calyx and a split corolla, a striking terato- logical form. The character is a simple recessive to the normal form in variety crosses but when crossed with N. sylvestris, a different species, the normal flower factor in N. sylvestris appears to possess a lower potency than that of normal flowered varieties of N. tahacum. Con- sequently the hybrids are intermediate with respect to the flower char- acter expression, all of the flowers on a given plant exhibiting some development of the calycine flower character. We interpret this to indicate that the normal flower factor of N. sylvestris does not interact normally with the set of factors which interact to determine the floral character expression in the hybrid, but that the calycine flower factor is able to interact normally and to its full extent with these factors. As a consequence the flowers of the Fi hybrid are strongly calycine. This interpretation is further supported by the fact which we have previously set forth in some detail that practically the entire set of characters are determined by the N. tahacum parent. It is conceivable that crosses with other species would show the same character of variability in po- tency as has been found for the sex-factors of Lymantria. At any rate a close analogy here exists between the behavior of sex as a character and the behavior of a character known to depend upon a simple factor difference. The evidence which has been presented with reference to the determi- nation of sex lends strong support throughout to the idea that sex-de- termination depends on the genotypic constitution of the individual. This does not, it must be clearly understood, mean that other external factors may not act to disturb the usual relations just as they occasion- ally do with other factors; but as in such cases these external factors must act in conjunction with the genotypic sex-factors. To assume that changes occur willy-nilly in the case of sex-factors is no more warrant- able than to assume that other factors change frequently in response to environmental conditions, an assumption that does violence to the high degree of stability which has been observed to obtain for factors in general. CHAPTER XII SPECIES HYBRIDIZATION In the preceding chapters an attempt has been made to show how character differences in a large number of plants and animals may be interpreted on the basis of differences in the unit factors which are dis- tributed to the germ cells during gametogenesis. The character differ- ences, however, which were analyzed, although often seemingly complex, were really rather simple, for rarely were more than four or five factor differences taken into account. In a few species of plants and animals the number of factors which have been investigated is considerable, but when compared with the number of factors which must constitute the entire hereditary material of a species it is an insignificant fraction of the total. The analyses which have been presented, therefoje, are for forms which possess an enormous number of factors in common. The differ- ences which they display are mostly unessential alterations in scattered loci in these systems. With the taxonomic question as to what constitutes a species differ- ence, we are not greatly concerned. It is clearly apparent that species as they have been named represent widely divergent differences with respect to the extent of separation from related species. It must be clear to the geneticist, therefore, that specific difference is a variable thing, sometimes meaning one thing, sometimes another. If we look at the question from the standpoint of the number of factors involved, we see clearly that races of plants and animals may differ in one or many genetic factors. Just where the line should be drawn which distinguishes varieties, forms, species, etc., would therefore appear to be almost wholly an arbitrary matter, usually to be decided from considerations of con- venience. Whatever it is, however, the distinction cannot well be viewed from the genetic standpoint, for ordinarily the systematist works with plants and animals which have not been investigated in such a fashion, and, in the case of the more widely separated forms, with those which cannot be so investigated. A genetic investigation of the difference between two species depends upon the possibility of crossing the species in question, and further upon the possibility of securing offspring from the progeny of such a cross. Not infrequently this latter condition is not fulfilled, for it often follows as a result of species hybridization that the individuals thus produced, 219 220 GENETICS IN RELATION TO AGRICULTURE although vigorous and normally developed, are totally sterile. The mule is a familiar example, many others could be given, but they will be con- sidered elsewhere along with the problem of sterility in species hybrids. For the present we shall consider one of the simpler cases in which the species hybridized, although differing very markedly in morphological characters, produce hybrids which appear to be fully fertile. Species Hybrids in Antirrhinum. — Baur crossed the wild Antirrhinum molle with the common garden snapdragon. Antirrhinum majus, and Baur and Lotsy have made extensive studies of the progenies obtained in successive generations of this cross and of other species hybrids in Antirrhinum. Antirrhinum majus and A. molle differ strikingly in a large number of morphological characters. The size proportions and general characteristics of the common snapdragon of the garden are fa- mihar to eVeryone. It is a strong growing erect herbaceous plant, about three feet high producing spikes of large zygomorphic flowers. Under the careful attention of commercial seedsmen it has produced a very large number of varieties which differ in the form and color of the flowers, in height and in other characteristics. Antirrhinum molle on the other hand is a low growing prostrate plant which is profusely branched and produces flowers about one-third as large as those of majus, but very like them in form and general appearance. The species differs from majus also in being apparently totally self-sterile, so that with respect to their genetic constitution plants of molle are normally heterozygous to some extent. Since 7nolle occurs in nature in a number of slightly different forms its self-sterility must not be lost sight of in interpreting the results of hybridization between it and majus. Fi of the reciprocal crosses molle X 7najus and majus X molle are completely self-fertile, and identical in every respect. Minor differences did occur but they were of such a nature that they could be accounted for as a result of the slight degree of heterozygosis of the particular plant of molle which was used as a parent. Baur employed a peloric majus for crossing with molle in order that he might follow a known factor difference throughout the investigation. The Fi plants in this experiment bore zygomorphic flowers, a fact which indicated a corresponding behavior as regards dominance for the factor for zygomorphic flowers in molle and majus. Six Fi plants differing slightly in their characters were selected as parents for the F2 generation. Lotsy grew the progeny of five of these, obtaining from them 624 Fo plants. The general conclusions which Baur and Lotsy have drawn from a study of these F2 plants is that the extreme range of forms dis- played, so great that no two plants resembled each other in all their characters, is a result of Mendelian segregation and recombination of characters; The diversity, however, was so great as to preclude the application of any exact factor analysis to the case. SPECIES HYBRIDIZATION 221 jPjQ 98. Flower types obtained in F- of a cross between Antinhinuin maju.-i (peloncj and A. molle. {After Lotsy.) Fig. 99. Beginning at the left, a peloric majus of the type used lacrosses with molle; a plant of molle; a plant much resembling molle obtained in F2; and on the extreme right the F3 progeny of such a plant. {After Lotsy.) 222 GENETICS IN RELATION TO AGRICULTURE In one F2 population of 255 plants Lotsy was able to distinguish about twenty-five different flower types as shown in Fig. 98. The flower types were not distinct, but represented merely different steps in an almost continuous series, save for the discontinuity incident upon the sharp segregation of a group of plants which bore peloric flowers. More- over, within any of these flower types the plants differed greatly in a number of other characters, such as size, color of flower, form of leaf, habit of growth, etc. As regards fertility there was segregation into self -fertile and self-sterile plants, the former being in the majority. Of the 255 plants, 135 produced zygomorphic flowers, 119 peloric flowers, and one plant produced both zygomorphic and peloric flowers. In color the flowers on different plants ranged from the deep red of the majus parent to the pale color of molle. Lotsy also grew several Fs populations. One of these from an F2 plant bearing hooded zygomorphic flowers consisted of 209 plants all of which were different, indicating again an extreme condition of hetero- zygosity. Not a single plant produced flowers displaying the hooded character of the parent plant. There was again a vast array of flower forms, twenty-three different types feeing represented. With respect to the peloric condition, 113 plants bore peloric flowers only, 94 zygo- morphic flowers, and 2 bore both types. Although several different colors were represented, Lotsy was able to arrange them in two classes; the first consisting of 153 plants approximating the red color of majus, and the second group of 56 plants of about the color of the pale molle parent. There was, therefore, a fair indication of Mendelian segrega- tion for color in this generation. In this population as shown in Fig. 99 a plant was obtained which very closely resembled the type of A. molle in all its characters, and re- produced these characters in its progeny. Other plants were obtained which strongly resembled majus in certain of their characters, but not so completely throughout. The important feature here is the fact that even in F2 segregation and recombination of factors have produced a plant which is practically identical with one of the parents. Other F3 populations were grown from F2 plants displaying different sets of characters. One of these from a zygomorphic F2 plant produced a population segregating for color and form (zygomorphic vs. peloric) of flower. Another population from a peloric F2 plant consisted entirely of peloric flowering plants, but in this population there were many different color classes. An F2 plant of the pale color of molle gave an F3 popula- tion consisting entirely of pale-flowering plants but showing segregation in form and for the peloric character. Obviously if results such as these are to be explained on a Mendelian basis, it must be assumed that a relatively large number of factor differ- SPECIES HYBRIDIZATION 223 ences exist between the two species under consideration. When we observe the number of differences in habit, form, size, etc., which are known to obtain between the two species, this assumption does not appear to do violence to actual facts in the case. Baur has sought by systematic hybridization investigations to determine which of the known factors of the hereditary material of A. majus are also contained in that of A. molle. From these investigations he concludes that A. molle certainly possesses the factors indicated by the incomplete formula BBDDEEFFll, in which B represents a factor for yellow flower color; D, a factor for extension of pigment to the tube of the corolla; E, the factor for zygomorphic flowers; F, a base factor for red flower coloration which is epistatic to B; and I, a recessive factor which determines a low intensity of flower coloration. His success in determining the presence of these factors in the hereditary material of A. molle has led Baur to conclude that it is entirely within the range of possibility to analyze completely the differences which exist between these two undoubted species. All the unusual flower forms, therefore, which are obtained by crossing them are to be regarded as the results of peculiar factor interactions. We have pointed out in previous chapters that it is not always possifcle to predict the character expression of a given set of factors from a knowledge of their known expression in certain combinations. That this condition is here operative is borne out in part by the fact that certain flower types which appeared in F2 did not reappear even among fairly large numbers in the F3 generation from such F2 plants. We consequently can state with assurance in spite of unsatisfactory ratios and peculiar character expressions that the results obtained in this species cross may reasonably be interpreted in harmony with Mendelian doctrine. Detlefsen's Cavy Hybrids.^ — ^A similar line of investigation in animals has led Detlefsen to similar conclusions. He crossed the tame guinea- pig, Cavia porcellus, of which many different races have been produced under domestication, with the wild C. rufescens. The latter differs from the tame guinea-pig in a number of respects. It is very much smaller, weighing about half as much as the tame guinea-pig, and in skeletal measurements and other characters it is definitely set off as a distinct species from C. porcellus. In color it is of the agouti type common to all wild rodents, but the agouti differed from that of the tame guinea- pig in having less power to exclude black and brown from the hair than has the agouti of the tame animals, consequently individuals of the wild C. rufescens have darker coats than those of the tame porcellus. By crossing C. rufescens and its hybrids with porcellus with various races of porcellus, Detlefsen was able to study the inheritance of the following factors in this species cross : A — the agouti factor, which operates by restricting the black or brown 224 GENETICS IN RELATION TO AGRICULTURE pigment in the hairs thus producing the gray or agouti pattern. There are variations in the regional distribution of the restrictive action. The allelomorphic condition a gives self-colored individuals. B — the factor for black. The allelomorph h conditions a brown col- oration instead of black. C — the basic color factor in rodents. The allelomorphic condition represented by c gives albinos. E — a factor conditioning the extended type of pigmentation of self- black or brown animals. The allelomorph e gives the black-eyed or brown-eyed red or yellow coat. R — the factor for rough or rosetted coat, as distinguished from the smooth coat determined by the allelomorph r. The work of a host of investigators has demonstrated beyond question the Mendelian inheritance of these factors in races of the tame guinea- pig. Castle in particular has demonstrated how these factors behave in Mendelian fashion, one among the first investigations estabhshing the general validity of Mendelian principles. Moreover, these conclusions have been abundantly confirmed by investigations with other rodents, which appear to possess a closely analogous series of color factors. Detlef sen's experiments were conducted by crossing tame female guinea-pigs to wild males, and then mating back the hybrid females to tame male guinea-pigs. This was necessary because the male hybrids were sterile until back crosses to the tame guinea-pigs had been made for two or three generations. Crossing back to the wild species was impos- sible on account of the scarcity of wild animals and their failure to breed freely under domestication. The investigations were carried through eight generations, during which many types of matings were made, and a total of 1160 hybrids were reared and studied. As a result of these investigations Detlefsen concludes that the wild rufescens is of the constitution AABBCCEErr with respect to the factors noted above. Moreover, the relation of these factors as respects domi- nance and segregation was throughout identical with the relations displayed in intervarietal crosses in the tame guinea-pig. Recombinations of factors occurred in the normal fashion so that it was possible to secure hybrids showing any type of coloration found in the tame guinea-pig. The conclusion, therefore, that interspecific crosses between C. porcellus and C. rufescens display complex Mendelian inheritance appears to be established by these investigations. It may be pertinent, however, to enquire whether homologous factors normal for the two species are really identical. If we assume that the two species possess similar genetic constitutions, i.e., have similar sets of chromosomes bearing the factors in like arrangement, it is entirely con- ceivable that, although the formal arrangement of factors in the heredi- SPECIES HYBRIDIZATION 225 tary material might be the same for the two species, the actual factors them- selves might differ in certain respects, for example in the exact type of character expression and in their power to react with a given set of factors. If the differences be relatively slight, the factors might still be able to interact with each other approximately in the normal fashion, and to display allelomorphic relations dependent upon their position in the hereditary material. On this point Detlefsen contributes very important data which we shall consider somewhat in detail. The first set of observations relates to the differences between the agouti factors of C. porcellus and C. rufescens. It is a common obser- vation that the agouti pattern in rodents in general is a variable one. Some of this variability is unquestionably due to the presence of modi- fjdng factors, but not all such variations can be interpreted in this fashion. Elsewhere we have pointed out that in mice a system of quadruple allelomorphs includes the factors for yellow, black, gray, and gray with white belly. In the rabbit, Punnett's results may be interpreted as establishing the existence of a triple system of multiple allelomorphs consisting of the factors for yellow, agouti, and black. Similarly in the tame guinea-pig there are apparently allelomorphic variations which affect the agouti pattern, but Detlefsen finds, nevertheless, that these never condition the type of agouti presented in C. rufescens. Detlefsen points out that agoutis in common restrict black or brown in the sub-apical band of individual hairs so that the dorsal hairs present a barred appear- ance. More powerful restriction is shown in the hairs of the belly, but there is always a close correlation between the amount of restriction in dorsal and ventral regions, for the darker the dorsal region, the darker is the pigmentation of the ventral surface. The wild agouti factor was distinguished by its weak restricting power, so that ordinarily the yellow sub-apical band in the hairs of these animals was distinctly narrower than in some agouti guinea-pigs. In some cases the lack of restriction was so marked that only a slight sprinkling of agouti hairs in the adult gave evidence of the existence of the agouti factor. Moreover, in some cases the wild agouti pattern carried with it a ticked belly, a condition appar- ently unknown in the tame guinea-pig. Some variation was observed in the agouti patterns of the original individuals of C. rufescens and this must not be forgotten in interpreting Detlefsen's results. Dark agoutis produced by constantly mating wild agouti hybrids to tame non-agouti guinea-pigs, were mated to tame agouti animals. We may represent the factor for wild agouti by A', that for tame agouti by A, and that for the allelomorphic condition in the tame guinea-pig by a. Following this formula, then those dark agoutis produced by mating wild hybrid agoutis to tame non-agoutis must have been of the genetic constitution A' a. When such animals are mated to tame agoutis two types of animals 15 226 GENETICS IN RELATION TO AGRICULTURE are produced, those of the genetic constitutions AA' and Aa respectively. Phenotypically these two classes of individuals are exactly alike for the powerful tame agouti factor is alike dominant to the wild agouti factor A' and to the tame non-agouti factor a. When the individuals of this population were bred to tame non-agouti animals of the genetic consti- tution aa the existence of the two above-mentioned genotypes was clearly demonstrated, for half the individuals gave progenies exhibiting sharp segregation into light tame agoutis and dark wild agoutis in approximately equal numbers and the other half gave progenies consisting of approxi- mately equal numbers of light tame agoutis and non-agoutis. Other tests satisfactorily supported this analysis so that it may be concluded that the agouti factor of the wild C. rufescens is different from the agouti factor of the tame C. porcellus, but that they are allelomorphic to each other. If we consistently follow up the hypothesis which we have devel- oped as to the constitution of the hereditary material and the operation of the chromosome mechanism, this can only mean that the factors for agouti, although different, occupy corresponding loci in the hereditary system of these two species. Aside from certain observations indicating differences between the rough factor of wild and tame guinea-pigs we have no evidence as to whether or not those other factors, the inheritance of which was investigated, are different, but we may safely conclude that factors of corresponding behavior occurred at exactly the same loci in the hereditary system of the wild C. rufescens. Evidence as to the difference between the two agouti factors is also provided by the irregular behavior of the wild agouti factor in the hybrids. Although the first hybrids between the wild agouti and tame non-agouti guinea-pigs are mostly of the dark wild type with ticked bellies in sub- sequent generations there appear agoutis which are so light as to approach closely the light agouti type of the tame parent and others are so dark that the individuals show only a slight sprinkling of agouti hairs. Indi- viduals displaying such modifications of the wild agouti pattern show no very regular type of behavior, for dark individuals sometimes produce some light individuals and the light individuals sometimes produce some dark ones. The dark modification, however, is most common and often becomes more pronounced upon successive dilutions with tame blood. An interpretation of such phenomena cannot be made satisfactorily unless we consider the agouti factor as a member of a com- plex system of factors which together operate to give the agouti type of coloration. From this standpoint it is not at all strange that the wild agouti factor acting in conjunction with a corresponding system of factors mostly derived from the tame guinea-pig should exhibit the full power of its customary restrictive action because of a failure to set up wholly harmonious relations with these factors. This modifying SPECIES HYBRIDIZATION 227 of the wild agouti pattern, therefore, lends additional support to the conclusion that these two agouti factors, although occupying homologous loci in the hereditary systems of the two species are different from each other. The Forms of Species Hybrids. — Thus far we have dealt with two species crosses which have given satisfactory indications of behavior essentially in accord with generally accepted Mendelian principles. The remainder of the chapter will be devoted to general considerations respecting species hybrids and to particular cases which do not give entirel}^ satisfactory evidence of Mendelian behavior. In common with most variety hybrids, species hybrids display marked uniformity in the first generation and equality of reciprocal crosses. Exceptions, however, occur to both these conditions and these we shall take up later in the discussion. With respect to the characters which they display species hybrids usually represent an intermediate condition as compared with the parents. We may refer this condition to a mixture of dominant factors derived from both parents and in* some cases to actual intermediate expression of contrasted allelomorphs, as is not uncommon in variety hybrids. The intermediacy of Fi in species crosses is a well-known phenomenon and is so common that it may be regarded as the rule. This condition was well known to the older hybridists, such as Kolreuter, Gartner, Naudin, and Focke, all of whom investigated extensive series of species hybrids with respect to the characters both of the immediate hybrid and of its progeny. The intermediate condition* however, is not universal, for examples are known of all conditions from that of strict intermediacy to a condition so nearly resembling one parent in certain cases that only slight character differences or sterility establish the existence of an actual cross. Intermediate species hybrids are so common that it seems super- fluous to call especial attention to them, nevertheless this will be done in order to point out the relation of the intermediate condition to other characteristic features of species hybridization. In the first place the intermediate condition is not associated with any particular degree of fertility in the hybrids. Partial sterility is a common characteristic of wide crosses, and in fact this sterility in some cases appears to be complete. The Antirrhinum species hybrids are intermediate in practi- cally all characters, but they are apparently completely fertile. Such cases are, however, uncommon in species hybridization, but neverthe- less a few others have been studied. Baur and Lotsy have reported other species hybrids in Antirrhinum which give fertile intermediate hybrids. Some species hybrids in Nicotiana are known to be very nearly com- 228 GENETICS IN RELATION TO AGRICULTURE pletely fertile. East has reported investigations of a cross between A'', alata and N. langsdorffii. N. alata has flowers the corolla length of which averages about 82 mm., whereas the corolla length of flowers of A^. langsdorffii averages not over 22 mm,, so that A^. alata flowers are nearly four times as large as those of N. langsdorffii. In addition to these differences there are other distinct differences between the two species. Nevertheless examination indicated that there was little, if any, diminution in fertility in Fi. A few other species hybrids in this group of Nicotiana give highly fertile Fi hybrids, for example A'', alata X A^. sandercB and A'', langsdorffii X A'', sanderce. There appears to be little reason for not regarding these as species hybrids, although it should be stated that some investigators feel inclined to restrict the species concept to forms which display a certain degree of partial sterility in Fi. Such a line of separation must, however, be purely arbitrary since it can be shown that fertile species hybrids merely represent one of the extremes in a continuous series extending from complete fertility to complete sterility. Since partial sterility is such a characteristic feature of species hybridi- zation, it is not surprising to find that diminution in fertility is not associated with any particular kind of character expression in the hybrids. Intermediate hybrids as well as those which more or less resemble one of the parents usually, therefore, display a considerably diminished fertility. Not much has been done with such hybrids for aside from exceptional instances sterility presents at once a bar to their further analysis and to their use for economical purposes. A familiar example, the mule, a cross between Equus cdballus and E. asinus, has given no authentic case of the production of offspring, although produced for many centuries under domestication and in vast numbers. Among plants so many examples occur that it is of no advantage whatever to attempt an enumeration of them here. The student who is particularly inter- ested in such matters will find that excellent compilations of species hybrids in plants have been made by Gartner and Focke; and Ackermann, Przibram, and Rorig have performed a similar service for the animal kingdom. In tobacco a large number of species hybrids occur which give partially sterile intermediate hybrids. The genus Nicotiana had been much employed in hybridization investigations providing as it did the first instance of hybridization in the plant kingdom when in 1760 Kolreuter crossed A'', rustica and A'', paniculata. The hybrid thus obtained was intermediate in its characters and was only slightly fertile. Varying comments have been made as to the exact expression of the characters of this hybrid as compared with those of its parents, but apparently careful scrutiny reveals the influence of both parents in practically SPECIES HYBRIDIZATION 229 every character, although to varying extents in different characters. All, however, have found it relatively infertile, although among some hundreds Lotsy, in one experiment, discovered one plant which possessed a rather unusual degree of fruitfulness. Although the condition of intermediate character expression includes by far the majority of species hybrids, there are some notable exceptions which very closely duplicate the set of characters of one parent almost to the exclusion of those of the other. This fact was recognized even by the older investigators, for Gartner states that any condition may be obtained from that of strict intermediacy to a condition so closely re- sembling one parent as to be distinguished from it only by increased vigor and partial steriUty. Gartner found examples of dominance of one parent particularly striking in some Nicotiana crosses. Thus N. paniculata X N. langsdorffii is reported to give a hybrid form almost indistinguishable from N. langsdorffii and N. suaveolens X N. macrophylla is predominantly N. 7nacrophylla in its character expression. Later in this chapter we shall describe crosses between N. sylvestris and a series of varieties of A^. tahacum which constantly yield hybrids resembhng the particular tahacum variety used in crossing. Curious instances of such predominance of one type are reported for triple hybrids. Thus N. rustica X N. paniculata polhnated with N. angustifola gives plants closely resembling A^. angustifola; if the same hybrid is pollinated with N. glutinosa it produces plants closely resembling N. glutinosa. There are authentic instances of species crosses which do not give equivalent results in reciprocal crosses. It is a common observation that some species crosses may be made in one way only. Crosses be- tween wheat and rye are sometimes successful when wheat is the female parent, but the reciprocal cross has never been obtained. But usually when a cross is possible in both directions the reciprocal hybrids are prac- tically indistinguishable. Among exceptions to this rule are crosses between Digitalis purpurea and D. lutea, strikingly different species, which constantly give hybrids resembling the female parent. In Oeno- thera such results are particularly common, and de Vries and others have investigated a number of such cases. A typical case is that of 0. biennis and 0. muricata which give strongly patroclinous hybrids in reciprocal crosses. The fact, however, that these hybrids breed true in further generations introduces a complication which places us on our guard against the operation of some as yet undiscovered factors. We can under- stand why reciprocal crosses should give different results, when there are differences in chromosome number or content in the two sexes as is generally the case among animals, but in plants it is more difficult to assign a reason for this type of behavior aside from a few cases in which apogamy is known to occur. It is, therefore, necessary for us to accept 230 GENETICS IN RELATION TO AGRICULTURE these cases with some reservations, looking to the future for experimental investigations which will provide us with a satisfactory explanation for them. The Vigor of Species Hybrids. — The increased vigor displayed by species hybrids has been frequently commented upon by investigators from the time of Kolreuter down to the present. In 1849 Gartner in his general treatment of this subject in species crosses especially notes that the luxuriance of hybrids frequently expresses itself in an unusual develop- ment of practically all plant parts. He also cites a considerable num- ber of the earlier hybridists who have noted this increased vigor, among them Kolreuter, Sageret, Berthollet, Herbert, Mauz, and Lecoq. Hy- brids which up to that time had been particularly noted for this sort of vigor represented such a large number of different famihes that there could be no question as to the generality of the phenomenon. For increase in length of stem Gartner notes especially Verhascum lychnites X V. thapsus which grows to a height as great as 15 feet; Althcea cannabina X A. officinalis which sometimes attains a height of 12 feet; Malva mauritiana X M. sylvestris which attains a height of 11 feet; Digitalis purpurea X D. ochroleuca which grows to a height of 10 feet; and finally Peiwnza nyctagini- flora X P. phoenicea and Lobelia cardinalis X L. syphylitica which attain a height of 3 to 4 feet, a significant increase as compared with their low-growing parents. Often the vigor is expressed in a general increase in size throughout as appears to be particularly true of hybrids between different species in the genera Mirabilis and Datura. In Nicotiana a number of hybrids such as N. suaveolens X A^. macrophylla, N. rustica X N. 7narylandica, and many others display such general hybrid vigor sometimes to a very marked extent. Tropceolum majus X T. minus, a hybrid of the tall and dwarf nasturtiums of the garden is another notable instance of hybrid development. Gartner also records many interesting ways in which this hybrid vigor expresses itself. Thus certain hybrids in Dianthus, Lavatera, Lobelia, Lychnis, Geum, and Penstemon while not displaying notable increases in vegetative vigor lend themselves much more readily to vegetative propagation than do their parents. In some cases the hybrids show an unusual tendency to produce side branches and suckers, and in other cases still other outlets of this hybrid vigor are found. Not all species hybrids, however, display hybrid vigor, and many indeed show a strikingly weakened condition accompanied by much less- ened vegetative vigor. In tobacco several species hybrids show less- ened vegetative vigor, as for example, Nicotiana grandiflora X N. gluti- nosa, N. glutinosa X A'^. quadrivalvis, N. rustica X N. suaveolens, and A^. suaveolens X N. quadrivalvis. Similarly Verbascum blattaria X V. lych- nitis gives weakened hybrids. Consequently within the same genus SPECIES HYBRIDIZATION 231 some species hybrids show marked increases in vegetative vigor, whereas others show just as marked decreases. Investigations since Gartner's time have simply extended observa- tions on the comparative vigor of parents and hybrids in species hybrids as well as in the less violent variety hybrids. Thus Focke who inves- tigated large numbers of species hybrids found many that were abnor- mally weak, but these usually represented rather wide crosses. Crosses between more closely related species, however, generally showed an in- creased vegetative vigor. The increased vegetative vigor, he regards as merely an extension of the same condition which Darwin had investigated in variety crosses, namely that crossbreeding is advantageous from the standpoint of the general growth of the forms involved. The idea that sterility may be the cause of this increased vigor is refuted on the one hand by the fact that some of the most vigorous species hybrids are also highly fertile, and on the other hand by the fact that most of the weak hybrid forms are nearly or quite sterile. East and Hayes have attempted to offer an explanation for these phenomena on the basis of heterozygosis. They have reached this con- clusion from extensive investigations of the effect of self-fertilization in maize and of cross-fertihzation in tobacco. In corn they have found, as we shall describe more in detail later, that continued self-fertilization results in the isolation of races which are very uniform as respects their character development, but which almost constantly show considerably decreased vigor of growth. This decrease in vigor is most rapid in the first generations and becomes less rapid as the races become more con- stant in their characters. Since the approach to constancy in char- acters may be regarded as evidence of approach to a homozygous condi- tion in this a normally highly heterozygous species, East and Hayes argue that the normal vigor of maize is largely an expression of its heterozygous condition and that the decrease in vigor is a consequence of reduction to a homozygous condition. This conclusion is in part confirmed by the evidence from crossing such homozygous strains of maize. The Fx of such crosses usually exhibits an immediate return to the vigor of the population from which the strains were isolated. However, it is not entirely clear why this behavior cannot be ascribed to the isolation of races possessing fewer dominant factors than most of the plants in the original population. When such races are crossed the original set of dominant factors would be reunited, and in consequence the normal vigor of the original population would be exhibited. Since the foregoing was written D. F. Jones has published an explana- tion of increased vegetative vigor of hybrids or ''heterosis," as it has been termed by Shull, which he has summarized as follows: 232 GENETICS IN RELATION TO AGRICULTURE 1. The phenomenon of increased growth derived from crossing both plants and animals has long been known but never accounted for in a comprehensible manner by anj^ hypothesis free from serious objections. 2. The conception of dominance, as outlined by Keeble and Pellew in 1910 and illustrated bj^ them in height of peas, has had two objections which were: a. If hetero- sis were due to dominance of factors it was thought possible to recombine in genera- tions subsequent to the F2 all of the dominant characters in some individuals and all of the recessive characters in others in a homozj^gous condition. These individuals could not be changed by inbreeding, b. If dominance were concerned it was con- sidered that the ^2 population would show an asymmetrical distribution. 3. All hypotheses attempting to account for heterosis have failed to take into consideration the fact of linkage. 4. It is shown that, on account of linked factors, the complete dominant or com- plete recessive can never or rarely be obtained, and why the distributions in F2 are symmetrical. 5. From the fact that partial dominance of qualitative characters is a universal phenomenon and that abnormalities are nearly always recessive to the normal condi- tions, it is possible to account for the increased growth in Fi because the greatest number of different factors are combined at that time. 6. It is not necessary to assume perfect dominance. It is only necessary to accept the conclusion that many factors in the In condition have more than one-half the effect that they have in the 2ti condition. 7. This view of dominance of linked factors as a means of accounting for heterosis makes it easier to understand: a, why heterozygosis should have a stimulating rather than a depressing or neutral effect; and b, why the effects of heterozygosis should operate throughout the Ufetime of the individual, even through many generations of asexual propagation. In order to extend their investigations to normally self-fertilized species, East and Hayes made numerous crosses between different species of Nicotiana. As a result they found, as indeed had previous investigators on this the favorite genus for hybridization studies, that the vigor of hybrids varied all the way from a condition so weak as to give embryos incapable of germination to a condition greatly exceeding in vigor that exhibited by either parent. Thus Nicotiana tabacum, the commonly cultivated tobacco, when crossed with A'^. sylvestris gives hybrids which exceed by 35 per cent, the average height of the parents and are estimated to be 20 per cent, more vigorous. Similarly N. tabacum, when crossed , with TV. rustica, gives hybrids exceeding by 80 per cent, the average height of the parents, but when crossed with N. alata grandiflora, the hybrids are only about 10 per cent, of the average of the parents, both in height and vigor. About fifty species crosses within the genus were made by these investigators with results in re- spect to vigor which bore out those already known for various specific crosses. It is obvious from these results that stimulation in the hybrid is a result of certain specific interactions. East and Hayes regard those hybrids which show decreased vigor as evidence of such great differences SPECIES HYBRIDIZATION 233 between parents that normal cell division is impossible. When on the other hand the differences are not great enough to obstruct normal cell division, the degree of stimulation is held to increase directly with the amount or kind of heterozygosis present. These conclusions, however, do not appear to be very firmly estab- lished on the experimental side for by no means the only explanation has been offered. When we consider the recent work with Drosophila it is clear that many factor differences are known which in addition to resulting in some definite character distinction display a rather ill-defined effect in decreasing vigor. Thus the factor for white eye color in addi- tion to determining white eyes has such an effect on the viability of the white-eyed phenotype that in segregation this class never comes up to Mendelian expectations. Similarly other factors have definitely a weakening effect in vigor, in sterility, and in other characteristics. This effect, also, is apparently cumulative, so that in Drosophila strains containing many recessive factors almost invariably must be carried on in a heterozygous condition on account of their low viability. Here very evidently the increased vigor of the heterozygous strains is to be attributed to a recombination of the dominant factors normal to the wild type, for the heterozygous forms display the characters of the normal wild type and a size and vigor approximating that of the homo- zygous wild type. Similarly in corn the occurrence of open pollination makes it possible for a relatively large number of such factors which lower vigor to exist in a variety. These show their effects in marked degree only on self-fertilization for such self-fertilization automatically results in a rather rapid reduction of strains to a homozygous condition. If the number of recessive factors affecting vigor is fairly large then it is evident that the mathematical probability of isolating some of them in continued self-fertilization is relatively great, but the chances that the same ones will be isolated in different pure strains is relatively slight. It follows that usually pure strains resulting from continued self-fer- tilization will display lessened vigor and productiveness, and that dif- ferent strains isolated in this fashion will give hybrids approximating the normal condition of fertility and productiveness. The increased or even decreased vigor of species hybrids of the wider type appears, there- fore, to belong to a distinctly different category for which we are not yet fully prepared to provide an explanation. To suggest that the increased stimulation depends on the specific interactions which occur between two different contrasted hereditary systems is, confessedly, falling back on a less definite explanation, but one which does not appear improbable when viewed in the light of our knowledge of the unexpected relations which certain factor combinations display when brought together. 234 GENETICS IN RELATION TO AGRICULTURE Sterility in Species Hybrids. — A common phenomenon of species hybridization is the marked degree of sterility which is exhibited. This fact of steriUty in speices hybrids has led certain investigators, par- ticularly Jeffrey, to lay great stress upon partial sterility as an evidence of hybrid character. This contention may be valid for a majority of cases, but obviously it would not follow even if all species hybrids dis- played partial sterility, that all cases of sterility are to be referred to hybridity. Specifically many instances are known for which simpler and more satisfactory explanations suffice. Thus Bateson has recorded a case of contabescence in the anthers of sweet peas which is strictly due to the presence of a definite factor for contabescence. The ratios obtained are approximately 3 : 1 ratios, contabescence being recessive, and moreover, the factor for contabescence is definitely linked with other factors, so that in every respect the factor analysis of this case of sterility is wholly satisfactory. In Drosophila similarly some cases of sterility are definitely referable to the action of specific factors which sometimes have effects so marked that strains homozygous for the factors in question cannot be maintained. This condition is well illus- trated by flies which are homozygous for the factor for rudimentary wings. Such flies are practically never fertile. Many other instances are known where slight effects on fertility result from factors which are intimately concerned in the expression of other characters. A some- what different type of sterility, but one which is also definitely established, is that which Bridges has reported for the males of Drosophila ampelophila which lack the F-chromosome. The evidence upon which this case of sterility is based appears to be conclusive, and to demonstrate that while the male Drosophila lacking the F-chromosome may develop a normal soma, it cannot produce functional germ cells. The sterility of wide crosses, however, appears to belong to a distinct category, an explanation for which we shall endeavor to give later on in this chapter. At this point we shall only take up some of the types of sterility displayed in such crosses. At the outset it may be well to note that the degree of sterility dis- played by hybrids varies from complete fertility to complete sterility. It is, therefore, readily apparent that sterility in hybridization as a means of species differentiation gives no natural divisions, but that arbitrary ones must be erected depending upon the degree of sterility displayed. Moreover, other factors such as those noted above com- plicate matters and render it extremely difficult to decide where to draw the line. Here again, therefore, the search for a universal species in- dicator has met with failure. From a genetic standpoint this is as it should be for it merely indicates that races of plants and animals display all degrees of genetic differences from simple differences in isolated factors to complex differences in entire series of factors. SPECIES HYBRIDIZATION 235 Sterility in crosses between apparently good species may be at times almost completely lacking. Thus the crosses between Antirrhinum majus and A. molle and some other crosses made by Baur within the genus Antirrhinum proved fully fertile. The same condition has been found in other species crosses. Thus Nicotiana alata grandiflora and N. langsdorffii, although they differ strikingly in their characters, give hybrids which are about as fertile as the parents. Certain orchid crosses are also reputed to display a high degree of fertility, but on the whole crosses between good species very rarely show even an approxi- mation to the full degree of fertility, and this is true of both plant and animal hybrids. The sterility displayed by species hybrids may not always be equiva- lent in both sexes. Thus one of Baur's Antirrhinum crosses, that of A. majus X A. siculum proved completely sterile as far as the production of good ovules is concerned, but some good pollen grains are produced which can be used in back crosses to the parents. In the case of Cavia porcellus X C. rufescens, we have already noted that the males are sterile and the females fertile. Detlefsen attempted to follow out the inheritance of fertility in this case, and attacked the problem from many angles. The fertility of the females appears to be complete, since the Fi females produce litters of approximately the average number of young of those of the two parent species. The offspring of the hybrid females when crossed back to the tame guinea-pig again produce fertile females and sterile males. With each successive back cross to the tame guinea-pig the percentage of fertile males rises in a fairly regular fashion as is shown in Table XXXVII. Detlefsen points out for this case that the assumption that the wild species carries eight disturbing dominant factors gives a very close agreement with the observed re- Table XXXVII.— Percentages of Hybrid Offspring with Many Mobile Sperm IN Matings of Female Hybrids with Tame Guinea-pigs, and Female Hy'brids with Fertile Male Hybrids {After Detlefsen) Generation Offspring of female hybrids and guinea-pigs Offspring of female hybrids and fertile male hybrids Calculated for of females Number Percentage with many mobile sperm Number Percentage with many mobile sperm eight factors Fi F2 1 8 49 99 150 49 15 00.0 00.0 14.3 33.3 60.7 69.4 73.3 1 2 7 17 11 1 00.0 00.0 14.3 58.8 63.6 100.0 00.0 00.4 Fs Fi 10.0 34.4 F, Fe 59.7 77.6 Fv 88.2 236 GENETICS IN RELATION TO AGRICULTURE suits. The Fi hybrids would then be of the genetic constitution AaBb- CcDdEeFfGgHh. Such individuals produce gametes of the constitu- tion ahcdefgh only once in 256 times, so that when crossed back to tame guinea-pigs which produce only gametes bearing the recessive factors, 0.4 per cent, of the males should be fertile. The percentage of fertile males in successive generations of back crossing should then increase pro- gressively as shown in the last column of Table XXXVII. As Detlefsen himself, however, points out the close agreement of these calculated figures with those actually observed is misleading as an indication of the significance of the analysis, for it is doubtful whether simple segregation of Mendelian factors provides an explanation of the entire phenomena. It is rather strange in fact that only the males display this sterility, and it is of interest to note, as Detlefsen points out, that several other analogous instances of male sterility in animal species hybrids are known. The yak, Bibos grunniens, crossed with the domestic cow. Bos taiirus, gives fertile female and sterile male offspring. Similarly the gayal, Bihos frontalis, the gaur, Bibos gaurus, and the American bison, Bison americanus, have been crossed with domestic cattle and have given fertile female and sterile male hybrids. There is strong evidence that hybrids of the banteng, Bibos sondaicus, and the zebu. Bos indicus, display similar relations. When we consider the physiological relations between factors and particularly the significant fact that probably no crossing-over occurs in the males of this species, we feel inclined to attribute the male sterility to other causes than to a mere sorting of factors having to do with fertility. Partially Sterile Hybrids of Wheat and Rye. — Thus far cases have been considered in detail in which the species hybrids display a consid- erable degree of fertility. At the other extreme stands a series of hybrids which display sterility which is nearly but not quite complete. Such are the hybrids between wheat and rye which Jesenko has subjected to thorough experimental study. There can be no question that wheat and rye are distinct species, in fact they have been universally assigned to different genera. They seem to represent about the extreme limitations of effective hybridization. Jesenko and others have been able to obtain hybrids between wheat and rye only when wheat is used as the female parent, consequently we are unable to compare the results of reciprocal hybridization in this case. Even pollination of wheat with rye is suc- cessful only about six times in one thousand as Jesenko found in over six thousand trials with different species and varieties. The Fi hybrids were intermediate in general characters, although the relations of dominance displayed in variety crosses was preserved in the species crosses. In Fig. 100 is illustrated one of these hybrids and its two parents. The in- creased size of the spike as compared with those of either parent is par- SPECIES HYBRIDIZATION 237 ticularly striking. These Fi hybrids are completely sterile with their own pollen. However, it was possible by pollinating the hybrids either with pollen from wheat or rye to obtain a few viable seeds. For wheat pollina- tion the ratio of success was about 3 in one thousand; for rye only one plant was secured from nearly five thousand trials. The pollen grains of the hybrids were apparently completely non-functional, and cytological examination indicated prevailingly irregular divisions and behavior in their production. The product of back-crossing the Fi hybrids to wheat gave plants •1 J, I ^U 1 '1 1 J 1 ' 'i^ i 1 1 i 1 j • / iV \ ' A B C Fig. 100.— Sterile hybrids be- tween wheat and rye, A, the wheat parent; C, the rye parent, and B, the Fi hybrid between them. (After Jesenko.) Fig. 101. — Sesqui-hybrids from thei^i wheat X rye crossed back to wheat. (After Jesenko.) very similar to wheat. This is illustrated in Fig. 101. Although all these plants resembled wheat in their general characters, they neverthe- less showed wide differences from one another, not only in morphological characters but in physiological ones such as fertility as well. A few of the plants were totally sterile, but some of them were more or less fertile and in general those were most fertile which most closely resembled the wheat parent. In the next following generation, the progeny of those plants which were most fertile consisted of plants which were apparently pure wheat and completely fertile and plants which were less like wheat and showed lessened fertility as the resemblance to wheat decreased. 238 GENETICS IN RELATION TO AGRICULTURE . For a few particular characters, Jesenko was able to establish close ap- proximation to a Mendelian analysis, so that it can scarcely be doubted that in the sorting out of the factors to establish the constant races of further generations, the phenomena displayed were such as to indicate clearly the operation of a Mendelian mechanism. But when we consider the phenomena in the light of the characters involved, then it may be seen that the results obtained are truly remark- able. Wheat and rye differ strikingly in their characters and the recovery of approximately the parental form so often in these back-crosses is out of the question from a strict Mendelian viewpoint, if all combinations are assumed to survive. As an explanation of these phenomena, Jesenko calls attention to the fact that there are eight chromosomes in the germ cells of rye and wheat, so that in the formation of gametes in the i^i some will possess eight wheat chromosomes, others seven wheat and one rye, and so on. When back- crossed to wheat, therefore, union with those gametes which contain only wheat chromosomes or at most two or three rye chromosomes results in wheat-like plants which are fertile, whereas greater proportions of rye chromosomes results in plants which are less like wheat and sterile. Similarly, as Jesenko in fact found, pollination with rye results in plants resembling rye, because of the union of the rye pollen with gametes which contain all or nearly all rye chromosomes. The sterility in these hybrids, therefore, Jesenko regards as the consequence of the inharmoni- ous action of a "plasma" built up of large proportions of both rye and wheat elements. Partially Sterile Hybrids in Nicotiana. — A similar state of affairs has been found to exist in hybrids between various varieties of Nicotiana tahacum, the commercial tobacco, and N. sylvestris, a very different species. N. tahacum occurs in a very large number of distinct varieties some of which are so different that they could justly lay claim to recogni- tion as distinct species. Goodspeed and Clausen have studied the hybrids of a number of N. tahacum varieties with N. sylvestris and have found that in all cases the Fi hybrid duplicates very closely the total set of characters of the particular tahacum variety used in the hybrid save on a very much enlarged scale, for these hybrids are conspicuous for the increased vigor due to hybrid stimulation. In Fig. 102 is illustrated a typical plant of N. sylvestris. N. sylvestris is a monotypic species and has been grown under cultivation for over thirty years without producing any distinct varieties. It is a strikingly beautiful plant with its stout, erect growth; stiff, broad ascending leaves; and its star cluster of long pure white flowers. Nothing even approximating its flower characters occurs in the numerous varieties of N. tahacum,, in fact it belongs to a totally distinct section of the genus Nicotiana. In spite of its distinct SPECIES HYBRIDIZATION 239 characters, however, it crosses freely with members of the tabacum group, and yields reciprocal hybrids which are equivalent throughout. In Fig. 103 on the right is illustrated a plant of N. tabacum angustifolia and beside it the Fi hybrid with A^. sylvestris. The figure shows clearly how faithfully the characters of the A'', tabacum parent are reproduced in the hybrid. The leaves are long, narrow and petioled, the upper ones strap-like and pendant, the flowers are narrow and have narrow, sharply Fig. 102. — Typical plant, of Nicotiana sylvestris. pointed lobes — these and the general habit of growth are all characters clearly referable to the A^. tabacum parent. A very different variety of tabacum, such as the variety known in the University of California Bot- anical Garden cultures as A^. tabacum "Cuba" gives corresponding results. This variety is tall and bears white flowers many of which are quadrimer- ous instead of pentamerous as is normally the case in Nicotiana, These characters are faithfully reproduced in the hybrid with sijlvestris as is shown in Fig. 104. A^. tabacum ''Cuba" is peculiar among the tabacum varieties in its ability to develop seed capsules in the absence of fertiliza- tion, and these may sometimes contain a few viable seeds. This is appa- 240 GENETICS IN RELATION TO AGRICULTURE rently a recessive character in crosses with A^. tabacum varieties which display a normal behavior in this respect, but it is manifested in the i^i hybrids with N. sylvestris in the remarkable way in which this hybrid retains its seed capsules, although there are very few or no seeds in them. Since all the other Fi hybrids of tabacum varieties and sylvestris shed their flowers, often before the corolla has withered, this feature has very conspicuously characterized the Fi hybrids of A^". tabacum "Cuba" and sylvestris. * f ' ^l-') Fig. 103. — Nicotiatia sylvestris (left), A'', tabacum angustifolia (right) and the Fi hybrid (center). (After Goodspeed and Clausen.) The significant feature of these hybrids, however, is the hereditary behavior which they display. They are almost completely sterile, but if the plants are grown under reduced conditions of culture and the flowers are hand pollinated with pollen from either of the parent species, a few seeds are set, but not more than about 1 per cent, of the number ordi- narily produced by the parents. If N. sylvestris pollen is used to polli- nate the Fi, the sesqui-hybrids thus obtained are of diverse types, most of them abnormal, but about 10 per cent, closely approximate N. SPECIES HYBRIDIZATION 241 sylvestris in all their characters. These latter plants are fertile and in succeeding generations give offspring which to all indications are pure sylvestris individuals. Similarly when pollen of the N. tabacum parent is used, the sesqui-hybrids are of a variety of forms, but all approximate Fig 104. — Nicoliatia tabacum "Cuba" (left) and its Fi hybrid with A'', syhestris (right). (After Goodspeed and Clausen.) the N. tabacum parent in their characters and no one could determine by studying them that they were only once removed from A'', sylvestris. Those which most closely approximated the N. tabacum parent in morphological characters are also most fertile and give rise to fertile races which do not differ significantly from the A^. tabacum parent. 16 242 GENETICS IN RELATION TO AGRICULTURE The behavior is truly remarkable when viewed in the light of modern Mendelian conceptions. The number of character differences between the two forms is very considerable, and the recovery of the parental forms with almost unimpared fertility is so frequent that subsidiary assumptions must be made to account for them on a Mendelian basis. Goodspeed and Clausen, therefore, have developed the conception of Mendelian reaction systems for an explanation of these phenomena. According to this conception the normal functioning of a gametic or zy- gotic set of factors depends upon the harmonious interrelations which the factors maintain with one another. The uniform resemblance of the Fi hybrids of A^. tabacum varieties with N. sylvestris to the N. tahacum varieties is held in these cases to indicate that the N. tabacum set of factors is dominant as a Mendelian reaction system to the set of factors contributed by N. sylvestris. The fact that these hybrids so completely resemble the N. tahacum parent indicates that the elements of the N. sylvestris system are throughout unable to interact normally with those in the opposed N. tahacum system. It is for this reason that a reces- sive factor which is practically completely swamped in Fi intervariety crosses in N. tabacum, expresses itself so strongly in the Fi hybrids with N. sylvestris for, if the corresponding element of the N. sylvestris system were unable to interact with the elements of the dominant reaction system, then it is clear that although the factor is dominant, the corre- sponding character cannot possibly express itself in the individual. The haploid number of chromosomes in these Nicotiana species and varieties is probably twenty-four. Consequently the recombination series is given by the expansion of the expression (1 + l)^*. Only one gamete in 16,777,316 would carry only N. tahacum chromosomes and the same proportion would hold for gametes carrying only N. sylvestris chromosomes. This is on the assumption that no crossing-over occurs in the formation of gametes in the Fi hybrid. If crossing-over should occur normally the proportion of pure N. sylvestris or N. tabacum gametes would then be correspondingly reduced. The further assumption is also tacitly made that there are some factor differences between N. tabacum and N. sylvestris in every chromosome, which is in all probability correct when we consider the striking differences between the two species. Accordingly the results of the back-cross with N. sylvestris which gives a relatively high percentage of what are apparently pure N. sylvestris plants are exceedingly significant. Developing the reaction system hypothesis, it would appear that, if the N. tabacum and N. sylvestris systems display a high degree of mutual incompatibility, any gamete containing elements derived from both systems would give a reaction system subject to profound disturbances incident upon the inharmonious relations set up between the N. tabacum and N. sylvestris elements. If SPECIES HYBRIDIZATION 243 the admixture be relatively slight, the inharmonious elements may not greatly affect the workings of the reaction system, and there would result individuals showing practically the entire set of characters of one or the other parent, and such individuals would be fully fertile. A slightly greater proportion of inharmonious elements in the reaction system would result in such profound disturbances in its functioning as to produce the abnormal individuals of various kinds which make up so large a propor- tion of the progeny from such parentage. When the proportions of inharmonious elements in the gametes becomes still greater, they fail to function at all. It is upon the formation of such non-functional gametes or the attempt to produce them, that the partial sterility of the hybrid depends, and since in this particular case these form by far the greater proportion of gametes, the hybrid is very nearly completely sterile. The relations may be illustrated by Table XXXVIII which represents Table XXXVIII. -Recombination Series in Gametes of Fi of A^. tahacum X A^. sylvestris Condition of gametes Tabacum: sylvestris chromosomes Proportionate number of gametes Piogeny when pollinated with A'^. tahacum Progeny when pollinated with A', sylvestris 24:0 1 Plants resembling Plants resembling 23:1 24 the N. tahacum the Fi and ab- Functional 22:2 276 parent and of vari- normal plants but 21:3 2,024 ous degrees of all nearly com- 20:4 10,626 fertility pletely sterile 19:5 42,504 18:6 134,596 , 17:7 346,504 16:8 2.36,321 15:9 1,307,-504 14:10 1,961,256 13:11 2,496,144 Non-functional . 12:12 11:13 10:14 9:15 8:16 7:17 6:18 2,705,456 2,496,144 1,961,256 1,307,504 736,321 346,504 134,596 No viable seeds No viable seeds 5:19 42,504 Plants resembling Abnormal, infertile 4:20 10,626 the Fi hybrid and plants and fertile 3:21 2,024 nearly completely plants closely re- Functional 2:22 276 sterile sembling N. syl- 1:23 24 vestris 0:24 1 244 GENETICS IN RELATION TO AGRICULTURE the recombination series obtained in the Fi hybrid on the assumption that the chromosome mechanism is operating normally and there is no crossing- over. Neither of these assumptions is correct, but the table will show the principles involved in the production of the progeny by back-crossing. If it be assumed that the presence of not more than five N . sylvestris chromosomes in a system containing mostly N. tabacum chromosomes or correspondingly not more than five A^. tabacum chromosomes in a system containing mostly N. sylvestris chromosomes will not com- pletely disturb the relations within the systems to the point of failure to function at all, then about 0.7 per cent, of the gametes will be functional and 99.3 per cent, non-functional. This accounts for the high degree of sterility displayed by Fi. Pollinated by A'', tabacum those gametes at the A^. tabacum end of the series produce some plants which closely resem- ble the N. tabacum parent and are fertile, and others less fertile and resem- bling the A^. tabacum parent somewhat less. Conceivably some of these give abnormal forms such as have been observed in the cultures. The A'', sylvestris end of the recombination series pollinated with N. tabacum gives sterile hybrids approximating the Fi in their characters and some of these might likewise be abnormal. On the other hand when the A''. tabacum end of the series is fertilized by A^. sylvestris, sterile individuals result which resemble the Fi and perhaps where there is any missing link in the chain of tabacum chromosomes, the resulting individuals are ab- normal. The A^. sylvestris end of the series, however, gives fertile in- dividuals closely resembling A'^. sylvestris and perhaps abnormal indi- viduals which have a tendency to resemble A^. sylvestris. The high pro- portion of fertile individuals resembling the parents in either case depends on the selective elimination of the greater proportion of the gametes which contain elements derived from both parents. The conception then that recombination gametes must form harmonious reaction sys- tems in order to function accounts in these nearly sterile hybrids for the high degree of sterility, for the quick recovery of either parent by back-crossing, and for the recovery of full fertility in subsequent genera- tions upon return to the parental type. It is a curious consequence of these phenomena that it is easier to recover the exact parental types from hybrids of A", sylvestris and A'', tabacum than from intervarietal hybrids of A'', tabacum, which are fully fertile and display all manner of recombinations. Species Hybridization in (Enothera. — Curious results have been obtained in (Enothera in which genus considerable attention has been given to the results of hybridization of a large number of different species. Since these results have often been cited as evidence of non-Mendelian behavior, it is well to consider some of them in detail. De Vries par- ticularly has made a thorough study of almost every conceivable com- SPECIES HYBRIDIZATION 245 bination of species within the genus, and also of the mutants of 0. lamarckiana with the parent species, with one another, and with other species. As an example of the type of behavior displayed by these hybrids we shall take the results of intercrossing lamarckiana and its two mutant derivatives ruhrinervis and nanella. When lamarckiana is crossed with rubrinervis, the phenomena are as outlined below: lamarckiana X ruhrinervis lamarckiana subrobusta lamarckiana subrobusta rubrinervis The Fi consists of two forms in about equal proportions, lamarckiana and subrobusta, the latter a form intermediate between lamarckiana and rubrinervis. In subsequent generations, the lamarckiana individuals breed true, but the subrobusta individuals produce both subrobusta and rubrinervis, the latter breeding true. To these results we may add those obtained by crossing lamarckiana and nanella, the dwarf mutant of latnarckiana. This cross gives in Fi approximately equal numbers of lainarckiana and naviella and both forms breed true in subsequent generations. Finally to complete the triangle we may consider the results of hybridization of rubrinervis and nanella which are given below in the form of a diagram. nanella X rubrinervis lamarckiana subrobusta lamarckiana rubrinervis subrobusta dwarfs The percentage of subrobusta individuals in the Fi of this cross is usually considerably below 50 per cent. In subsequent generations the subro- busta individuals segregate in the same fashion as those of the Fi. The dwarfs obtained in this experiment unite the characters of r-ubrinervis and nanella and are consequently designated rubrinervis nanella to dis- tinguish them form the true nanella. Like the lamarckiana and rubri- nervis individuals, they breed true in subsequent generations. The actual results of this series of experiments are given in Table XXXIX, from which data on those forms which bred true is omitted. It is at once ap- 246 GENETICS IN RELATION TO AGRICULTURE Table XXXIX. — Results of Various Matings of rubrineruis (R.) and nanella (N.) AND THE Forms Produced prom Such Matings (compiled from de Vries, '^Gruppenweise Artbildung") Parentage Number of plants Lamarckiana Rubrinervis Subrobusta »r , Rubri- ^f"" nervis *"" nanella Nanella X rubrinervis Nanella X rubrinervis Rubrinervis X nanella (N. X R.) subrobusta (R. X N.) subrobusta {R. X N.) subrobusta (A'^. X R.) subrobusta {R. X N.) subrobusta Lamarckiana X R. nanella R. nanella X lamarckiana (A''. X R.) lamarckiana X nanella {N. X R.) lamarckiana X nanella (R. X N.) lamarckiana X nanella Nanella X (R. X A'^.) lamarckiana Nanella X (N. X R.) lamarckiana {R. X A^.) lamarckiana X R. nanella Nanella X (A'". X R.) subrobusta. . R. nanella X {R. X A''.) subrobusta (N. X R.) subrobusta X nanella... (N. X R.) subrobusta X R. nanella (N. X R.) subrobusta X R. nanella {R. X A''.) subrobusta X R. nanella 105 79 70 160 160 56 230 234 152 152 266 70 112 68 27 84 45 204 138 246 214 289 Per cent. 73 59 59 3 25 86 80 76 62 55 33 51 Per cent. 10 3 34 21 15 87 16 33 20 75 72 72 Per cent. 27 41 41 80 85 52 70 73 77 32 Per cent. 10 12 14 9 12 20 43 14 20 24 38 45 51 67 29 25 28 28 parent that the phenomena exhibited, although complex, are very orderly; but no very consistent Mendelian interpretation has been advanced to account for all of them. The hypothesis of de Vries while ingenious does violence to many of our most cherished conceptions of the general nature of hereditary phenomena. De Vries assumes that pangens exist in three forms; active, labile, and inactive. Two pangens are concerned in the above series of forms, the rubrinervis pangen for strengthening of the vascular bundles and the nanella pangen for stature. These pangens exist in lamarckiana in the labile condition in which they occasionally change to the inactive condition and thus produce the corresponding muta- tions rubrinervis and nanella. Labile pangen X inactive pangen then gives according to de Vries in Fi the ascendency of either one or the other condition to the complete exclusion of the other form in later generations. Accordingly lamarckiana X nanella gives in i^i lamarck- iana and nanella which breed true in further generations. Similarly SPECIES HYBRIDIZATION 247 when lamarckiana is crossed with ruhrinervis, the rubrinervis pangen in lamarckiana is in the labile condition, but in ruhrinervis it is in the in- active condition. Here, however, a difficulty is introduced by the fact that the form corresponding to rubrinervis in Fy is intermediate between rubrinervis and lamarckiana, it is the form which de Vries calls sub- robusfa. Must we assume a fourth condition for the pangens in this form? An additional difficulty is introduced when we consider crosses of rubrinervis and nanella. Rubrinervis has arisen from lamarckiana by mutation, by a change of the labile rubrinervis pangen in lamarckiana into the inactive condition. But when rubrinervis is crossed with nanella, Fx consists entirely of lamarckiana and subrobusta plants. As we pointed out, crosses of nanella with lamarckiana show that the nanella pangen NN rubrinervis N'N' Nn lamarckiana subrobusta I N'N' NN Nn nn lamarckiana rubrinervis subrobusta nanella Fig. 105. — Results of crossing two "mutants" of CEnothera lamarckiana. in lamarckiana is in the labile condition. How, then, should this pangen have become inactive in rubrinervis which was supposedly de- rived from lamarckiana by a change in the rubrinervis pangen? For according to de Vries the behavior of the nanella pangen in such an experiment is illustrated in Fig. 105 in which the active pangen is designated by N, the labile pangen by A'"', and the inactive pangen by n. Those who have attempted to apply a rigid Mendelian analysis to the ffinothera phenomena have failed to do so without making assumptions which thus far remain beyond the limits of experimental verification. Nevertheless the work of such investigators as Heribert-Nilsson, Renner, Davis, and others demonstrates that Mendelian analyses may be applied to particular cases and that when the difficulties which occur in GEnothera are considered, the facts thus far discovered do not preclude an ex- planation on an essentially Mendelian basis. Davis in particular has pointed out that thus far no species of ffinothera has been found which will stand trial as of strict genetic purity. In all species apparently 50 per cent, or more of the pollen grains are abortive and similar 248 GENETICS IN RELATION TO AGRICULTURE proportions of the ovules are non-functional. To this category of facts must be added the high percentage of seed sterility which is common in the genus. If any of this pollen, ovule, and seed sterihty is selective, then obviously it will be impossible to analyze the progeny successfully, unless the exact nature of the non-functional gametes and zygotes may be determined. The importance of this point has been indicated in the explanation of the frequent occurrence of parental forms among the sesqui-hybrids of rye and wheat and of Nicotiana tabacum with A^. sylvestris, and it has been definitely established for many cases of albinism in plants and for peculiar sex ratios and consequent disturbances of Mendehan ratios in Drosophila. Until, therefore, a satisfactory account can be given of the difficulties which have been enumerated above it will be impossible on the one hand to offer a satisfactory Mendelian interpretation of the ffinothera investigations and illogical on the other hand to advance the results of these investigations as evidence of non- Mendelian inheritance. Moreover, considerable success has attended the efforts to produce by species hybridization strains of ffinothera which behave like lamarck- iana. It is not without significance that Davis has been able to pro- duce forms by crossing 0. biennis and 0. franciscana so much like lamarckiana as to be indistinguishable from it taxonomically. Tower also has taken pure species of Leptinotarsa, the Colorado potato beetle, and by mating them has produced strains which breed approximately true, but which under the stress of unusual conditions may throw off small percentages of aberrant forms. In his species crosses in Anti- rrhinum, Lotsy has reported the occurrence of races which give small proportions of aberrant forms. Since at present we have no certain knowledge that lamarckiana is not a form of hybrid origin and that its so-called mutants are not really segregants from a race possessing a peculiar hybrid constitution, these analogous cases assume considerable importance as an indication of the hne of attack which may be followed for an explanation of the Oenothera phenomena. Conclusions.— If we attempt to outline the present status of our knowledge of the phenomena of species hybridization, we see thus far no clear evidence of non-conformance to an explanation which is essen- tially Mendehan. The strict Mendelian explanation must be modified to take into account the peculiar relations which obtain in species hy- bridization. For an explanation of such relations the reaction system conception has been advanced. According to this conception the total set of factors in any species forms a reaction system in which the factors display harmonious interrelations with one another. Variety hybridi- zation, since it is concerned only with isolated differences in systems which are fundamentally identical, usually produces no disturbances in SPECIES HYBRIDIZATION 249 the reaction system relations. Consequently strict Mendelian analyses may be applied to such phenomena, and the reaction system relations need not be considered. But when species are crossed we must look to reaction system relations to account for the fact that not every set of factors which can be obtained by recombination is capable of establish- ing the harmonious interrelations which are necessary for normal func- tioning in a reaction system. As a consequence species hybrids exhibit a peculiar set of phenomena including sterility, whether partial or com- plete, production of abnormal forms, and apparent lack of conformance to established principles of hybridization. Underlying all these surface phenomena, however, is a behavior essentially Mendelian, if we take Mendelism to include all those phenomena consequent upon the shuffling and recombination of factors which possess at least a relatively high degree of stability. Since any irregularities in the distribution of factors or chromosomes, which may be occasioned by the inharmonious re- lations within the hybrid reaction systems acting upon the chromosome mechanism, can hardly be considered to give rise to results which should not be included under the term Mendelism, it is very evident that simple assumptions such as we have outlined above will account for a con- siderable array of phenomena. CHAPTER XIII PURE LINES For half a century succeeding Darwin, it was assumed that by selecting a certain type of individual for propagation, the species or variety would be continually transformed in the direction of the selec- tion. Such a conception was a natural result of the widespread acceptance of Darwin's theory of the method of evolution and later of Galton's ''law of inheritance" as applied to selection. Experience seemed to bear out this idea also, inasmuch as continual selection of the best plants for seed and the best animals for mating was found to be profitable. But it was not until Johannsen decided to test the power of selection by keeping the pedigrees of individual plants and their descend- ants that the truth concerning the composition of varieties of cultivated plants became known. Heterogeneity within single botanical species had already been discovered, but that horticultural varieties were also heterogeneous but with respect to less easily distinguishable characters had not been realized. Definite knowledge concerning the composition of horticultural varieties threw light on the problem of selection by ex- plaining why continuous selection within a variety is necessary in some crops while it has little or no effect in the case of certain other crops. This discovery was of tremendous significance to genetics, particularly to breeding. For this reason the following account of Johannsen's classical experiments is based directly upon his own presentation of the matter. Discovery of Pure Lines. — Johannsen chose a certain brown variety of the common garden bean {Phaseolus vulgaris nana) known as the Princess bean. In 1901 he harvested 287 plants which had grown from selected seeds of very different sizes and of known weights. The har- vested beans from each plant were weighed separately. They were then divided into classes with an interval of 10 eg., the class center values ranging from 30 to 80 eg. Next he determined the mean weights of all the beans from the plants grown from mother beans falling in the first class (25-35 eg.) and similarly for the progeny of each of the groups of the mother beans. The result is shown in the following table. Weight of mother beans 30 40 50 60 70 80 Mean weight of progeny 37.1 38.8 40,0 43.4 44.6 45.7 250 PURE LINES 251 These two series may be expressed in terms of percentage by multiplying each scries by a factor that will change the value of the middle class to 100. The mean weight of all the mother beans was very nearly 50 eg. while that of the progeny is approximately 40 eg. Thus the first series is multiplied by 2 and the second by 2.5 giving the following result. Weight of mother beans 6C 1 Sf ) 100 120 140 160 1 ""' Mean weight of progeny 93 97 100 108 111 114 Now the deviation of each progeny class can be compared directly with the deviation of the mother class. Deviation of mother beans -40 -20 ! 20 40 60 Deviation of mean weights of progeny . . — 7 -3 8 11 14 Thus the ratios of the minus deviations of the progeny classes to the minus deviations of the mother classes are %o ^^^ %q, the mean of which is i%o or 0.163. Similarly for the plus deviations, %o, ^}io> ^%0 X H, 0.303. The average of these two values is 0.233 which is about 3^ as compared with Gallon's observation of ^3 inheritance in size of seed in the sweet pea and stature in man. During these preliminary experiments, however, Johannsen noticed that plants grown from similar sized beans produced beans of very differ- ent sizes. Thus, for example, the plants grown from the largest mother beans (about 80 eg. in weight) yielded seeds of strikingly different sizes. The average weight of the seeds of these individual plants varied between 35 and 60 eg. and when the weights of all the individual beans of this series were arranged in a frequency distribution it produced a series that differed considerably from the normal frequency distribution. The distribution of 598 seeds, all progeny of beans about 80 eg. in weight, when arranged in classes of 5-cg. intervals, was as follows : Classes 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 Number of seeds 5 18 46 144 127 70 70 63 28 15 8 4 Theoretical numbers 1 3 11 26 53 85 109 112 91 59 30 13 4 1 M = 45.44 ± 0.43 eg.; a = 10.40 eg. Clearly this distribution if plotted would produce a skew polygon with the mode to the left of the theoretical mode. This observation caused Johannsen to have serious doubt regarding the biological justification of Gallon's law. For such a distribution did not appear to be the expression of only one "type"; on the contrary, it seemed more likely that the material was mixed. 252 GENETICS IN RELATION TO AGRICULTURE This state of affairs was the starting point of further critical study. In order to take account of the effect of selection supposedly in the opposite direction, he next examined the progeny of the smallest mother beans (about 30 eg.) and found that they displayed no such striking irregularity as did the progeny of the largest beans. (Possibly this was due to the fact that about 20 plants were grown from the smallest beans while the progeny of the largest beans came from only 11 plants.) The progeny seeds from the smallest mother beans were weighed individually and the data put in the form of a frequency table as in the former case. Classes 15 20 25 30 35 40 45 50 55 60 65 Number of seeds 8 18 71 156 172 127 35 15 3 6 Theoretical numbers 1 6 27 77 139 162 121 57 17 3 1 M = 36.68 ± 0.30 eg.; a = 7.33 eg. This distribution does not indicate a mixture. Instead it suggests that a single original "weight type" of bean was set apart by selection in the minus direction. The general result of this preliminary study was cer- tainly a sort of confirmation of Galtonian regression ; but at the same time the doubt was aroused whether the original population was not a hetero- geneous mixture from which selection simply sorted out already existing "types". Hence came the question: Will selection of plus or minus variants within pure lines bring about the isolation of types and cause Galtonian regression? This question was answered the following year (1902). A series of 19 pure lines was used for this investigation. Each of these pure lines originated from a single bean from the crop of 1900. In the fall of 1901 each line was represented by the seeds of one plant. In 1902 he planted 524 seeds. Every seed was given a number and each plant was harvested separately. Each pure line, each plant and every single bean was sepa- rately numbered. Thus each individual could be compared with every other individual. Johannsen first compared his material as a whole with the results of his preliminary study. Having recorded the weight of each bean, he arranged the data in groups corresponding to the classes of mother beans as in the previous year. Weight of mother beans 20 30 40 50 60 70 Mean weight of progeny 44.0 44.3 46.1 49.0 51.9 56.1 Number of progeny seeds ISO 835 2,238 1,138 609 494 Again he found about 34 inheritance and ^ regression of progeny on mother beans. He next divided each of these six groups of progeny beans into classes according to weight as shown in Table XL. PURE LINES 253 Table XL. — Showing Variation within Classes in a Population Composed OF Pure Lines. (From Johannsen) Classes of progeny seeds in eg. n M beans, eg. 5 15 25 35 45 55 65 75 85 95 15-25 1 15 90 63 11 180 43.78 + 0.56 7.47 25-35 15 95 322 310 91 2 835 44.47 + 0.31 9.03 35-45 5 17 175 776 956 282 24 3 2,238 46.17 + 0.19 8.93 45-55 4 57 305 521 196 51 4 1,138 48.94 + 0.28 9.34 55-65 1 23 130 230 168 46 11 609 51.87 + 0.42 10.42 65-75 5 38 5 53 175 180 64 15 2 2 — 494 56.03 + 0.45 10.02 Total 15-75 eg.. 370 1,676 2,255 928 187 33 5,494 47.92±0.13 • 9.87 It is true that each of the six progeny series corresponds closely to the normal frequency distribution. There is no distortion such as would be expected from mixed material. Nevertheless it becomes evident that the material is heterogeneous as soon as the data are arranged by pure lines as shown in Table XLI. Table XLI. — Survey of the Effect of Selection in Pure Lines (The dark-faced figures indicate mean weights in eg.; the light-face figures designate respective numbers of seeds.) {From Johannsen) Weight in eg. of the mother beans Mean The pure weig hts he lines of t 20 30 40 5C 60 70 lines I 63.1 . . .. 6 4.9 91 64.2 145 II 57.2 86 54.9 195 56.5 120 i 5.5 74 56.8 475 III 56.4 144 56.6 40 e 4.4 98 55.4 282 IV • 54.2 32 53.6 163 i 6.6 ] 12 64.8 307 V 52.8 107 49.2 29 t 0.2 ] 19 61.2 255 VI 53.5 20 50.8 111 42.6 10 60.6 141 VII 45 9 16 49.5 262 48.2 27 49.2 305 VIII 49.0 20 49.1 119 47.5 20 48.9 159 IX 48. 5 117 47.9 124 48.2 211 X 42.1 28 46.7 412 46.9 93 46.6 533 XI 45.2 114 45.4 217 46.2 87 45.5 418 XII 49 6 14 45.1 42 44.0 27 46.5 83 XIII 47.5 93 45.0 219 45.1 205 45.8 95 45.4 712 XIV 45.4 21 46.9 51 42.8 34 46.3 106 XV 46 9 18 44.6 131 45.0 39 46.0 188 XVI 45.9 147 44.1 90 41.0 36 44.6 273 XVII 44 78 42.4 217 42.8 295 XVIII 41 54 40.7 203 40.8 100 40.8 357 XIX 35.8 72 34.8 147 35.1 219 I-XIX 44.0 180 44.3 835 46.1 2238 49.0 1138 51.9 609 < S6.1 ' 194 47.9 5494 254 GENETICS IN RELATION TO AGRICULTURE The above analysis not only demonstrates that Johannsen's material was a mixture of different "weight types" but it also gives striking proof that selection within a single pure line has no effect. Johannsen points out that in certain lines (/, X, XI) there seems to be a slight effect but that in others (VI, IX, XII, etc.,) an opposite tendency appears; while still others (//, ///, 7/77). are irregular. Generally speaking then no effect of selection is seen for there is no significant difference between the means of the several groups in each pure line. The apparent indications of selection effects are merely fortuitous variations. In each of these lines, therefore, the offspring of plus and minus variants exhibit complete regression to the mean of the particular line. In short, individual varia- tions were not inherited, only the characteristic modifiability of the particu- lar line was inherited. Johannsen did not rest here but continued to test his pure lines of beans during successive years. He found a certain amount of seasonal fluctuation in the range of variation and in the variation constants, yet each pure line maintained its own individuality as indicated by the varia- tion in weight of beans produced. And this maintenance of entity was accomplished in spite of repeated selections of smallest and largest beans so that each year every pure line was represented by two lots of plants, a "plus strain" grown from the largest beans and a "minus strain" grown from the smallest beans. Complete failure of such repeated selection to cause significant change in the mean weight of either strain was observed in each pure line. As illustrations the data on Lines I and XIX are presented in Tables XLII and XLIII. From these data it is evident that six years of selection of plus and minus strains within Line I produced no permanent departure in either direction. In fact the last column (B-A) actually shows an inverse effect during three of the six years. Moreover, if the average of the means for the six years in both strains be compared this conclusion is verified. Table XLII. — Selection-effect Duking Six Generations in Line I op the Princess Beans. {From Johannsen) Harvest years Total number of beans Mean weight of mother beans of the select strains Differ- ence b - a Mean weight of progeny seeds of select strains Difference B - A a-minus 6-plus A-minus B-plus 1902 145 60 70 10 63.15±1.02 64.85 + 0.76 + 1.70 + 1.27 1903 252 55 80 25 75.19 + 1.01 70.88±0.89 -4.31 + 1.35 1904 711 50 87 37 54.59 + 0.44 56.68±0.36 +2.09 + 0.57 1905 654 43 73 40 63.55 + 0.56 63.64 + 0.41 +0.09±0.69 1906 384 46 84 38 74.38 + 0.81 73.00 + 0.72 -1.38 + 1.08 1907 379 56 81 25 69.07 + 0.79 67.66 + 0.75 -1.41 + 1.09 PURE LINES 255 Table XLIII. -Selection-effect During Six Generations of Line XIX of the Princess Beans. {From Johannsen) Harvest Total number of beans Mean weight of mother beans of the select strains Differ- ence b - a Mean weight of progeny seeds of select strains Difference B - A a-minus 6-plus A-minus B-plu8 1902 1903 1904 1905 1906 1907 219 200 590 1,657 1,367 594 30 25 31 27 30 24 40 42 43 39 46 47 10 17 12 12 16 23 35.83 + 0.44 40.21 + 0.65 31.39 + 0.29 38.26 + 0.16 37.92 + 0.22 37.36 + 0.30 34.78 + 0.38 41.02 + 0.43 32.64±0.21 39.15 + 0.17 39.87 + 0.16 36.95 + 0.21 -1.05 + 0.58 +0.81 + 0.78 + 1.25±0.36 +0.89±0.23 + 1.95±0.27 -0.41+0.37 The mean for the progeny of the plus strain is 66.12 + 0.28 and for the progeny of the minus strain, 66.66 + 0.33. The difference is -0.54 + 0.43 (the probable error of the difference in all cases being found by taking the square root of the sum of the squares of the two probable errors). In Line I, therefore, there is no positive effect of selection; on the con- trary there would appear to be a slight inverse effect! Line XIX was characterized by beans of the least weight. The data for the results of six years of selection in plus and minus directions, particu- larly the difference between the progeny means (B-A), reveal somewhat larger fluctuations in the plus direction than in Line I but it will be noted that the probable errors of the differences are smaller, hence the validity is the more certain. Comparing the means of the means of the progeny seeds as before, for the plus strain we have 37.40 ±0.11 and for the minus strain, 38.83 + 0.15, the difference being +0.57+0.19, which is certainly small although in the plus direction. Now, if we compare the summaries of the data from these experiments, -0.54 and +0.57, we are forced to conclude that selection was without effect in these pure lines. Finally Johannsen conducted similar experiments with the Princess beans, using the characters, length and breadth. He came to the same general conclusion, to wit, that he found no trace whatever of selection effect within pure Hues and that the variations in pure line individuals are merely fortuitous modifications and are not inherited. Conditions Necessary for the Existence of Pure Lines.— Johannsen defined a pure line as the progeny of a single self-fertilized individual of homogeneous factorial composition. Unless mutation takes place none of the descendants of such an individual can differ from the parent in their genetic factors. Two important conditions are imposed by this definition, viz., homozygosity and self-fertilization. The latter of these is the more fundamental inasmuch as it is mathematically demonstrable that self-fertilization, if continued generation after generation, leads 256 GENETICS IN RELATION TO AGRICULTURE rapidly toward a homozygous condition in all descendants. Thus, Jennings shows that in the case of the original cross, A A by aa giving all Aa, if thereafter all breeding is by self-fertilization, then, after n genera- tions, the proportions of different genotypes in the population may be calculated by the following formulae: AA = Aa = aa = 2n+ 1 2"' 2" - 1 2n+ 1 • Therefore, within six self-fertilized generations after a cross involving a single pair of factors, the proportion of homozygous individuals in the population for one or the other of the two factors will be 98.4 per cent. Hence it is clear that, even though many genetic factors are concerned, as is undoubtedly the case in any crop plant or domestic animal, yet in those species where self-fertilization is the method of reproduction, the fundamental condition necessary to the existence of pure lines is met. Although by definition every pure line is a genotype, yet every genotype is not a pure line, for any heterozygote belongs to some genotype whereas a pure line is necessarily homozygous. Upon the basis of Johannsen's definition, it would be impossible to obtain pure lines from obligatory allogamous species, to which class belong all domestic animals and certain cultivated plants. However, it is clear that continual inbreed- ing in such organisms would tend to produce a homozygous genetic composition. Isolation of Pure Lines from Mixed Populations. — In order to obtain pure lines from mixed populations the method employed will de- pend upon the method of reproduction of the organism. In autogamous species the method adopted by Johannsen in working with beans is adequate. The individual plant being capable of reproducing the species through self-fertilization and incapable of natural cross-fertilization, it is only necessary to isolate the progeny of single individuals to establish pure lines. However, in supposedly autogamous species natural hybrids sometimes occur. Hence in critical work it is always advisable to pro- tect the flowers even of autogamous plants. In dealing with allogamous species, in which it is necessary to mate two individuals, when starting with a mixed population of unknown genetic factors the original selections must be made on the basis of phenotypic similarity. With domestic animals the repetition of such selection for a large number of generations has produced the "pure" or pedigreed breeds, which approximate more or less closely to pure lines and hence should be expected to breed fairly PURE LINES 257 true to type. With plants the method of procechire depends upon the details of reproduction in the species under consideration. For example, corn is naturally cross-fertilized but is also self-fertile, while the common sunflower is self-sterile and so must always be cross-fertilized. With such plants as the sunflower, then, the procedure will be as with animals and the length of time required to produce approximately pure lines will depend upon three things: (1) the number of genetic factors for which each of the selected individuals is heterozygous; (2) the number of genetic factors with respect to which the two selected individuals differ; (3) the number of chromosomes in the species. The specific chromsome number is an important consideration because of its direct relation to the number of linked character groups or in other words to the possible number of freely assorting pairs of factors. Sufficient has been said concerning the comparative ease of isolating pure lines from populations of autogamous species and the relative difficulty of obtaining pure lines from allogamous species to make it clear that the material under consideration is of the highest importance in all critical discussions of the effect of selection within pure lines. Finally, it is to be noted that a vegetatively pro- pagated phenotype may or may not be a pure line according to its genetic constitution. A group of individuals thus propagated is known as a clone. In strictly allogamous species a clone would hardly ever be homozygous. The Effect of Selection Within Pure Lines. — There is now con- siderable evidence in support of the theory that selection within a pure line is without effect. This evidence comes from the results of practical breeding as well as scientific investigations of certain autogamous species of plants, such as wheat, oats and barley; also from thoroughgoing re- search on a few allogamous species, especially on certain insects and pro- tozoa, particularly paramecia. The constant maintenance of head type in wheat is strikingly portrayed in Fig. 106, which shows two heads from each of four varieties which were first isolated by Louis de Vilmorin between 1836 and 1856. The plants according to Vilmorin were found to be identical in all respects "although separated by an interval of 50 years during which annual selection had been continued. This fixity is shown not only in the characters of the ear but also in all the other characters of the plant even that of precocity, which would appear to be most dependent on climate." The use of this case as evidence in support of the pure-line theory has been criticised upon the ground that the selec- tion practised had for its purpose the preservation rather than the altera- tion of the type. But from the experience of many investigators and breeders we may safely conclude that within true pure lines selection is without effect on the type unless mutations occur. After subjecting a variety of barley known as Glorup to plus and minus selection for eight 17 •258 OEXETICS I\ UKl.ATTOX TO ACincn.Tr Ix'K s2;oniM;Uions, \\\c charartcv uiuUm' observatiiMi biMui:; dogro(> of mealiness of the kernel i^Sehartigkeit), .lohannsen eonehuled ihat (he seleelion had prodneeil no etVeel. ^Moreover (he Swedish plan(-breeiling s(a(ion at Svalof has been guided for years by (he knowledge (hat (heir pedigree eul lures, i.e., pure lines, were not ehangeil by seleetion. A sinnlar eon- olusion was reaehed by Tower after four to ten generations of rigorous selection of albinie indi\iduals in three dilVerent attenijits to establish an albinie raee from a stable raee (^pedigree material) of the C\ilorado potato beetle {lA'ptinofa)\^a (hccniJiiicata). The history of these three Fig. 100. — Four pure linos of wlioat which have boon grown by Vilniorin for 50 years. Tho orisxinal spocimon iu the seed museiun is shown on the left iu each case. The close siniihuity of the pairs of heads indicates that pure lines remain constant indefinitely. (.\j'tcr Iltiijidooni.) experiments are shown at A, B and (' in Fig. 107. The small black polygons show for each generation the imlividuals selected to become the parents of the next generation. It will be noted that neither the range nor the mode of the population is permanently shifted in the direction of the selection. Thus we find that in races or varieties which are constant (homozygous) selection has no etl'ect unless mutations occur. Vaiious evidence has been brought forward to show that the principle does not hold for all organisms. But in all such cases among sexually propagated species we may assume that the material used was hetero- zygous for certain factors. Such has been shown already to be a satisfac- tory explanation of Castle's results in selecting for phis and minus strains in the hooded rats which is one of the cases originally advanced as evidence against the pure line theory. I'nia-: fJNf'js 250 Significance of the Pure Line Theory in Breeding. The, quoHtjon UiijH urisfs: Wow does the pun; lino (Jioory explain tlic fact that man ha.s wroiij.'lii f»ofoijnr] changes in fJonicsticated animals and plants by Holec- " Normal' Uarnje of Variation "Norrnal" KariKc of Variation Mode Mode I'k;. 107. — Diagrammatic represftntation of results of three experiments in selecting beetles in an effort to create an albinic strain from a pure strain, (from Twjjer.) tion? It is well known that as a rule a mixed population coasists of a n limber (probably quite large) of distinct biotypes and that in autoga- mous species these biotypes are pure lines to begin with, while in alloga- mous species it is only by continued inten-sive selection that existing 260 GENETICS IN RELATION TO AGRICULTURE biotypes can be differentiated from one another so that they "breed true." How these distinct biotypes originate will be considered in the following chapter, the fact that they exist is the chief consideration here. The effect of "mass" selection in causing temporary changes in heteroge- neous varieties of plants and races of animals is easily understood by the aid of the diagram shown in Fig. 108. The area within the la,rge curve represents a mixed population or phenotypically similar group containing a number of distinct genotypes indicated by the small curves A-Z. Every genotype has its own variation curve and is distinct from each of the others, but they intergrade with each other so completely that the population appears as an entity. Now if one should select individuals from either extreme of the population, say at 90 or 70, it is clear that such individuals might belong to any one of four or five geno- JOM 62 C3 64 as eOWOii 687071 727374 Fig. 108. — Schematic diagiam showing the relation of a population to the biotypes composing it, or of a phenotype, to the genotypes or pure lines within it. (After Lang from Goldschmidt.) types. If selection in the same direction were continued a strain would be established with a mode distinct from the mode of the original popu- lation. These strains could be maintained by continual selection and in time a single genotype might be isolated when selection would be said to have changed the type permanently. But selection changed nothing — it only isolated a certain genotype or genotypes from the origi- nal mixture. Tower's results in selecting for the purpose of creating albinic and melanic strains of beetles as illustrated in Fig. 109 may be explained in this way. The original population shown at A consisted of a number of distinct biotypes. By the isolation of several extreme variants Tower separated plus and minus strains which he was able to maintain for eight generations by practising intensive selection. In the eighth generation he divided each population in half, continuing in- tensive selection with one portion and stopping all selection in the other. By this method he was able to maintain the plus and minus strains and at the same time to observe that in the ninth generation the mode of the PURE LINES 201 progeny of the unselected eighth generation population lay much nearer to the mode of the original population. Within three generations the unselected strains had moved back to the mode of the species. Now it is to be remembered that Tower was dealing with an obligatory allo- gamous species. Moreover, what is now known concerning body pigmentation in Drosophila makes it altogether likely that quite a large "I " " "■" A ^1- &K / \ ~l V y^ \ 3"i ^i 3 \^ \i ^1 ]JL \ ;-:4-:V>f=^=y ■4 4--^'^ i ,1 P^ .3l II ^-|Sl ^31 X > = ^ ^ :-r- ^j ^ =>■ "-i^rv/'^ j I tf^'^V > ^, 1 ' ! II e a— -'^ V '^^^^t' ' / M lY ^s||^;^:i^^^^^ V ^^ ^ V i 31 Z 5 li_ IeL' V-;! = l vn -,--^. T^a^ TO 2===-^ ^--= t irV- ^t l| /Pt y = = VII 1 ffl^ M k A -d- t^ -^m^ Se^ L^aJ -f#% l+l fjt M ^^ /^^ y^f=*L V i^. J^^ ty=f^ 4 V 4 i,^ -it mr M^ ^^ IV ^^ ^t4^ i^ ii^^ >H=f=N 4- -^4 A ^^g ih^ ,^-.f^, TTT M^, [ym\ r ]>-Kn N I iVt 1 4 iiJ?^ Aj 4At^. ^^=-i II -/i i N^rf^-3EL, ^--i lX^---^ 41 ! .^--'N. i- A k'/^^sf A \\c i X t miNim^BrriiNiiNN Fia. 109. — Diagrammatic representation of the results obtained by the creation of albinic and melanic strains from a mixed population of beetles. {From Tower.) number of genetic factors are concerned in the degree of pigmentation of these beetles. Hence selection of phenotypes for a number of genera- tions did not isolate genotypes, i.e., the plus and minus strains were not homozygous. As this is an allogamous species undoubtedly most of the individuals in the original population were heterozygous for many factors. Furthermore, Tower did not select single pairs but always took several 262 GENETICS IN RELATION TO AGRICULTURE pairs as parents for each generation. While the selection of similarly pigmented individuals would tend gradually toward a homozygous condition,with respect to the specific factors conditioning pigmentation, yet it is altogether likely that under the conditions of the experiment a considerable degree of heterozygosity was maintained. In other words the selection practised did not isolate pure lines, the plus and minus strains did not become homozygous. Much of the work done in the past in ameliorating animals and plants has been by this method of selecting phenotypes but not genotypes, which accounts in part for the frequent necessity of continuous selection in maintaining improved strains or breeds. In reviewing the development of plant breeding we shall note certain cases of early recognition of the effects of genotypic selection, a principle which is now accepted as fundamental in all breeding operations. CHAPTER XIV MUTATIONS Baiir's third category of variations comprises all inheritable changes due to causes other than segregation and recombination of genetic factors. Although comparatively little is known concerning the specific causes of mutations, yet it is possible to distinguish between two general classes of such inheritable variations according to the nature of the genetic units involved. These classes are (1) alterations in genetic factors, and (2) deviations in the number of chromosomes. We designate the first group as factor mutations and the second as chromosome aber- rations. Since the first group is of vastly greater importance to agri- culture than the second, we shall consider the latter very briefly before engaging in discussion of the former, which we deem worthy of recognition as mutations in the strict sense. Chromosome Aberrations. — By the aid of cytology it has been demon- strated that inheritable changes are occasionally induced, in plants at least, by irregularities in the behavior of the chromosomes during mitosis or meiosis, such that certain germ cells contain fewer or more chromo- somes than the number typical of the species. Aberrant forms in several plant families are now known to differ from the parent species in chromosome number. Some have only a single chromosome more or less than the parent, while a few are known in which the original number is doubled. It is possible that aberrations occur involving all combina- tions of numbers between these two extremes. In various forms of La- marck's evening primrose {(Enothera lamarckiana) , whose typical number is 14, according to Gates the following aberrant numbers have been reported— 15, 20, 21, 22, 23, 27, 28, 29, 30. Aberrations involving the doubling of the number of chromosomes typical of the species is known as tetraploidy because there are four times the haploid number typical of the parent. Occasionally aberrations or hybridization between diploid and tetraploid forms result in triploidy. There is a limited amount of evidence which indicates that groups of species have arisen by progressive alterations in chromosome number. Thus in Drosophila, Metz has found ten species in which the chromosome numbers range from 6 to 12 and the larger numbers appear to have arisen by subdivision of the large dumbbell-shaped chromosomes found in the species having smaller numbers. Evidence 263 264 GENETICS IN RELATION TO AGRICULTURE 1 that doubling of the chromosome number may occur during somato- genesis has been found by Farmer and Digby in the interesting hybrid, Primula kewensis. The original plant, which was sterile, "had 18 and 9 chromosomes in its premeiotic and postmeiotic nuclei respectively," but in the fertile plants which were propagated asexually from it, as well as in similar fertile hybrids which were produced in later experi- ments, the diploid and haploid numbers were 36 and 18 respectively. Having found by means of careful measurements of the chromosomes in the two forms that the nuclei in both forms contain the same volume of chromatin, the authors conclude that the increase in number may be attributed to transverse fission of the 18 larger chromosomes and not to the fusion of two nuclei. From a study of chromosomal dimensions in relation to phylogeny, Meek "arrived at the conclusion that the widths of chromosomes are successively greater in higher zoological phyla, and that this dimension is constant for very large groups of animals." But Farmer and Digby have shown that such a conclusion is without foundation since "closely related forms may possess chromosomes differing widely in shape and size and character." Hence they conclude "that phylogenetic affinity is not, necessarily, correlated with chromosome width." They also point out that "unfortunately we know practically nothing about the phylogeny of the chromosomes. No convincing hypothesis has been put forward to explain how these remarkable bodies have become organized, nor how their peculiarities have either been brought into existence or are kept so true for a given species." However, we are reminded by Glaser that chromatin is present in bacteria though not in the form of a nucleus and it may not be too much to hope that cytology may yet discover the principal stages in the development of the chromo- somes and establish such correlation as may exist between this develop- ment and organic evolution. Certainly extended investigations of chromosome numbers must be made before chromosome aberrations can be considered an important factor in evolution. Except that certain chromosome aberrations, such as tetraploidy causing gigantism, might be of economic value, in general this class of mutations is of minor importance in breeding. Factor mutations, on the other hand, are of prime importance and of general occurrence. Factor mutations have appeared in controlled cultures of many animals and plants and the character differences con- ditioned by them are as a rule such as distinguish varieties of a single species. Moreover, varietal characters are Mendelizing characters in the narrow sense and the existence of simple Mendelian phenomena among all classes of sexually propagated organisms proves that factor mutations are of general occurrence. Although it is probable that every MUTATIONS 265 factor mutation has a certain effect upon every character in the organism, yet the visible effects of some factor differences are restricted to a single character. According to their visible effects, therefore, we recognize two classes of factor mutations: (1) those conditioning apparently only single characters; (2) those having a visible manifold effect on the soma. Cases Fig. 110. — A seedling of the oak-like walnut (left) and of the California black walnut, the parent species (right.) involving mutations of the second class are known in several species of animals and plants. An interesting example is the oak-like walnut, Juglans calif ornica var. quercina, which appears to differ from the parent species by a single factor difference, Fig. 110. But this variety is distinct from the species type in nearly all gross morphological characters. 266 GENETICS IN RELATION TO AGRICULTURE The Nature and Causes of Factor Mutations. — Our knowledge of genetic factors is entirely of an inferential sort and it is probable that these ultimate hereditary units are no more likely to be objectively perceived than are the atoms of which all matter is generally believed to be composed. But our present understanding of biochemistry and the chromosome mechanism of heredity leaves no room for doubt concerning the theoretical nature of these factors. Living protoplasm is generally considered as composed of very complex organic compounds. The phenomena of stereochemistry, especially the substitutional or cyclic changes which occur within various compounds under proper con- ditions, suggest that similar compensatory relations exist between the substances composing the living cell. Yet cytological observations indicate that the chromatin is the only permanent constituent of the nucleus and that the chromosomes are unaffected by the regular physio- logical processes of metabolism, growth and reaction to stimuli even though they play a very definite role in all these activities. As was explained in Chapter IV, the chromosomes are linear series of loci whereat are located specific factors. According to the multiple allelomorph hypothesis more than one factor may exist at a given locus. Since the chromosomes appear to consist of the only permanent substance in the nucleus, it is conceivable that at each locus there exists a unique chemical system; yet it is not unreasonable to suppose that occasionally substi- tutional changes similar to those known to take place in less complex organic compounds may occur here. The contributions of Eeichert on the specificity of proteins and carbohydrates as a basis for the classification of animals and plants are based on the fact that such substances as serum albumin, hemoglobin, glycogen and starch exist in stereoisomeric forms. That is, ''each kind of substance may exist in a number of forms, all of which forms have the same molecular formula and the same fundamental properties in common, but each in accordance with- variations in intramolecular configuration has certain individualities which distinguish it from others. ... It has been found that the number of possible forms of each substance is de- pendent upon the possible number of variations of the arrangements of the molecular components in the three dimensions of space, or, in other words, of variations of molecular configuration, the possible number in case of each substance being capable of mathematical determination. Thus, we find that serum albumin may exist in as many as a thousand million forms. Hemoglobin, the red coloring matter of vertebrate blood, is a far more complex carbon compound than serum albumin, and theoret- ically may exist in forms whose number is beyond human conception, running into millions of millions. The same is true of starch." Having in mind this complex molecular structure of protoplasmic constituents MUTATIONS 267 and the phenomenon of substitutional changes of atoms or radicals by which such complex compounds are transformed, we can express a conception of the nature of factor mutations. To be specific let us suppose that some unusual condition occurs in a certain germ cell of a normal female Drosophila such that a single atom in each of the very complex molecules of the substance unique for the locus W in the X-chromosome changes place with a different atom in the surrounding nucleoplasm — the substance unique for the locus W is no longer capable of conditioning the laying down of red pigment in the eyes and, if the affected ovum is fertihzed by a F-bearing sperm, a white-eyed male appears, the result, as we say, of a factor mutation. This conception of factor mutations is useful as a basis for the multiple allelomorph hypothesis. In order to explain how two or more factors may have the same locus in a chromosome, it is only necessary to assume as possible the substitution of two or more different atoms or radicals in the molecule of the complex organic substance unique for the given locus by other atoms or radicals in the nucleoplasm. Factors are relatively stable entities however. It has been shown al- ready that any organism must possess thousands of factors, yet mutations are comparatively rare even in Drosophila. These facts are rather difficult to harmonize with our conception of the nature of factor muta- tions. If substitutions of atoms or radicals occur why do they not take place more frequently? Such questions must remain obscure until we know something about the chemical constitution of the hereditary factors. Only then can we expect to understand clearly the nature of the altera- tions which occasionally are made in them. In this connection the behavior of factor mutations in inheritance is of decided interest. As a rule they are recessive to their normal allelo- morphs and for some time they were thought to be due to the loss of factors, this idea being associated with the presence and absence hypothe- sis. But on rare occasions dominant mutations have appeared. Among 150 mutations from the normal type of Drosophila ampelophila several, such as bar eye, dark streak on thorax, abnormal abdomen and CIII, a factor which modifies eosin eye color, are dominant over their respective allelomorphs. A few other mutant characters have been found to be dominant, such as hornlessness in cattle and red buds in the evening prifti- rose (CEnothera rubricalyx), but the great majority are recessive as is indicated by the ratio in F2 from crosses between mutants and normal individuals. The condition in Fi by no means always indicates complete dominance of the normal character. Hence it is clear that whatever the nature of the mutation-producing chemical change may be, as a rule it is either completely subordinate to the normal condition or else it merely modifies the effect of the normal state in heterozygous individuals, making 268 GENETICS IN RELATION TO AGRICULTURE its own distinctive manifestation in one-fourth of the progeny of such individuals. When we enquire as to what are the particular conditions or specific antecedent events that make possible or cause the assumed substitution of atoms or radicals, we find ourselves again confronted by an almost total lack of knowledge. One thing is certain however, namely, that factor mutations are not fortuitous in occurrence, because, if they were the outcome of wholly indeterminate series of events, they would be as likely to occur in one species or race as in another at a given time and with the same relative frequency under all conditions, but such is not the case. On the contrary, certain species appear to be much more prolific in factor mutations than others and, as stated in Chapter II, it would appear that inheritable variations can be induced under controlled environmental conditions in pedigree strains that have bred true for a number of generations. Furthermore, even though our knowledge of the occurrence of factor mutations were so meager as to furnish no basis for reasoning and even though future observations of the same might seem to indicate that they are fortuitous, we should still be justified in assuming the existence of specific causes for factor mutations. It has been clearly shown by Pearl that, while natural phenomena are the result of long series of antecedent events or con- ditions, yet these are not all of equal determinative value; but rather that there are always specific causes which are few in number, immediate in time and large in relative quantitative effect. It does not seem necessary to present here the course of reasoning on which this con- clusion rests. The important thing for agriculture is the fact that factor mutations are caused and the possibility that some of the deter- minative antecedent conditions are external to organisms, i.e., that they exist in the environment and are controllable by man. The prob- lem of the exact nature of factor mutations is only a phase of the general problem of the nature of living protoplasm, the solution of which is one of the ultimate aims of biology. But it is possible at least that experimental research may reveal methods by which factor mutations can be induced in both plants and animals. Factor Mutations Both Germinal and Somatic. — Factor mutations appear to occur in undifferentiated cells, the germ plasm or embry- onic tissue in animals and either the germ cells or any meristematic tissue in plants. Occasional discontinuous variations are found in animals which might seem at first to be due to factor mutations in the developing soma. But most of these abnormalities are more satis- factorily explained in other ways. Thus, gynandromorphism, or the condition of having one side of the body male and the other female, has been reported in insects more than a thousand times according to MUTATIONS 269 Morgan. Without doubt it is caused by some irregularity in the proc- ess of fertilization. Homeosis, or the replacement of one organ by another, is known to have followed mutilation. Examples of the modi- fication of characters by environmental conditions are given in Chapter II. There are many similar variations in animals, none of which are hereditary. However, we shall again refer to the possibility of somatic mutations in animals. There is no direct evidence as to the cytological time of factor muta- tions, but the stage in the germ cell cycle of animals at which factor mutations are most likely to occur would seem to be shortly before or during the process of maturation. This is indicated by the sporadic appearance of mutants. The first observed mutation in Drosophila ampelophila was white eyes, which were found in a few males among several hundred individuals in a pedigreed red-eyed race. Similarly with other sex-linked mutant characters that have been observed in this species, they have appeared either singly or at most in a few individuals. Had these mutations occurred at an earlier stage in the germ cell cycle, more gametes would have been affected and more mutant individuals would have been found. Obviously the length of time that must elapse before a factor mutation can manifest its existence depends upon two things in addition to the stage in the germ cell cycle at which it occurred : (1) its relation to its normal allelomorph, i.e., whether it is dominant or recessive; (2), its relation to sex determination, i.e., whether it is sex- linked or not. A mutation from W to win an X-chromosome of a normal male Drosophila would have produced a heterozygous red-eyed female in the next generation and no white-eyed flies whatever. One-fourth of the progeny of such a female would in turn be white-eyed if she mated with a normal male. Similarly with any non-sex-linked recessive cha'racter which upon its first appearance in pedigree culture is found in more than a single individual the probable order of events is as follows. A muta- tion occurred in a single germ cell of a single individual, which mated with a normal individual, thus giving rise to one heterozygote among its progeny. This heterozygous individual mated with a normal individual, producing heterozygotes among one-half of their progeny. Finally some of these heterozygotes mated together and one-fourth of their progeny bore the recessive mutant character. It would seem, therefore, that factor mutations in animals occur in the germ cells shortly before or during maturation and the time of appear- ance of a mutant character depends upon the relation of the mutant factor to its normal allelomorph and whether or not it is contained in the sex chromosome. In plants factor mutations may occur in any meristematic tissue as well as in the germ cells. Observations on the occurrence of mutant 270 GENETICS IN RELATION TO AGRICULTURE seedlings indicate that, as in animals, germinal mutations usually occur just before or during the maturation process. The strongest evidence for this conclusion is the fact that, so far as known, new dominant char- acters appear first in only one or two individuals. The following cases illustrate this point. The red-leaved evening primrose, (Enoihera rubricalyx (Fig. 118) has been known to occur but once in all ffinothera cultures and then in a single plant. The red sunflower, Helianihus lenticiilaris coronatus, as reported by Cockerell, first appeared as a single plant which proved later to be a heterozygous dominant. A purple-leaved mutation in hemp, Cannabis saliva, is reported by Dewey to have first appeared in two pistillate plants in a closely inbred strain of normal green plants. Had these mutations occurred at some preliminary stage in germ-cell formation, the change in chemical constitution would have been transmitted to several or many gametes and a considerable number of individuals would have appeared instead of only one or two. Factor mutations in meristematic cells, or vegetative mutations, as distinguished from those originating in the germ cells, give rise to simple bud sports or to chimeras according to the location of the mutating cell. A bud sport is a shoot or branch which differs genotypically in one or more characters from the remainder of the plant. Here the factor mutation must occur in one of the undifferentiated cells of the very young shoot. Just as in the case of factor mutations in germ cells, so in vegetative muta- tions the somatic effects range from single visible character differences to manifold effects in which many structural details are different. An example of bud sports in which the factor mutation induced a single character difference is shown in Fig. 111. The early gladiolus known as ''The Bride" is a white variety of Gladiolus colvillei, a red-flowered form, and doubtless originated from it as a seed or bud mutation. In 1915 there appeared in a row of ''The Bride" a single stalk bearing partly red and partly white flowers. That this grew from a corm which was an offshoot from a typical white-flowering corm is certain. Furthermore, that the mutation occurred very early in the development of this corm and not sometime during the growth of the flower stalk is proved by the following observation. In the autumn Fig. 111. — Bud sport from a white flowered gladiolus bearing red flow- ers on ore side of the stalk and showing one flower half red and half white; a sectorial chimera (see Chapter XXII). MUTATIONS 271 following the discovery of the mutant stalk it was carefully lifted and the corm from which it grew was separated from the cluster of white- flowering corms. It was observed that there were smaller corms located very close to the mutant corm. The following spring one flower stalk bore red and white and the other only red flowers. In gladiolus the young corms push out from near the base of the old one. Hence the original mutant corm must have consisted partly of cells capable of producing red pigment in the flowers. That the cells having this altered chemical constitution comprised about one-half of the corm is indicated by the position of the red and white flowers on the stalk. This illustra- tion is hardly typical of all bud sports in that the mutation occurred too late in the development of the young shoot to change all the cells in the corm and so make all the flowers red. It was chosen first, because the mutant character is dominant/ which makes it certain that the sport was due to mutation rather than to segregation, and second, because it also illustrates the origin of chimeras. In many cases of discontinuous bud variation the entire shoot is affected. Cases of bud variation pre- sumably caused by factor mutations which condition manifold character differences are occasionally found in the citrous fruits. The so-called Australian Navel orange has undoubtedly arisen a number of times from the commerical variety, the Washington Navel orange, from which it differs in its propensity to rank vegetative growth combined with low productivity. Also the fruits are rough and of poor quahty. Numerous other distinct types of oranges and lemons have been discovered, usually as a single tree or merely a branch on a tree of the commonly cultivated variety (see Fig. 161). A chimera is a mixture of genotypically diverse tissues in the same shoot. The nature, categories and artificial production of chimeras and graft hybrids are discussed in Chapter XXII. Here it is only necessary to point out that as they occur in nature they undoubtedly owe their origin to factor mutations. In the red and white flowered gladiolus an entire shoot became composite in nature through a factor mutation in a meristematic cell very early in the development of the stem. If the mutation had occurred later on at just the right point in the vegetative cone, it might have produced a single red and white flower. This is apparently the manner of origin of the odd stripes on certain fruits such as the lemon shown in Plate II. In this case it is evident that mutations occurred in two different cells. In one case the factor change resulted in the laying down of yellow pigment of a deeper shade ("deep chrome," No. 176 of Ridgway's Color Standards) than that normal for the variety, which is lemon chrome. In the other case the mutation 1 G. coluillei is a hybrid between G. cardinalis, which has bright scarlet flowers and G. tristis, which has white or yellowish flowers. 272 GENETICS IN RELATION TO AGRICULTURE resulted in the production of some red pigment along with the yellow, thus causing the narrow sector of deep orange chrome (Ridgway, No. llh). That each of these changes occurred in a single cell is indicated by the fact that the differently colored sectors are sharply defined through- out and that the extremities of the orange red sector are extremely narrow. J. B. S. Norton reports the origin of a color chimera in the Acme tomato in which a branch of lighter green foliage appeared and the lighter colored tissue could be traced down the stem to a point where it had ap- parently originated in a single cell. Expanding as the stem grew, first a portion of a leaf was involved and finally an entire bud was included, thus giving rise to the sport branch. Undoubtedly this is the usual manner of origin of natural chimeras. We have examined several typical cases of factor mutations in animals and plants. From this evidence it is clear that factor mutations occur in undifferentiated cells — the germ cells in animals and either the germ cells or any meristematic cell in plants. There is, of course, no a priori reason why mutations should not occur in the somatic cells of animals. A fairly common meristic variation is the reduplication of repeated parts and it is possible that this departure from normal development is con- ditioned by a factor mutation. The discovery of a germinal mutation causing reduplication in animals would support this idea. Such a mutation has been discovered by Miss Hoge who reported a recessive factor for reduplication of the legs in the Drosophila. This possibility of somatic factor mutations in animals has little practical significance on account of the impossibility of propagating domestic animals asexually. It has considerable theoretical interest, however, in its possible bearing on the origin of certain diseases such as cancer. Vegetative Mutation Versus Somatic Segregation. — Since the ma- jority of bud sports are characterized by the replacement of a dominant with a recessive character, it is not strange that both bud sports and chimeras have been generally considered as due to ''somatic segregation" in heterozygous individuals. It is not yet known whether bud sports occur more frequently in heterozygous than in homozygous individuals. But this consideration is of less importance than the fact that somatic factor mutations do occur, which seems to be well established. To mention an illustrative case, Emerson has shown that the experiments of de Vries, Correns, Hartley, East and Hayes, and himself, ''all indicate that certain somatic variations are inherited in strictly Mendelian fashion. All these somatic variations consist in the appearance of self-colors on plants that are normally variegated in pattern. The fact that variegated plants occasionally throw both bud sports and seed sports with self- colors is not, in general, to be taken as an indication that the variegated plants in question are heterozygous. Such behavior seems to be insepa- o. Plate II. — Chimera in a Lemon. The broad sectoi of orange and the narrow sector of orange red were caused by factor mutations which occurred presumably in single cells at a very early stage in the development of the fruit. MUTATIONS 273 rably associated with variegation. Correns has pointed out that variegated Mirabihs plants cannot be considered mosaics of green and 'chlorina' types due to heterozygosis, since they do not segregate into chlorina and green, but into variegated and green. The same reasoning apphes to variegation in the color of maize ears. Variegated-eared plants do not throw reds and whites, but reds and variegates. The conclusion seems irresistible that self-color occurring as a somatic variation is due to the change of a Mendelian factor for variegation into a factor for self-color. If this be granted, the behavior of these variations in later generations is a mere matter of simple Mendelian inheritance." If bud sports are caused by mutations and if most bud sports involve a change from the dominant to the recessive condition of a certain factor, it follows that the change in chemical constitution must affect both of the Fig. 112. — Bud sport and chimera in an ear of corn. This ear appeared in a field of white dent corn. The apparently white Ivernels, occupying about 3^ of the surface, were actually variegated, being marked with "fine red lines, or streaks, radiating from the caps down the sides of the kernels." (After Hartley.) duplex factors present in the somatic cell in order that the recessive character may appear. To those who think of mutations as fortuitous events, this may seem an obstacle to the conception that bud sports are the result of factor mutations. But from the point of view that factor mutations are caused, probably by some specific internal condition, it would seem most natural for the cause to have the same effect on both factors. Obviously this conception assumes that in such cases the specific cause, whatever it is, has the same potentiality in all parts of the nucleo- plasm, and there is no a priori logical objection to such an assumption. At the same time there is good evidence that mutations do sometimes occur in only one of a duplex pair of factors. Hartley reports "a remark- able ear (Fig. 112) occurring in a field of white dent corn which had for many years been grown as a reasonably pure corn, but which occa- sionally, as many white corns do, produced a red ear." But this ear was only partly red since about one-fifth of its surface was occupied by varie- gated grains which appear to be white in the picture. Hartley tested all the grains on this ear and found thai, the red grains produced a crop of 84 red ears and 86 pure white ears, while the variegated grains 18 274 GENETICS IN RELATION TO AGRICULTURE produced 39 variegated ears and 36 pure white ears, which is clearly a 1 : 1 ratio in each case. This proves that both types of grains were heterozygous for a dominant mutant factor and that both of the factor mutations occurred in only one member of a duplex pair of factors. Pre- sumably the mutation from white to variegated occurred first, and later the mutation from variegated to red in a cell so located that, as the shoot developed, only a portion of the ear was affected. There appears a very important obstacle to the conception of "somatic segregation" in that the mechanism of cell division is apparently one of the most nearly perfect and regular of natural systems and that the order- liness of procedure is especially notable in undifferentiated tissue, where bud sports and chimeras commonly originate. To assume that the oc- currence of self-colored flowers on variegated plants is due to chromosome aberrations in mitotic divisions is much less plausible than to explain such phenomena by assuming a simple factor difference as responsible for self- color and variegation, and that changes from one state to the other are possible under certain conditions. This is the only reasonable hypothesis by which to explain mutations from the recessive to the dominant condi- tion of a pair of factors, as we have seen in the case of Hartley's ear of red and variegated corn. Therefore, while chromosome aberrations are known to occur during mitosis and aberrant numbers of chromosomes have been found in senile and diseased tissues, yet, in general, bud sports and chimeras are satisfactorily explained on the basis of factor mutations ; whereas "somatic segregation" as the term has been used by Bateson, Gates and others implies the common occurrence of breaks in the mechanism of mitosis such as are not known to occur in normally functioning somatic cells. It should be remembered that horticultural literature contains nu- merous peculiar cases of discontinuous variation, many of which have been described or "explained" as "somatic segregations" resulting from hybridization. We believe that most of these cases can be explained much more reasonably in terms of factor mutations. But certain discontinuous variations in plants are undoubtedly the result of neither factor mutations nor chromosome aberrations in vegetative tissues. For example, persistent and deciduous calyx lobes are sometimes found on fruits of the same plant especially in the rose family. Tufts has described the occurrence of this phenomenon in the Le Conte pear and the Transcendant crab-apple as "somatic segregation," assuming that some sort of segregation-mechanism exists in the division of somatic cells. Data from the pear tree gave a ratio of 3.15 deciduous to 0.85 persistent lobes. But to assume irregularities in chromosome behavior such as would cause segregation preceding the formation of nearly one-fourth of the calyx lobes on the tree is unwarranted in view of the general regularity MUTATIONS 275 of the process of mitosis (see p. 60). It has been shown by Babcock and Lloyd that no special significance should be attached to the occur- rence of a ratio which, under the laws of simple sampling, could not occur oftener than once in 1,155,000,000,000,000 times, especially in view of the fact that these two varieties are presumably complex hybrids, and the persistency and deciduousness of the calyx lobes were variable in the parents. Hence to use the term somatic segregation in attempting to explain phenomena such as these is not only unwarranted but posi- tively misleading. The multifarious manifestations of dimorphism in plants arc, in general, the result of alternative expression of inherited characters rather Fig. 113. — Transition from..one form of leaf to another on the same branch in (a) Euca- lyptus globulus (b) Hedera helix. than alternative transmission of different factors. There are, to be sure, various cases of dimorphism within species, such as the different forms of flowers described by Darwin or the zygomorphic and peloric snap- dragon flowers, which usually do not appear together on the same plant and which exhibit alternative inheritance when crossed. But there are many species which bear different forms of branches, leaves, flowers or other organs on the same plant. Cook has described dimorphic branches in cotton, coffee, cacao, the Central American rubber tree and the banana, also dimorphic leaves in cotton, hibiscus, okra and allied genera. The open and cleistogamous flowers of the violet make a familiar example'of dimorphism in the same plant. In all these cases it appears that the individual plant contains all the factors conditioning the expres- 276 GENETICS IN RELATION TO AGRICULTURE sion of the alternative forms. It seems reasonable then to explain the variations in somatic expression of the genetic factors present by internal changes of some sort. Frequently these variations appear as localized stages in ontogeny and it is possible that internal secretions (hormones) play a more important role in plant development than has been realized. The recent experiments of Loeb on Bryophyllum calycinun indicate not only the association and possible identity of root-forming and geotropic substances in this plant, but also that the leaves produce growth inhibit- ing substances which pass downward through the stem and which may accompany or may be identical with the root-forming hormones. Cook has shown that sometimes two extremely different forms of leaf occur on adjacent nodes but even such abrupt transitions might result from an internal reaction occurring in the interim between the development of the two successive leaves. Moreover, the transition from one leaf form to the other is frequently gradual as in the two series, each from a single branch, shown in Fig. 113. We conclude, therefore, that most cases of dimorphism in the same plant are not caused by factor mutations but rather that they should be classified with those cases of '' fixed dimorphism " so frequently found in insects and illustrated by the earwigs in Fig. 20. " Mutations" in the Evening Primroses. — Credit for directing atten- tion to suddenly appearing new forms of animals and plants both as material for origin of species and for improvement of domesticated races belongs to the Dutch botanist, Hugo de Vries. Other naturalists had previously noted such aberrant or anomalous organisms but without attaching much significance to them. Thus in the works of Darwin, especially in "The Origin of Species" and "Animals and Plants under Domestication," there are frequent references to aberrant individuals or sports and to curious groups of plants and animals like the niata cattle, which Darwin admits probably originated as definitely distinct individuals among the typical species group. Yet Darwin never con- sidered such aberrant individuals or groups as playing any significant role in evolution. On the other hand, de Vries became so convinced of the general occurrence and significance of suddenly appearing, heritable variations that he proposed a theory of evolution by mutation in which he applied Darwin's great principle of natural selection to these mutations as the general method of origin of species. The investigations which led him to this conviction extended over a period of nearly 20 years, dur- ing which time he brought under experimental cultivation some hundred species of plants that grow wild in Holland. They all exhibited more or less continuous variation; also he was able to isolate numerous strains which differed from the normal wild type with respect to some peculiar feature. But de Vries was searching for evidence of species "in the making" and he believed that by sufficient searching he should locate MUTATIONS 277 a species in which the transformation into new forms was proceeding on a scale large enough to make possible the direct observation of species formation. In none of the particular races that he collected did he observe profound discontinuous variations until in 1886 he discovered a feral group of large-flowered evening primroses {(Enothei'a lamarckiana) growing in a suburb of Amsterdam. They had escaped into an abandoned potato field from a nearby park. The source of this particular even- ing primrose has been traced by de Vries. About the middle of the 19th Fig. 114. — CEnolhera lamarckiana. {From a i^aiiitiiiy. See de Vries, Gruppenweist Art- bildung.) century seeds of CE. lamarckiana were imported into England from Texas. De Vries' race came from an estate near Hilversum, the seed having been obtained originally from an establishment in Erfurt, which de Vries thinks must have obtained their seed from England. It has never been found as an indigenous species either in Europe or America. This beautiful plant is much prized as an ornamental and is known to have escaped from cultivation in various places. "Lamarck's evening primrose is a stately plant, with a stout stem, attaining often a height of 1.6 meters and more (see Fig. 114). When not crowded the main stem is surrounded bj^ a large circle of smaller branches, growing upward 278 GENETICS IN RELATION TO AGRICULTURE from its base so as often to form a dense bush. These branches in their turn have numerous lateral branches. Most of them are crowned with flowers in summer, which regularly succeed each other, leaving behind them long spikes of young fruits. The flowers are large and of a bright yellow color, attracting immediate attention, even from a distance. They open toward evening, as the name in- dicates and are pollinated by bumblebees and moths. Contrary to their con- geners they are dependent on visiting insects for pollination. (E. biennis and CE. mvricata have their stigmas in immediate contact with the anthers within Fig. 115.^ — Leaf, flower bud, flower and essential organs of (Emothera rubrinervis (1-4) and CE. brevistylis (5-8). The specimen of brevistylis came from a red pigmented strain grown by Dr. R. R. Gates; the original brevistylis had no more red pigment than lainarckiana. the flower buds, and as the anthers open in the morning preceding the evening of the display of the petals, fecundation is usually accomplished before the insects are let in. But in CE. lamarckiana no such self-fertilization takes place. The stigmas are above the anthers in the bud, and as the style increases in length at the time of the opening of the corolla, they are elevated above the anthers and do not receive the pollen. Ordinarily the flowers remain sterile if not visited by insects or pollinated by myself, although rare instances of self-fertiliza- tion were seen .... Ordinarily biennial, it produces rosettes in the first, and stems in the second year" (de Vries). De Vries' original discovery consisted of the location of two aberrant groups among several thousand lamarckiana individuals. One of these new forms had smooth leaves and was named Icevifolia, the other had MUTATIONS 279 very short styles and was named hrevistylis. Each differed somewhat with respect to other characters as well (see Fig. 115) but were named for their most striking difference from the parent species. De Vries next proceeded to hunt for more new forms. By transplanting rosettes from the original locality to his garden he carefully compared them when they flowered the second year and saved guarded seed from the "mutants" so that he might test the inheritance of the new forms. He also gathered seed from two different lamarckiana plants in the open and these were the source of his lamarckiana "families." Of one of these families he raised several thousand plants from self-fertilized seed in each generation for seven generations and in each successive population he discovered a number of "mutants." This experiment is summarized in Table XLIV. Table XLIV.- -Pedigree of a Family of CE. Lamarckiana , 1886- -1899 Generation Gigas Albida Oblonga Rubrinervis Lamarckiana Nanella Lata Scintillans I 9 II 15,000 5 5 III 1 10,000 3 3 IV 1 15 176 8 14,000 60 73 1 V 25 135 20 8,000 49 142 6 VI 11 29 3 1,800 9 5 1 VII 9 3,000 11 VIII 5 1 1,700 21 1 Total 1 56 350 32 53,509 158 229 8 This summary shows that among a total of some 50,000 seedlings of self-fertilized lamarckiana plants seven different new forms appeared with varying frequency. The first two to be recognized and isolated for testing were nanella and lata (see Fig. 116). The dwarf variety, nanella, was also found blooming among the typical lamarckiana plants at the original station and these plants, like those dwarfs that appeared in the experimental garden, always bred true if self-fertilized. Lata on the other hand proved to be self-sterile because entirely devoid of viable pollen. When fertilized with larnarckiana pollen it produced 15 to 20 per cent, of lata and the remainder lamarckiana. Later on de Vries discovered a hybrid strain of lata that produced some viable pollen and when self-fertilized these plants produced the same proportion of lata and lamarckiana progeny. For this reason de Vries considers it an inconstant species. In the third generation another new form appeared, which unlike nanella and lata was more robust than lamarckiana. It 280 GENETICS IN RELATION TO AGRICULTURE also had considerable more red pigment in the epidermis. This was especially marked in the marginal region of the sepals (see Fig. 118) and on the developing fruits. This form was named ruhrinervis and since it bred true when self-fertilized it was considered a well-defined "progressive" species, i.e., a species capable of maintaining itself in the Fig. 116. -CE. lamarckiana, lata and nanella. {From colored plate iv de Vries' Mutatiorts- theorie, vol. 2.) wild state. In the fourth generation four additional forms were found. One of these, gigas, appeared only once in de Vries' cultures but the one plant found in 1895 produced nearly 300 plants of gigas type from self-fertilized seed and the strain bred true thereafter. In recent years it has appeared a number of times in other strains of lamarckiana. This form is decidedly more robust than lamarckiana and the leaves are MUTATIONS 281 broader and of a deeper green, as is shown by Fig. 117. De Vries classified this form also as a progressive species. Albida and ohionga were classified as distinct but weak species, incapable of perpetuating themselves in a state of nature and hence were called "degressive." The seventh form, scintiUans, proved to be inconstant from the beginning, self-fertilized seed always producing some lamarckiana seedlings, as Fig. 117. — (E. lamarckiana {left) and gigas (right), flower stalks and rosettes. {From de Vries.) well as scintiUans, ohionga, lata and sometimes nanella. The two new forms found growing wild in 1886, Icevifolia and hrevistylis, never appeared among the experimental cultures but because they, like nanella, appeared to have lost something that characterized lamarckiana and to be dis- tinguished from it by one definite character, de Vries classified these together with nanella as "regressive" species or "retrograde" varieties at the same time pointing out that they possessed the qualifications of — nanella, loBvifolia, brevistylis. 282 GENETICS IN RELATION TO AGRICULTURE elementary species. De Vries, therefore, classified the "mutants" as follows: I. Progressive species — gigas, ruhrinervis. II. Degressive species — albida, oblonga. III. Regressive species or retrograde varieties IV. Inconstant species — lata, scintillans. These were considered the important mutations although several others were recognized and given names but because of sterility or lack of space and time for growing them they were not preserved. The Mutation Theory of Evolution. — Based on the observations and experiments above reviewed, de Vries formulated a set of "laws of mutability" for the evening primroses which serve as an epitome of his theory of evolution. Omitting further discussion of the evidence for the present, the laws themselves are stated as follows : I. New elementary species appear suddenly without intermediate steps. II. New forms spring laterally from the main stem. III. New elementary species attain their full constancy at once. IV. Some of the new strains are evidently elementary species, while others are to be considered as varieties. V. The same new species may be produced in a large number of individuals. VI. Mutability is entirely independent of fluctuating variability. VII. The mutations take place in nearly all directions. To these an eighth must be added in order to complete the theory: VIII. Natural selection eliminates all unfit mutants originating in the wild. De Vries found many examples of the sudden origin of new forms in the history of domesticated animals and plants and pointed out various practical applications of his discovery, to some of which we shall have occasion to refer later. For the present it is necessary to give furtheJT consideration to the evidence in the case of the CEnothera "mutants" and to the interpretation thereof in order to arrive at a definite conception of the true nature of these aberrant forms. For this purpose it will be necessary to summarize in a general way the researches which have been made since de Vries' original work. The publication of "Die Mutationstheorie " aroused widespread interest and brought forth certain criticisms concerning the biological basis for de Vries' conclusions. The opponents of the theory assumed CEnothera lamarckiana to be of hybrid origin and pointed out that upon such a basis the so-called mutations are merely recombinations of ances- MUTATIONS 283 tral characters. The controversy which arose over these questions stimulated investigation to such an extent that the CEnothera Uterature of the past decade woukl fill many volumes. These investigations have proceeded along three definite lines: (1) crossbreeding experiments, (2) cytological studies, (3) observations and collections in the field to- gether with pedigree cultures. The last of these can only be mentioned. See the recent reviews of Davis and Bartlett where further references are given. The results of the other two bear directly upon the problem of classifying de Vries' original "nuitations." Concerning the first method of investigating the genetics of Oenothera, there have been many crossbreeding experiments in both Europe and America. Until recently most of the data derived therefrom have been viewed as impossible of interpretation on a Mendelian basis. But since 1914 certain investigators have come to believe that breeding experiments with Q]]notheras have very little value as a means for exact genetical analysis unless complete germination of all viable seeds is assured. In that year Renner pointed out that seed sterility in the evening primroses may cause apparent noncomformity with Mendelian principles. Following up this suggestion with seed germination tests and breeding experiments in which all viable seeds were germinated, Davis came to the conclusion "that large proportions of Q^^nothera seeds sprout in the earth only after many weeks or even months and that this habit of delayed germination must have given in many of the cultures described in the CEnothera literature hardly more than glimpses of the genetical possibilities. We cannot feel certain that the records of any cultures of Oenothera so far reported are complete for their possible progeny, and consequently the ratios of classes described in breeding experiments and the percentages of 'mutants' calculated cannot be accepted as final in exact genetical work. We are not in a position even to guess what may be the change of front when exact data become avail- able. . . . Consequently we have at present in the ffinotheras no standard material of genetic purity with which forms under suspicion -may be confidently mated to determine by crossbreeding the uniformity of their viable gametes. Until such material is discovered we shall be working largely in the dark in our attempts to analyze the genotypic constitution of CEnotheras." The same author, is inclined to interpret data from his most recent CEnothera breeding experiment {biennis and franciscana) as giving "positive evidence of a segregation of factors in the F2 generation of a character to be expected in Mendelian inheritance." This inference is the more noteworthy inasmuch as in the past this author has not committed himself positively to a Mendelian interpretation of any particular data on the CEnotheras. That certain characters in this group are conditioned by specific genetic factors, seems to be generally accepted. For example, it is highly probable that the deeply pigmented 284 GENETICS IN RELATION TO AGRICULTURE character of (E. ruhicalyx is conditioned by one or two specific factors. This new form (see Fig. 118) was discovered by Gates in 1907, among a population of over 100 ruhrinervis plants grown from self-fertilized seed of ruhrinervis. The original ruhricalyx plant when self-fertilized produced 12 plants, 11 ruhricalyx and 1 ruhrinervis, which would indicate that the original ruhricalyx plant was heterozygous for one or more factors for excessive production of anthocyanin and that ruhricalyx is dominant to ruhrinervis. Gates has raised ten generations of ruhrinervis (more than Fig. 118. — Flower bud and leaf of A, (Enothera ruhrinervis; B, CE. ruhricalyx. The deeper pigmentation of ruhricalyx is not confined to the bud and lower side of the leaves, but is also present_ in the stems. The rosette leaves also show more or less color on the vaiArih in j-uhricalyx. one pedigreed strain) and found it to breed true and he has one strain of ruhricalyx that has bred true for five years, but as yet there are no data on the results of a cross between them where the F% progeny were grown on a large scale and with controlled seed germination. However, in Fi ruhricalyx is dominant. Another O^^nothera character which is inherited in simple Mendelian fashion is the typical feature of hrevistylis (Fig. 115). Brevistylis is known to breed true when self-fertilized and the results of various crosses indicate that the short style is conditioned by a single factor, although it is not always completely recessive in Fi. Thus we find that, while most of the experimental breeding data on Oenotheras cannot be interpreted in terms of ordinary Mendelian concepts, neverthe- MUTATIONS 285 less characters have been discovered that appear to be inherited according to simple Mendelian rules. We conclude, therefore, that some of the lamarckiana derivatives are the result of factor mutations. The cytological studies on ffinothera have yielded important infor- mation concerning the chromosome numbers of various species and "mu- tants." With reference to (E. lamarckiana and its derivatives especially the chromosome counts of Miss Lutz, Gates, Davis and others are of great interest. Lamarckiana has 14 chromosomes as have also most of the "mutants" which have been derived from it, but the sexually deficient and inconstant form, lata (see Fig. 116) has been found always to have 15 chromosomes. Furthermore, actual cases of a distribution of 6 + 8 chromosomes in the heterotypic division of pollen mother cells have been observed in lamarckiana and rubrinervis. It is safe to assume, therefore, that lata-\ike "mutants" result from the union of a gamete containing 8 chromosomes and one containing the normal number, 7. There is also good evidence that (E. gigas is the result of tetraploidy. Several different plants of this type have been found to contain 28 chromo- somes or thereabouts. However, there is a giant race of the Chinese primrose which has only 24 chromosomes, the number typical of the species, while another has 48 chromosomes. It seems then that gigan- tism is associated with tetraploidy but that it is not necessarily caused by an aberration in chromosome number. Thus we find that at least one, and perhaps another of the original lamarckiana derivatives are due to chromosome aberrations during meiosis. Of the nine original mutants we have now definitely classified two — hrevistylis as a factor mutation and lata, the result of a departure from normal chromosome number, and we have found that a third, gigas, exhibits an extreme chromosome aberration. What about the remain- ing six — Icevifolia, albida, ohlonga, rubrinervis, nanella and scintillansf There is no evidence of a simple factorial relation between them and the parent species. One of them, scintillans, must remain in the doubtful class until its chromosome numbers have been determined, but the inconstancy of this form suggests that it should be classed with lata and gigas under chromosome aberrations. The remaining five, Icevifolia, rubrinervis, nanella, oblonga and albida, are known to have 14 chromo- somes. Based on the evidence set forth in Chapter XII, it seems to us that one and only one category is open to these five forms and that probably albida, oblonga and most of the new forms that have appeared not only in cultures of lamarckiana and its derivatives but also in other species of Q^^nothera, are the result of chance recombinations of factors due to a condition of substrate hybridity. This expression, as has already been explained, is meant to imply that "mutating" species such as ffi". lamarckiana are merely species hybrids which happen to result from combinations of different reaction systems such that the majority of 286 GENETICS IN RELATION TO AGRICULTURE their germ cells are similar. Hence they breed true in the main but occasionally throw the new combinations of diverse elements which have come to be known as ''mutants." In conclusion, it may be well to state our reasons for restricting the term, mutation, to those changes in specific factors, which result in the appearance of new Mendelizing characters. This term was used by de Vries to designate a more or less comprehensive change which appeared suddenly, without warning, giving the impression that a full-fledged new species had sprung from a pure, constant, old species much as Athena sprang from the head of Zeus. We cannot conceive of new species originating in this way except in certain exceedingly rare cases, which fall under the two categories already described and illustrated, viz., (1) single factor mutations having such a profound manifold effect that the new form would be generally recognized as a distinct species, and (2) chromosome aberrations during mitosis or meiosis. We have found that the majority of the new forms derived from (Enothera lamarckiana do not fall into either of these categories and that the most reasonable explanation of their origin is based on the assumption that (E. lamarck- iana is of hybrid origin. Therefore, if the term, mutation, is to retain the meaning originally given it by de Vries, we cannot continue to classify the majority of new (Enotheras or other organisms resulting from hybridi- zation as mutations. On the other hand, the fact that most discontinuous, inheritable varia- tions are caused by alterations in genetic factors and that these factor mutations play an important role as one means for organic evolution, seems to justify their recognition as mutations in the strict sense. By limiting the meaning of mutation as we propose all the objectionable im- plications previously connoted by the term are removed. The desir- ability of accomplishing this has been indicated by Agar, who states: "The greatest opposition to modern views of genetics has come from those who consider that they have taken away the philosophical basis of the theory of evolution and especially of the evolution of adaptation. For, while mutation could quickly bring about specific diversity, the evolution of complex adaptive structures is undoubtedly most easily grasped when the inheritable variations presented to natural selection are minute and abundant. This difficulty, though real, would undoubtedly have assumed smaller proportions had it not been for the natural fact that the earliest mutations studied were large morphological ones, and consequently that these have become fixed in many minds as types of mutational change." There is now abundant evidence that genetic diversity is expressed in minute morphological and physiological differences, and hence that mutations produce those small inheritable differences logically required for the explanation of adaptation through natural selection. PART II.— PLANT BREEDING CHAPTER XV HISTORICAL INTRODUCTION Plant improvement is nearly as old as agriculture. Our earliest agri- culturists must have protected the trees or plants that yielded food or shelter. Under protection the desirable forms among the chosen species were preserved. The finest example of this earliest plant improvement is found in rice, which has been cultivated for 5000 years or more in India and China and has long been grown in Egypt, East Africa, Japan, the Philippines, Java; Turkey and Italy. The remarkable plasticity of this species has enabled it to produce literally thousands of locally adapted forms. The oldest records of intentional preservation of superior plants are found, according to Darwin, in ancient Chinese encyclopedias that were translated by the Jesuits during the 18th century. The best plants and fruit trees were used for propagation; an imperial edict recommended the choice of large seed ; and even the Emperor Khang-hi is said to have originated the imperial rice by preserving and propagating a form which he noticed in a field. The original progenitors of our most important crop plants are mostly lost in antiquity, their descendants having been preserved by man's conscious or unconscious selection of desirable mu- tants or natural hybrids. The Beginning of Plant Breeding. — ^Long before any one thought of making a philosophical study of plant improvement the hybridization of flowers and the preservation of choice strains or favorite varieties was a common practice among gardeners and husbandmen. According to Fruwirth hybridization was practised in ancient times in China with various flowers, in Italy during the Roman Empire with roses, and in the 17th century in Holland with tulips and primulas; and the artificial pollination of the female date palm was mentioned by Theophrastus as the beginning of the study of plant culture. The earliest syste- matic work in the production of new varieties, of which we have authentic records, was done by the Dutch flower fanciers. The hyacinth, ac- cording to Darwin, was introduced into England in 1596 and in 1629 eight varieties were known. During the next hundred years or more the selection of varieties was carried on by the Dutch growers until, in 287 288 GENETICS IN RELATION TO AGRICULTURE 1768 nearly 2000 sorts were known in Holland. But in 1864 only 700 varieties were found in the largest garden in Haarlem, which fact in- dicates a gradual process of elimination of the less desirable selections of earlier years. Pioneers in Plant Breeding. — The systematic breeding of crop plants was begun in Europe during the latter part of the 18th century. Jean Baptiste Van Mons, a Belgian physician and professor of physics and chemistry in the University of Louvain, pursued plant breeding work as an avocation. But so great was his zeal in an effort to demonstrate certain theoretical ideas which he held concerning the improvement of fruits that the results of his labor were extensive. His experiments were begun in 1785. Thirty-eight years later he had 80,000 seedling trees in his "Nursery of Fidelity," as he called it, at Louvain. He dis- tributed cions without charge to many countries including America. He specialized on pears and his first catalogue, issued in 1823, lists 1050 varieties. Altogether he originated nearly half that number. Van Mons' service to agriculture, especially to pomology, has been widely recognized.^ Three other pioneer breeders who began their work during Van Mons' life are Thaer, Knight, and Cooper, representing Germany, Eng- land, and the United States respectively. During the latter portion of the 18th and the earlier years of the 19th centuries each of these men carried on experiments in plant breeding and made contributions of tremendous importance to agriculture. Thomas Andrew Knight was the first to show the value of hybridization in plant improvement. Accord- ing to Bailey, in the variety, accuracy, significance, and candor of his experiments. Knight stands to the present day without a rival among horticulturists. He was also a successful breeder of livestock and author of papers on plant physiology and breeding. Albrecht Daniel Thaer also made hybridizing experiments but emphasized the value of selection. Plant breeding was only one of his many agricultural interests and he is credited with having laid the foundation of scientific agriculture of today. Joseph Cooper disproved the current fallacy as to the entire necessity for changing seed and showed the American farmer the impor- tance " of selecting seeds and roots for planting or sowing, from such vege- tables as come to the greatest perfection, in the soil which he cultivates." Like Van Mons, each of these men had his theories, but only experience revealed the truth in those theories. Based on their experience they formulated certain rules which they knew would yield results, but fre- quently the conclusions reached by them were only partially true. At least five other men deserve to be mentioned among the earlier ^ For a discussion of Van Mons' theories and contributions (also of the work of Knight and Cooper) see Bailey, L. H.: "Survival of the Unlike," 1906, pp. 141-159. INTRODUCTION 289 breeders of agricultural crop plants. Three of these worked with grains and two with fruits. John Le Couteur, during the early part of the last century, was raising what he supposed were pure and uniform varieties of wheat, when Professor La Gasca of the University of Madrid, after examining one of his fields, pointed out 23 distinct forms. This was the beginning of Le Coutcur's collection of 150 varieties of wheat, some of which were introduced to the trade. One of them, "Bellevue de Tala- vera," is still known as a pure and uniform variety. De Vries points out that Le Couteur simply assumed that the progeny of his selected plants would be like the parents and experience justified the assumption. Thus be became the first to discover the importance of selecting in- dividual plants in the improvement of cereals. Patrick Shirreff was also celebrated about the middle of the centurj^ as a breeder of cereals. His method differed from that of Le Couteur only in that he searched for very exceptional plants as the starting points of new varieties. During his lifetime he discovered seven new varieties, which according to Darwin, were grown extensively in Great Britain, but only four of them had permanent value. He also proceeded on the assumption that his single selected plants would breed true and each did so. According to de Vries, he considered the occasional appearance of a distinctly superior plant as merely accidental. Frederic F. Hallet, like Le Couteur and Shirreff, practised the rigid selection of individual plants in breeding wheat. Although he proceeded on the theory that by choosing the best spike on a certain plant and the best grain in the spike he would obtain corresponding improvement in the variety, yet he did not rely on mere apparent superiority, but tested each grain on each spike. He then selected the finest plant of all. He began his work in 1857 and made important introductions during the 60's. While Le Couteur and Shirreff assumed that the selection of a single plant was sufficient and thenceforth gave their attention to multiplying the new variety, Hallet practised continuous- selection within his selected strains. He obtained considerable increase in yield as a result of his early selections but little or no increase due to continuous selection within pure strains. The success of these three pioneer wheat breeders was unquestionably due to the fact that they practised the isolation of pure lines some of which were superior to ordinary varieties. Charles Mason Hovey was the "father of the American strawberry." As early as 1830 Hovey "had a list of 30 strawberries of his own origina- tion, all springing from the "Hovey," which, together with "Boston Pine," had been introduced a few years earlier. Hovey crossed a native American species with the imported "Pine" variety, which is supposed to have sprung from the beach or sand strawberry of the Pacific Coast some years after its introduction into Europe. Some of Hovey's 19 290 GENETICS IN RELATION TO AGRICULTURE new varieties stood the test of years and his work served to stimulate further efforts to improve the most important horticultural crop of America. Ephraim Wales Bull produced the Concord grape as a result of eleven years of patient work in crossing the native species, Vitis lahrusca, with European varieties, raising the seedlings and testing selections. "From over 22,000 seedlings there are 21 which I consider valuable," he writes. Although the hybrid nature of the Concord and other deriva- tives of Vitis lahrusca has been questioned, the evidence from extensive tests of selfed seedlings of this and several other standard American varieties as reported by Hedrick and Anthony seem to indicate that they are really hybrids between American species if not between V. lahrusca and V. vinifera. Whatever the origin of the Concord may have been, its sterling value is evidenced by its history. Introduced in 1853, "ten years later the Concord grape was spread over the entire northern part of the United States and is now widely used in the temperate regions of most parts of the earth." Ephraim Bull's service to his fellow men seems to have been all but forgotten while he was still living, since "he died neglected, in poverty, broken in spirit." Vast as would be the value of his contribution if it could be computed, even more valuable was the inspiration he gave, "which has helped to make plant breeding one of the great forces in cheaply feeding the world. ^" The demands and possibilities of developing agriculture aroused the ambitions of two far-sighted agriculturists — Martin Hope Sutton and Pierre Louis Frangois Leveque de Vilmorin. A student of botany from his boyhood, Sutton had already made improvements in a number of plants when the Irish potato famine of 1847 drew public attention to his work through the substitutes which he suggested for the devas- tated potato crop. Later on the introduction of the Golden Tankard mangel, the Magnum Bonum potato, and the Marrowfat pea helped to establish the high reputation which the firm of Sutton and Sons came to hold throughout the world. They greatly improved many flowers as well as crop plants. Sutton's "Permanent Pastures" is still a standard work on grasses. In 1843 Vilmorin took charge of the seed establishment which had already passed through the hands of two generations of this remarkable family. His father, Andre Leveque de Vilmorin, had conducted a selec- tion experiment with carrots about ten years earlier. Besides the main- 1 The earliest hybridizers of grapes in America, according to Waugh, were Dr. Wm. Valk of Long Island (1845) and John Fisk Allen of Massachusetts (1846 or '47). Waugh also states that the two foremost American grape hybridists are E. S. Rogers of Massachusetts, who began in 1848 and distributed many numbered seedlings for trial in 1858, and T. V. Munson of Texas, who has probably added more to the prac- tical American fruit list in his hybrid grapes than has any other plant breeder. INTRODUCTION 291 tenancc of work already iiiulor way, Vilmoriii undertook two projcctvS which yielded results of the greatest importance to the entire worhl. One was the collection of wheats and other grains from many countries in order to compare them and to select those of greatest value. In con- nection with this work on grains he invented the pure-line method of selection and progeny test which came to be known as the "Vilmorin Method" and which has been used so successfully with wheat and other self-fertilized plants. From our present knowledge of pure lines we can understand why this method was effective. Vilmorin's other im- portant project was the improvement of the sugar beet. Previous to 1850 the beets had been selected according to form only. This method of selection began as early as 1787 on the seed farm of H. Mette in Qued- linburg, Germany, according to Legier. Selection on the basis of specific gravity was practised from 1850 to 1862, when the method of determining sugar content by means of polarized light was introduced. His success made beet-sugar production a commercial i)()ssibility and gave his name an enduring place in history. More Recent Progress in Plant Breeding. — The great world move- ments of the 19th century following the improvement of transportation facilities, the migration of peoples, industrial development and the growth of international trade, together with the improvement of farm machinery, resulted in the extension of agricultural industries and gave a greater impetus to plant breeding. This activity was manifested first in Europe and later, particularly in the United States Department of Agriculture and the state experiment stations, in America. Naturally the efforts at improvement were concentrated in the main on the crop plants pro- ducing the raw materials of importance in the world's markets, such as wheat and other small grains, sugar beets, corn, cotton, forage plants, the apple and other fruits. The methods employed were those which had been used in the past for the most part, but they were systematized and combined for more effective utilization. These methods may be classified under the following heads: 1. Mass selection. 2. Line selection antl progeny test. 3. Hybridization followed by direct utilization or selection and fixation of new varieties. 4. Clonal selection. Mass Selection. — The method of mass selection consists simply in picking out choice plants from the main crop and sowing the seed from them 671 masse. It has long been used, especially in improving small grains, but it has also been used with many other crops. With this method it has usually been found necessary continually to repeat the 292 GENETICS IN RELATION TO AGRICULTURE selection of best plants in order to maintain the improvement already gained. One of the earliest breeders to use this method was Andre Leveque de Vilmorin, who began selecting carrots about 1830. Soon thereafter selection of sugar beets for seed production was begun in France and Germany, first according to form of the root alone, but later according to specific gravity and actual analyses of sugar content. Mass selection later became the principal method of improving small grains in Germany, and it has been known as the German method of "broad breeding." The earliest prominent breeder of small grains was W. Rimpau, who began his work with rye in 1867 and developed the famous Schlanstedt variety. Later he worked with wheat extensively, first by mass selection and, more recently, by hybridization of varieties and subspecies. Although there have been scores of successful breeders of each of the important small grains in Germany, Rimpau was the first to engage in this work on a large scale. Mass selection in maize was begun as early as 1825, when J. L. Leaming, of Ohio, began the selection of best ears from his field for seed corn. By repeating this process he soon developed a superior strain that came to be known as the Leaming variety. The same simple method was employed in originating Ried Yellow Dent (1847), Morley Prolific (1876), and Boone County White (1885). The famous Illinois corn-breeding experiments, which will be described in later chapters, were begun in 1896 by Cyril G. Hopkins, then Professor of Agronomy in the University of Illinois. Among the other investigators who have participated in this undertaking are East, Shamel and L. H. Smith. The general result of the project has been the most convincing proof of the efficacy and practicability of mass selection in corn, not only for the chemical and physical characters of the grains but for other characters of the corn plant as well. The improvement of cotton by mass selection has doubtless been practised for centuries. Authentic records of the earlier methods used in foreign countries are scarce, but the characteristic variability in length of fiber, combined with the very practical value of increasing the average length, must have appealed to growers, at least in the more prog- ressive cotton growing regions of the world. In the South Carolina islands according to Webber the sea island types of cotton have been developed by consistent mass selection for early maturity, increased length of lint, and greater productiveness from a West Indian perennial type which was originally unsuited to conditions under which its derivatives •are now grown so successfully. Mass selection in cotton has been resorted to also in the campaign against various plant diseases, particularly cotton wilt, and for early maturity to avoid the ravages of the boll weevil. INTRODUCTION 293 Line Selection and Progeny Test. — Turning now to the .second of the four general methods, we find that the progeny test of individual plants was first used by Le Contour and Shirreff . But it was Louis de Vilmorin who first gave special attention to the value of the progeny test (1856) and, contemporaneously with Hallet, practised the selection of single plants, i.e., of pure lines in wheat, oats and barley, followed by separate tests of their progeny. This method was first used in America by Willet M. Hays who began the improvement of small grains at the Minnesota Experiment Station in 1888. Convinced by the results of extensive variety tests that systematic breeding would be required in order to secure a marked increase in yield of first class wheat. Hays devised the centgener method of grain breeding, which, briefly, consists of planting about 100 seeds from each selected plant in trial plots; the more promising centgeners being selected for testing on a larger scale. Hays' work re- sulted in the isolation in 1892 of two plants whose progeny within a decade were grown on thousands of acres. Although many new strains were secured, the rigid tests of several consecutive years in which the most promising strains were compared with each other and with the best commercial varieties, resulted in securing but few really superior varieties. However, these made possible an increased production of wheat through- out the northern states and in Canada. The Swedish Seed Association was organized in 1886 and established an experiment station at Svalof. During the first 5 or 6 years only mass selection was practised, but soon after Hjalmar Nilsson became director in 1891 the ''Vilmorin Method" was introduced. At Svalof it came to be known as the "System of Pedigree" or ''Separate Culture." Nilsson was led to adopt this system as the method for originating new varieties by the accidental discovery that the only wheat plots that were entirely uniform were grown from single plant selections. The new varieties produced at Svalof are now grown throughout the agricul- tural portion of Sweden. This station is also engaged in the systematic improvement of peas, clovers, grasses and potatoes. All this work is based on mass and line selection followed by field tests and distribution. The first application of the pure-line conception to a naturally cross- fertilized plant was made by Shull and by East working independently with corn. By guarding and self-pollinating individual plants for suc- cessive generations, a number of morphologically distinct strains were isolated, thus proving that the original population was a mixture of biotypes. These same methods, however, had been employed for a number of years by Webber, Hartley, and probably others in working with corn, cotton, and other naturally cross-fertilized plants. In recent years the plant-row test has been used for the improvement of old strains or production of new ones. In Germany, von Lochow in 1894 294 GENETICS IN RELATION TO AGRICULTURE adopted a modified form of line selection in the improvement of rye, which is also naturally cross-fertilized. Timothy breeding was undertaken by the New York (Cornell) Experi- ment Station, under the direction of Hunt, Gilmore and others in 1903. To begin with, samples of seed were secured from 22 states and 11 foreign countries. Although it had been long cultivated in certain parts of Europe, there were no distinct varieties of this species of grass {Phleum pratense) because it is normally cross-fertilized. Many interesting variations were found among the plants grown from the various samples, some of them being of great commercial value. After several years of experimental work 17 new sorts were selected as most promising. These had been increased vegetatively by division and subjected to progeny tests with both cross-pollinated and self-pollinated seed. In two years tests the 17 selections gave an average increased yield of 36% per cent, above ordinary timothy. If such an increase in production of timothy could be extended throughout the country, it would, according to Webber, add over .S90,000,000 to the value of the annual hay crop. Hybridization. — The third general method of plant breeding in the light of genetical science holds great promise of future possibilities. In spite of Knight's early demonstration of the value of varietal crosses in breeding, this method did not come into general use until the latter part of the nineteenth century. According to Darbishire, another English horticulturist, John Goss, made some of the identical crosses used by Mendel, and noted the phenomena of dominance in Fi and recombination in F2, but failed to grasp the significance of the facts he observed. Accord- ing to Munson, it was the horticulturist, A. J. Downing, who in 1836 first called the attention of American breeders to the possibilities in hybridization. After his success with strawberries, Hovey, in 1844, definitely championed the cause. The achievements of Hovey, Downing and others soon led to the general adoption of cross-fertilization as a method of breeding. In their efforts to secure varieties having certain combinations of desirable characters, the crossing of varieties of small grains was employed to advantage by Rimpan, Blount, Pringle, Hays, Nilsson and others in later years. The remarkable Marquis wheat which has proved so valuable in the northern wheat regions is a hybrid according to Carleton which was probably made by A. P. Saunders at the Agassiz (British Columbia) Experiment Farm in 1892. The applica- tion of this method in the production of disease resistant commercial strains has been attempted. R. H. Biffin began his study of wheat breed- ing in 1909 in the service of the National Association of British and Irish Millers. The demand was for a beardless, rust-resistant variety of high yielding power and good milling quality. Not being able to discover any single variety which combined all these characters. Biffin attacked the problem from the Mendelian standpoint and has attempted to secure the INTRODUCTION 295 desired combinations through the liyl)ri(Hzation of a low (luality, rust-re- sistant form with a variety very susceptible to rust but whose characters are otherwise superior. Biffin found that susceptibility to yellow rust (Puccinia glumarum) is dominant to rust resistance, in the cross between Rivet and Red King, but that resistant forms appeared in the F^ genera- tion which bred true for resistance. This discovery marks a definite forward step in the breeding of disease resistant plants. However, the problems of disease resistance are complicated by the variability of the parasitic organisms involved. Hybridization of maize was begun as early as 1878 at the Michigan experiment station and was taken up from time to time at certain other stations. In 1900 the U. S. Department of Agriculture began a large series of experiments in crossing corn, using "all types obtainable." This work has resulted in the distribution for trial of many promising selections. Following the striking experiments of East and Shull in crossing strains of corn that had been inbred for several generations, many experiment stations began the crossing of varieties and strains for increased production as well as for new combinations of characters. With cotton, the recent work of Balls in Egypt has furnished a basis for the pedigree and hybridization method of breeding. Although cotton is self-fertilized to a large degree, yet it is visited by insects during the early morning hours so that there is always a certain amount of natural crossing. It is very susceptible to environmental effects and its chro- mosome number is large (haploid number 20). These conditions make improvement by crossing a difficult matter. Cook noted the fact that parent characters are sometimes intensified in the Fi in cotton and recommended the use of Fi hybrid seed of proved crosses as a means of enhancing the quality of the lint. He also suggested a practic- able and economical method of producing and utilizing such hybrid seed. Apple breeding by crossing varieties was begun by Knight but this method has not been used extensively in Europe. In America the cross- breeding of apple varieties probably was begun by Charles Arnold of Ontario, Canada, about the middle of the last century. Other early hybridizers who worked with varieties of the common apple, Pyrus malus, were F. P. Sharp of New Brunswick, who began crossbreeding in 1869 and C. G. Patten of Iowa, who commenced somewhat later, but who has worked continuously with apples and pears since 1879. In this connection recognition is due Peter M. Gideon and the host he represents, who have produced new varieties of apples b}' raising seedlings and selecting the best. Most of the new sorts obtained in this way are of hybrid parentage. More recently important work on variety crossing of apples has been done by Macoun in Canada, Hedrick in New York, and Evans in Missouri, 29G GENETICS IN RELATION TO AGRICULTURE The composite crossing of three or more varieties in an attempt to effect a desired combination has been used successfully in small grains, as well as in many flowers. Referring to grains alone, William Farrer of Australia, A. N. Jones of the United States, and John Garton of England have used this method successfully. In the opinion of Carleton, Farrer leads all breeders in the production of hybrids that have come into practical use. He continually practised composite crossing, as many as six different varieties or subspecies entering into the ancestry of many of his new sorts, some of which are of superior production or milling quality as well as being disease resistant. Interspecific hybrids have frequently been produced by breeders seeking some definite goal, occasionally with striking success, especially among fruits. Even intergeneric hybrids have been reported, but the known cases, with the exception of orchids, are of slight importance to agriculture. For example, van der Stok secured a fertile hybrid between corn and teosinte in the hope that some of the hybrids would bear good sized ears and be resistant to chlorosis, a hope which was not, however, realized. A few similar cases are known, particularly among cereals, but very little use has been founcF for them. In fact, utilization of wide crosses is rather definitely restricted to direct employment of the Fi in cases where conditions of seed production are particularly favorable for producing large quantities of hybrid seed or where the hybrid may be propagated by clonal multiplication. Alfalfa culture appears to be capable of still further extension through crossing of species. According to Fruwirth hybrids between common alfalfa, Medicago saliva, and M. falcata are easily produced and occur abundantly wherever plants of the two species grow near each other, the crossing being effected by insects, especially bees. This was reported in 1877 by Urban. These hybrid forms are known as M. 7nedia Pers. {M. varia Martyn., M. versicolor Ser.). Seeds of these hybrids produce forms that can be considered M. 7nedia, and, while flower color and pod form are inconstant, the plants bear more seed and grow more luxuriantly than M. falcata and adapt themselves to varied soil conditions. West- gate has found good evidence that the well known hardy Grimm alfalfa originated as a natural hybrid between these species, and that it was not a product of acclimatization. Piper in 1908 called attention to the desir- ability of securing hybrids between M. sativa and the yellow-flowered Siberian species, and Hansen has recently determined the practicability of producing such hybrids on a large scale by mixed field plantings. Experts of the Bureau of Plant Industry of the U. S. Department of Agriculture have made an enormous number of attempts to cross different species of Medicago but utterly without success except in the case of falcata and satim. Selfed mtiva and especially media, {falcata x sativa) INTRODUCTION 297 give many aberrant forms. A very common one under greenhouse conditions is a form with very short internodes and very small leaves. This is presumed to be the form which Southworth mistakenly reported as a hybrid between M. sativa and M. Inpulina. In various fruits and in many flowers the crossing of species has yielded many valuable varieties. Some cases among flow(M's will be discussed in the following chapter. Among tree fruits the next hybridi- zers of species after Bull were the men who undertook to combine the hard}^ character of the Russian apples, which had been introduced during the 80's. Dr. William Saunders, then Director of the Dominion Experi- mental Farms, began this work in 1894. Similar work, with apples, cherries, plums, etc., has been carried on very extensively, and already with important results, by N. E. Hansen of the South Dakota Experi- ment Station. The production of a list of peach varieties adapted to the Gulf Coast States was the work of H. H. Hume, then of the Florida Experiment Station, and of P. J. Berckmans in Georgia. This was accomplished largely through the hybridization of the Chinese Saucer or Peen-to peach, Amygdalus platycarpa, with commercial varieties of the common peach, Amygdalus persica. The work of Webber and Swingle with crosses between various species of Citrus has received inter- national recognition, not only because of the results secured but on account of the possibilities in the improvement of citrous fruits which it revealed. The production of aphis-resistant plums among hybrids of distinct species, as reported by Beach and Maney, exemplifies an impor- tant line of attack in breeding disease-resistant plants. No small part of Luther Burbank's fame is due to his success in crossing species. Among the many interspecific hybrids which he pro- duced should be mentioned plumcots (hybrids between plums and apri- cots), the Royal walnut {Juglans Calif ornica X J. nigra), the Primus and Phenomenal berries (hybrids between species of Rubus) , many valu- able plums and a host of flowering plants. In his work with plums, as well as in the production of certain flower novelties, Burbank practised composite hybridization. An illustration taken from de Vries' account of Burbank's work, is the pedigree of the Alhambra plum, shown in Fig. 119. An equally if not more important phase of Burbank's work is his discovery of novelties and his perfection of the same by means of selection. His method is hardly to be classified as mass selection, nor is it line selec- tion in the strict sense. An important feature has been the use of very large numbers of seedlings either of introduced species, commercial varieties, or his own hybrids. It is by the use of his unusual power of observation, which Wickson thinks amounts to a gift of intuition, in choosing say a dozen seedlings from as many thousand, that this one man 298 GENETICS IN RELATION TO AGRICULTURE has accomplished so much. His methods of hybridization, also, have been such as to economize time rather than insure certainty as to ancestry d. Alhambra Nigra [ Americana [c < Triflora Simoni French Prune [ Pissardi [a I Kelsey Fig. 119. — Ancestry of the Alhambra plum. His aim has always been the tangible result rather than advancement of scientific knowledge. Clonal Selection. — Under the term clonal selection is included all methods of plant improvement based upon the utilization of asexual means of multiplication, whether by selecting the most favorable clones from a mixed population, or by selecting and propagating favorable varia- tions within clones. In potatoes many commercial varieties are definitely known to be mixtures of different clones, and improvement may be effected by simply selecting those which are most productive and most desirable from a market standpoint. A unique instance of clonal selection is that followed in Oklahoma and other regions along the north- ern limits of the range of Bermuda grass. There the cold winters kill off the less hardy strains ; those that remain are propagated by distribu- ting sod. In alfalfa many improved strains have been produced by the selection and multiplication of superior individuals. This work has been carried on by the Bureau of Plant Industry of the U. S. Department of Agriculture and various stations, especially those in South Dakota, Kansas and Arizona. The propagation of improved strains by means of cuttings is of great practical value, and Hansen recommends the use of tobacco planting machines for the setting of rooted alfalfa cuttings. A phase of clonal selection which has recently come into prominence is hud selection, although the occurrence of bud variations, particularly of bud sports, has long been a matter of common knowledge. Munson (1906) seems to have been the first to call attention definitely to the pos- sibilities in fruit improvement by selection of buds from superior indi- viduals or vegetative parts, although Bailey had on several occasions previously pointed out that varieties sometimes originated from buds. During the past ten years many practical experiments in bud selection have been conducted, but with diverse results. INTRODUCTION 299 The aim of the foregoing review has been to present the more promi- nent historical examples of the four general methods of plant breeding. Further details can be obtained from the authors cited and from Fru- vvirth's Die Ziichtung der landwirtschajtlichen Kultiirpflanzen (The Breed- ing of Agricultural Crop Plants). This useful work, consisting of five volumes, is partly in its second and third revised editions and is the most complete and thorough treatise on plant-breeding methods. Organization of Plant-breeding Work. — Growing appreciation of the importance of plant improvement to agriculture has led to organized effort along certain lines, some of which are discussed briefly below. Seed and Plant Introduction. — The first teacher of plant breeding in America was also her first agricultural explorer. In 1882 Budd went to Europe and Asiatic Russia for the purpose of studying horticultural problems. He was accompanied by Thero Gibbs of Canada, and the expedition was financed by the Iowa State Legislature and the Canadian Government. As a result of this exploration many hardy shrubs and trees were introduced into America. The Russian cherries and apples were of especial importance as they have been used, notably by Saunders and Hansen, in the production of new varieties, which are sufficiently hardy to resist the cold winters of the northwest portion of the great interior plain, Bailey, in 1894, called attention to the similarity in climates and floras of eastern America and eastern Asia and emphasized the "abundant reason for looking toward oriental Asia for further acquisitions, either in other species or in novel varieties." His wise foresight in this matter has received repeated verification in the numerous valuable introductions of Wilson and of Meyer. About this time the U.S. Department of Agriculture began to give serious attention to the intro- duction of seeds and plants from foreign countries under the supervision of Galloway. A few years later this important work was put in charge of Fairchild who has organized the present efficient system of agricultural exploration, seed and plant introduction, trial gardens and distribution of promising material. Collections of Plant-breeding Material. — The importance of bringing together a working collection of all available species and varieties within a group in which improvement is desired has been increasingly appreci- ated since the work of Vilmorin. The importance of local variety trials has long been realized and the collections of cultivated varieties at various experiment stations have proven very useful for purposes of selection of better adapted sorts as well as for some work in hybridization. Well- known examples are the sweet pea, peony, and chrysanthemum collec- tions at Cornell University and the collections of apples, plums, and grapes at the Geneva, N. Y., Experiment Station. But, on account of the time and expense involved in the work of hybridization, it is highly im- 300 GENETICS IN RELATION TO AGRICULTURE portant that the most promising forms which exist be secured, if possible, at the beginning of such projects. Some of the older collections of living plants, such as Arnold Arboretum and the New York, Brooklyn, and Missouri botanical gardens, as well as the Government Office of Seed and Plant Introduction, have given valuable assistance in supplying new and rare material to breeders. The transportation of pollen has also been resorted to, especially by the U. S. Department of Agriculture in its breeding of Citrus and it is known that, with proper precautions, some kinds of pollen can be sent by mail half-way around the world and still be viable. However, certain much desired crosses can be secured only after repeated efforts and the trial of various methods. Moreover, the response of introduced forms to local conditions is a most important con- sideration. All too often a supposedly promising new plant has proven entirely unfit for certain localities. These considerations are leading to the establishment of large working collections of our more important semi-permanent crop plants, especially the tree fruits. For example, the University of California Citrus Experiment Station is accumulating a collection which will include all the known species and varieties of Citrus and allied genera which will endure local conditions. Research on Plant Groups. — A breeding program such as that con- templated by the institution just mentioned involves the necessity of ex- tensive botanical investigations. In this particular instance it is fortu- nate that extensive work has already been accomplished by the U. S. Department of Agriculture since already a large amount of data on the botanical relationships and geographical distribution of the members of the Citrus group has been collected. As a result of these studies and explorations several new and very promising forms have been introduced and have already been utilized in breeding experiments by the Bureau of Plant Industry. Similar investigations of the genus Prunus are also under way by the Department. The recent explorations of date growing countries and studies on the varieties of dates is another illustration of the sort of work that is needed, not only among fruits in general but in the field crops as well. Organization of Plant Breeders. — In December, 1903, the American Breeders Association was organized under the auspices of the American Association of Agricultural Colleges and Experiment Stations. During the first seven years of its existence the publications of this organization were restricted to the annual reports of its meetings. These reports contain the papers which were presented at the meetings either in full or by title. In 1910 the Association undertook the publication of a quar- terly journal, the American Breeders Magazine, and discontinued the publication of annual reports. This magazine in January, 1914, became the Journal of Heredity, which is published monthly. At the same time INTRODUCTION 301 the American Breeders Association changed its name to the American Genetic Association. With its policy of unrestricted membership from the beginning this organization has done great service in fostering the common interests of geneticists and practical breeders. There are state associations of plant breeders in New York, Wisconsin, Minnesota, Illinois, Pennsylvania, Ohio, Nebraska, and Georgia. In certain other states the agricultural and horticultural societies have fostered plant- breeding work to a greater or less extent. The meetings held with their addresses and discussions, the exhibits of new introductions and occasional tiemonstrations in plant improvement by the experiment station or other agencies, have aided in bringing to the seed growers and farmers of the United States the knowledge of superior plants and their practical value. The Canadian Seed Growers' Association has fulfilled a similar mission. Summary. — Starting with the sporadic efforts of a century or more ago to find some better varieties of fruits and grains, there has been a gradual broadening of the great movement to increase agricultural output and raise the quality of raw materials by means of plant improve- ment. Throughout the later stages of this development scientific knowledge has become increasingly important until now the specialist on a particular crop plant may invoke the aid of every branch of agri- cultural science in selecting his material for breeding operations, making the desired crosses and selecting the progeny. All this has been done without much, if any, definite knowledge concerning the heredity of the plant in question. Within a decade the science of genetics has developed to a stage where it is capable not only of furnishing a rational explana- tion for the phenomena of variation and heredity which in the past seemed obscure and contradictory, but also of guiding the breeder who will familiarize himself with the established principles of the science, so that he may reach his goal with greater speed and economy. It is the purpose of the following chapters to set forth these principles in as clear and practical a manner as possible. It will be assumed, of course, that the reader is familiar with the fundamental treatment of the preceding chapters. CHAPTER XVI ON VARIETIES IN PLANTS The multiplicity and diversity of the varieties of cultivated plants never fail to impress the thoughtful observer. The cereals, fiber plants, legumes, root crops, and tree fruits which comprise most of the important agricul- tural crop plants include some 30 species. It is safe to assume that within this small group of species over 5000 distinct varieties are known at pres- ent. Of rice alone there are thousands of varieties in cultivation. Among flowering plants we find the same diversity. The rose, lily, chrysanthe- mum, violet, carnation, sweet pea, dahlia, gladiolus, tulip, and hyacinth of our gardens and greenhouses represent not more than 200 species, while of roses alone as many as 1000 named varieties are now listed in European catalogues. In general the longer and more widely culti- vated species contain the larger groups of varieties, partly because of the greater opportunity for their discover}^ and partly because these species have been subjected to conditions most favorable for the pro- duction of varieties. Before attempting to discuss the conditions or operations that lead to the production of new varieties it is necessary to enquire into the natural processes by which varieties have been produced. The Origin of Domestic Varieties of Plants. — Agriculturists have made use of three general methods in creating new varieties of cultivated plants, viz.: (1) the utilization of mutations or sports; (2) the employ- ment of hybridization and selection methods; and (3) the utilization of clonal diversity. The utilization of mutations should be interpreted to include not only the discovery and multiplication of mutant forms, but also the recombination of mutant characters in new varieties by hybridization. We include selection in the same category with hybrid- ization, because according to the hypothesis which we have championed throughout this text, its effectiveness usually depends upon the existence of germinal diversity such as follows hybridization. In certain cases, of course, selection methods have depended for success upon the utiliza- tion of mutations having minor character effects. The origin of varieties by these three different methods may be illustrated by considering in some detail the horticultural history of certain plants. Since the ancestors of most of our crop plants are now extinct, we may turn for this purpose to some of the more recently domesticated species, the histories of which are known more precisely. 302 ON VARIETIES IN PLANTS 303 Origin of Sweet Pea Varieties. — The sweet pea, Lathyrus odoratus , provides an excellent illustration of the origin of varieties by the utili- zation of mutations. Its history as a horticultural plant is known from the beginning and has been thoroughly reviewed in publications of the Cornell Station. The sweet pea was introduced into Holland and England from Sicily via Italy in 1699, and was first illustrated in a description published in 1700. The drawing is reproduced in Fig. 120. It will be noted that in habit it was similar to the cultivated sweet peas of the present day and the height to which it would climb was "6 or Fig. 120. — -Commeliii's drawing of the sweet pea in Hort-Medici Amstelodamensis, 1700. (After Bcal.) 7 feet," but the flower stems were short and bore only two flowers, while the flowers themselves were relatively small, with erect or reflexed standard and conspicuous, depressed wings. In color the standard was reddish purple and the wings light bluish purple. From this modest beginning there have been developed several distinct types of plant and flower forms and a list of named varieties, even within the most highly developed type of flower (the Spencer or waved form), which includes over 500 entirely distinct colors, tints, shades, and combinations. By far the greatest amount of this work has been accomplishetl during the past 50 years, during which period hybridization has been used 304 GENETICS IN RELATION TO AGRICULTURE extensively in creating improved varieties. But before hybridization was resorted to there were a dozen distinct color varieties which had arisen by mutation. Besides color mutations there have occurred spontaneous changes in flower form, flower size, and number of flowers on the stem, in stature and habit of the plant and in season of bloom, some of which are described below. Flower Color in Sweet Peas. — The chronology and probable ancestry of the color varieties of the sweet pea which appeared during the first 180 years of its horticultural history are shown in condensed form in Table XLV. This summary is based upon Beal's excellent historical review, from which citations to original sources have been obtained. Apparently the course of events was about as follows. From the original type form there appeared first white mutations (Plate III, 3.) If we call the simple flower-color factor complex CRB, in which C and R are complementary factors producing red, and B an epistatic factor which modifies that color to purple, then these mutations apparently depended upon a change in either C or Rto the recessive, white condition. The Painted Lady variety, red instead of purple, shown in Plate III, 2, appeared very soon after this, apparently as an independent mutation in the factor B from purple. By the close of the eighteenth century two other color types, black and scarlet, had been added to the list. The wild form and Painted Lady are bicolors, that is, the wings are lighter in color than the standard. The new color type scarlet (Plate III, 5,) apparently resulted from a recessive factor mutation conditioning the development of full color in the wings along with a certain intensification of color in the standard. Black (Plate III, 6,) was probably also merely a factor mutation for more intense pigmentation from the wild color type. Early in the eighteenth century a "blue" form, var. caenileus, was described in the trade, but its genetic relationships have not been clearly defined. Plate III, 8, which is taken to represent it has not been copied from a particular variety as was done in the case of the other types. Further additions shortly followed in the form of a "striped" variety, and of a "yellow" variety. The latter (Plate III, 4) unquestionably originated as a factor mutation from white, the former may have arisen as a factor mutation in purple. Plants with primrose yellow flowers have since been observed a number of times in white cultures, but never in red ones. This practically closes the account of the origin of color mutations up to the year 1880, after which time hyl^ridization was resorted to extensively in the creation of new varieties. Form and Size in Sweet Peas. — The changes in form and size of flower in the sweet pea have been no less striking than those in color, and they have been responsible for a large portion of the popularity which it enjoys. Today one can scarcely recognize in the favorite varieties of Plate III. — Oldest Varieties of the Sweet Pea. 1. The original wild form. 2. Old Painted Lady. 3. White. 4. Yellow. 5. New Painted Lady or "Scarlet." 6. "Black." 7. "Blue Edged" ( = purple picotee?"). 8. "Blue." Numbers 4-8 >ie reconstructions based on modern varieties because the original varieties listed under these names cannot be identified with absolute certainty. But it is highly pi obablc that they were very similar to the types shown above and that they originated by mutation in the order indicated. ON VARIETIES IN PLANTS 30'i Table XLV. — Okigin of tue Earlier Color Varieties of the Sweet Pea^ 1700 1718 white- (cRBE or CrBE) 1 73 1 white purple and blue — CRBE 1793 180G 1817 white white 1824 yellow white • (= primrose?) 1840 white 1845-49 1850 1860 yellow ( = primrose?) 1865 Painted Lady ( = pink and white) CRbE Painted Lady scarlet- I ( = deep rose) ( = dark violet) Old Painted New Painted dark purple Lady Lady or Scarlet Old Painted Lady New Painted Lady Scarlet In- vincible dark purple and deep violet New Large Purple New Large Dark Purple striped (= purple with brown, lavender or white?) New Striped blue dark bluish purple and pale blue Blue Edged Blue Edged 1870 List f yellow, white Painted Lady, Scarlet In- black, Imper- purple {Crown Prin- vincible, ial Purple, striped cess of Prussia scarlet, scar- black with with white, (pink and rose let striped light blue, pink). with white. yellow, white Painted Lady, New Painted Black, Black purple striped Blue Edged^ 1880 List Crown Prin- Lady, scarlet, Invincible, with white, Butterfly^ cess of Prus- Scarlet Invin- black with Invincible Captain sia, Fairy cible, scarlet light blue. Striped Violet Clarke* Queen ( = striped with Large dark Queen. Hetero- pink on white. purple. Im- sperma^ white). Queen perial purple, ( = light pink Purple In- and pink pur- vincible. ple). 1 In each case the color of the standard or banner is given first and of the wings second; the descrip- tive terms and variety names are identical with those in the original descriptions. - Described by Bailey and Wyman as purple-lilac in color ( = purple picotee). ' Quite similar to Blue Edged according to Beal ( = purple picotee). 4 = "white merging into pink and purple, wings white with purplish cast, wings edged with blue" ( = purple picotee) . 6 No description available; mottled seeds? 20 306 GENETICS IN RELATION TO AGRICULTURE the garden traces of the early pecuhar form of the flower portrayed in Plate III. In the original form the standard was erect, narrow at the base, notched at the top, and reflexed or slightly rolled at the sides. From it have been derived three distinct flower types; the grandiflora, the hooded, and the popular waved Spencer forms. The origin of the first two named is in some doubt. The hooded character was found in some of the earlier varieties. It was sometimes associated with notches in the sides as in the Butterfly (Fig. 121), and this character is found also Fig. 121. — Forms of sweet pea flowers — the standard or banner. Open or grandiflora form (upper row left to right) — Alba Magnifica, Shasta, Golden Rose. Hooded form (middle row) — Butterfly, Admiration, Dorothy Eckford. Waved form (lower row) — Elsie Herbert, Apple Blossom Spencer, White Spencer. (From Beal.) in some of the present day favorites. Bateson reports that hooded is recessive to grandiflora or erect type of standard. Some of the earliest varieties of improved grandiflora form were Queen of England (1888), Blanche Ferry (1889) and Alba Magnifica (1891). The waved or Spencer form is of more recent origin, and authorities are agreed that it arose as a "sport" from a beautiful, pink, hooded variety, Prima Donna. The pronounced waviness of standard and wings which characterizes this type had not appeared before in sweet peas. The two upper series in Fig. 121 indicate the more recent progress in enlarging flower size. Alba Magnifica and Butterfly were great acqui- ON VARIETIES IN PLANTS 307 sitions in their day iiiid were doubtless considerably larger than the oldest varieties. The first definite reference to size is found in New Large Purple, listed in 1845. As this occurs in the darkest color group and 15 years before the hybrid origin of a new variety, ]Mue Edged, was even suggested, it probably represents a factor mutation. That such nuitations ac^tually occurred in the sweet pea is proved by the fact that Countess Spencer and (iladys Unwin were both decidedly larger than Prima Donna fiorn the very first. The same is true as regards number of flowers in the cluster. Prima Donna, according to Beal's description, bore two or three, usually three, flowers on a stalk, while Countess Spencer has three to four flowers in a cluster. Many of the recent Spencer Fig. 122. — On the left, Suapdnigon sweet peas. On the riglit, double sweet pea, White Wonder. {From Beal.) varieties bear almost uniformly four-flowered clusters. The original form and earliest varieties had two flowers in the cluster. The oldest varieties definitely known to bear more than two flowers on a stalk are Invincible Scarlet (1865) and Crown Princess of Prussia (1868). As these antedate the era of hybridization it is probable that the increased number arose by mutation. Novelty forms have also arisen from time to time. In double sweet peas there are two standards instead of one. In some varieties this character has been fixed by selection so that most of the flowers come double. It gives the eff"ect of increased size (Fig. 122), In the snap- dragon type of flower (Fig. 122) the standard is folded around the wings. It is recessive to en^ct standard and gives a simple Mendelian ratio of 3 erect to 1 snapdragon in Fg. 308 GENETICS IN RELATION TO AGRICULTURE Habit in Sweet Peas. — There are several distinct types of plant in the sweet pea the origin of which may be definitely ascribed to mutation. The first Cupid plant (Fig. 123a) appeared among plants of the tall, '^^i'SS^WV. ^^^s^m -^i Fig. 123. — a, Cupid or prostrate, dwarf sweet pea; b, bush or erect, tall form; c, Cupid X bush Fi, the ordinary tall form (folded over in order to photograph). (From Bateson.) white-flowered variety, Emily Henderson, in 1893. The growers, C. C. Morse & Co. of San Francisco, raised seven acres of the new variety in 1895 and every plant was true to type. This mutation has since oc- FiG. 124. — Dwarf or Cupid sweet peas. I, ordinary or prostrate Cupid; II, erect Cupid, the F2 double recessive from bush X Cupid. (From Bateson.) curred in a number of widely separated localities. The bush type also originated as a mutation from the tall form. The investigations of the factor relations of bush and Cupid sweet peas have been described in ON VARIETIES IN PLANTS 309 a previous chapter. Scuui-dwarf, early-flowering sports have appeared even more frequently than those of the Cupid type. They have been made the basis of the winter-flowering types of sweet peas. Ordinary sweet peas pass into a semi-dormant condition for a time after germination, growing very slowly until sideshoots have been developed. The winter- flowering sorts, however, promptly send up a central axis which begins blossoming as soon as it has attained a height of from two to four feet. The Blanche Ferry group of varieties apparently had their inception in a mutation of this sort which a woman in northern New York noticed among some plants of the Old Painted Lady. She selected them for about twenty-five years after which they passed into the hands of a seedsman. From this stock a series of early flowering mutations have arisen in the order shown below. Black-seeded varieties are indicated by (6) and white-seeded ones by (iv). Old Painted Lady (h) I Bright-flowered sport {b) (30 years later) Blanche Ferry (/>) Extra Early Blanche Ferry (b) Emilv Henderson (white, w) I ■ I . Earliest of all (b) Mont Blanc (early white, w) I . , I . . Extreme Early Earliest of all (b) Earliest Sunbeams (primrose, iv ) I Earliest White (6) Fig. 125.— New varieties of sweet peas which originated by mutation among the progeny of Old Painted Lady. Hybridization and Selection in Sweet Peas. — The era of extensive hybridization in sweet peas dates from about the year 1880, consequently we can say but little of definiteness after that time with respect to the origin of new factors in the sweet pea save in a few particularly favorable cases. Laxton's Invincible Carmine was the earliest recorded new variety which was produced by crossing, and its parents are reputed to have been Invincible Scarlet and Invincible Black. We can easily understand, therefore, how it originated, for it is apparently merely an improved form of Invincible Scarlet resulting from the inclusion of the factor for intense pigmentation of Invincible Black in the factor complex of Invincible Scar- let. Similarly by hybridization it has been found possible to establish families of varieties such as the Spencer, the hooded, the grandiflora, and the winter-flowering sorts. Hybridization has throughout been merely a means of fully utilizing germinal differences which have arisen by mutation. It is true that in most cases we cannot say just when the 310 GENETICS IN RELATION TO AGRICULTURE particular features of form, color, and habit have arisen but we know that there was only one original form, and fragments of the history (Beal and Hurst) are sufficiently clear to give us assurance in advanc- ing this explanation of the role of hybridization in the creation of varieties of sweet peas. There is no authentic instance of a variety having origi- nated from hybridization of the sweet pea proper, Lathyrus odoratus, with any other species of Lathyrus, consequently^ the possibility of such germinal diversity is precluded. Similarly in the case of selection for more obscure characters such as number of blossoms in the cluster, size of flower, and vigor of growth, apparently the things that have been utilized in cases of improvement are mutations and new combinations of mutant factors. Fig. 126. — Four types of rose: a, typical modern Hybrid Tea rose, b, typical Hybrid Perpetual rose; c, the Damask rose, which was popular in old gardens; d, the old single Rosa gallica. (Reproduced from The Garden Magazine by permission.) Creation of Varieties of the Rose. — No finer examples of the origin of horticultural varieties by means of hybridization could be found than the garden roses of today. The genus Rosa is widely distributed in the Northern Hemisphere and contains several hundred species of which, according to Wilson, twenty-six have been utilized in the production of our garden roses. But these twenty-six species fall into fifteen distinct groups, and in habitat they represent Asia, Europe, and North America. The most important group of modern roses are the Hybrid Teas for ON VARIETIES IN PLANTS 311 tliey include garden and forcing varieties which combine marvellous beauty of form and color with vigor and hardiness (Fig. 126a). Four or possibly five distinct species enter into the ancestry of the group, as shown by the following pedigree. The Hybrid Ferpetuals (Fig. 12G^) are of mixed ancestry, all being hybrids of the Damask Rose (Fig. 126c) crossed either with Hybrid Bourbon or Hybrid Chinese varieties. The hardy, disease-resistant Japanese species, Rosa rugosa and R. wichuriana have entered into the ancestry of some of the best modern roses. Thus, the American Pillar variety is a hybrid between a red Hybrid Perpetual crossed with a hybrid between R. wichuriana and R. setigera, the Prairie Rose of America. Again, the Silver Moon variety is a result . Teas derivatives of. . Rosa chinensis var. odoratissima Hybrid , Teas . . . Hybrid Perpetual si Hybrid Chinese Hybrid Bourbons \Rosa damascena (Damask Rose) Fig. — 127. Pedigree of the hybrid tea roses. Rosa gallica (French or Provence Rose). See Fig. 126f/. Rosa chinensis (Chinese Monthly or Bengal Rose). Rosa centifolia (Cabbage Rose). of crossing R. IcBvigata, the Cherokee Rose, with a hybrid between the Tea Rose, Devoniensis, and R. wichuriana. These examples will serve to illustrate the composite ancestry of our best roses. The practicability of this method of procuring new varieties has of course been enhanced by the possibility of vegetative propagation. Occasionally valuable varieties have arisen as bud mutations but these usually differ from the parent variety only in some definite character, like flower color or habit of growth. In passing it is of interest to note how extensively this method of variety creation has been used by horticulturists, particularly in species which are normally propagated by clonal multiplication. The hybrid varieties of the rhododendron rival in diversity and floral magnificence even those of the rose, and like them they have been derived from the mingling of a number of different species. But it is among the Rosaceae particularly that horticulturists have found the most favorable subjects for hybridization. It is necessary in this connection merely to mention such familiar examples as varieties of plums, apples, strawberries, and other rosaceous fruits in the production of many of which extensive hybridization has been employed. In seed plants, also, there are many 312 GENETICS IN RELATION TO AGRICULTURE examples of like improvement. Unquestionably the amateur plant breeder can find no more fascinating or productive line of activity than that of selecting and working with some particular group of species from this standpoint. Origin of Varieties in the Boston Fern. — In 1915 Benedict reported that he had accumulated about 40 different forms of the Boston Fern, all of which had originated so far as is known from bud sports. The following statements regarding the source of these new varieties are based on Benedict's account. The original Boston Fern arose as a bud mutation from the tropical species, Nephrolepis exaltata. It was first Fig. 128. — 1. The original Boston fern, Nephrolepsis exaltata bostoniensis; 2, the first bud sport from the Boston, A^. exaltata bostoniensis Piersoni; 3, the Pierson fern next pro- duced elegantissima; 4, N. compacta, a sport from elega^itissima. {Courtesy Brooklyn Botanic Garden.) recognized as different from exaltata by F. C. Becker of Boston, and in 1896 it was named N. exaltata var. bostoniensis. The typical form of the species and the first sport, bostoniensis, are large growing ferns with uni-pinnate leaves (Fig. 128, 1). In the remarkable series of bud muta- tions that have been derived from bostoniensis within two decades, the principal characters undergoing transformation are, first, form of pinna and hence form of frond; second, size of frond; third, form of frond con- sidered independently of pinna-form; fourth, color of foliage. The original sport from the Boston fern was bi-pinnate; i.e., each pinna was subdivided into little pinnae or pinnules (Fig. 128, 2). This form appeared about 1900 in the establishment of F. R. Pierson of Tarrytown on the Hudson, and was named Piersoni or Tarrytown fern. It did not produce satisfactory plants because only part of the fronds were bi-pinnate; the remainder resembled the original Boston variety. ON VARIEriES IN PLANTS 313 But Piersoni soon produced a tri-pinnate sport which was more regularly- divided. Its fronds were somewhat shorter and much broader at the base, thus making the plant more compact. It was named elegantis- sima (Fig. 128, 3). Although it was unstable like Piersoni, its uniformity was considerably improved by selection. Soon it produced a sport of quite similar characters except that it was more dwarf which was named compada (Fig. 128, 4). In both elegantissima and compacta there was variation from the tri-pinnate to the quadri-pinnate condition. The Pierson fern also gave rise to another interesting series of new forms which exhibited variation in two more characters. In the ele- FiG. 129. — The fronds of modern commercial varieties differ greatly from those of the original Boston fern. The varieties shown here are relatively stable, although they are all likely in turn to produce new sports some of which may prove valuable, a, viridissima; b, Millsii; c, muscosa; d, vcrona; e, tnagnifica; f, superbissima. (After Boshnakian.) gantissima series the color of the foliage is similar to that of the original Boston form, but in the new sport, which was named superbissima (Fig. 129/), the fronds are not only shorter and the pinnae three- or four- divided, but the foliage is of a deeper green color. Moreover, the fronds and separate pinnae are twisted so as to give the individual frond an irregular appearance although an entire plant appears fairly symmetrical. Although superbissima was unstable, producing uni-pinnate fronds occa- sionally, it soon produced a sport that is more compact in form and which proved to be more stable. This was named muscosa (Fig. 129, c). Other distinct uni-pinnate forms that have sprung as bud mutations either directly or indirectly from the Boston fern are the dwarfs, such as Scotti, Dwarf Boston, and Teddy Jr., and the vigorous, broad fronded variety, Roosevelii. There is no regularity in the production of larger and 314 GENETICS IN RELATION TO AGRICULTURE smaller forms. That is, a dwarf form may spring from a large form or from another dwarf form as shown in Fig. 130. Another distinct group Fi(i. 130. — Bud mutations in sports of the Boston fern. At the right ih) is the form, magnifica, a dwarf, asexual descendant of the variety, bostoniensis. The fern in the center (a) is a sport from this dwarf. It has a tendency to produce further sports and so could not be depended upon to breed true. At c is shown a small plant whose single frond resembles magnifica. At d is another sport that already displays instability in having two sorts of fronds. {After Boshnakian.) Fig. 131. — A series of pinnse illustrating progressive variation in division. 1, Var. bos- toniensis; 2, Piersoni; 3, Whitmani; 4, Goodi (or gracillima) ; 5, Magnifica; 6, Craigi; 7, Amerpohli. (Courtesy Brooklyn Botanic Garden.) contains the more delicate, open, lace-Uke forms, such as Millsii and verona (Fig. 1296, d). The latter has an advantage over several earlier ON VAlilETlEH IN PLANTS 315 varieties of this group in that its raehis is strong enough to support the fully developed frond. As Benedict has shown. the bud mutations occurring in these ferns are more commonly regressive (showing more resemblance to bostonien- sis than to their parent forms), Ijut progressive mutations are found from time to time. These progressive changes take place along three main lines, viz., increase in leaf division (see Fig. 131), increase in ruffling or crisping, and dwarfing (see Fig. 132) ; and any form which has not Fig. 132. — A series of fronds illustrating progressive variation in ruffling and dwarfing. 1, N. exaltata; 2, var. bostoniensis ; 3, Harrisi (or Roosevelti) ; 4, Wm. K. Harris (or new sport of Roosevelti) ; 5, Teddy Jr. ; 6-8, dwarf sports of Teddy Jr. ; 7, Randolphi. {Courtesy Brooklyn Botanic Garden.) reached the limits of possibility in variation along the first and last mentioned lines, may be expected to give rise to new forms showing further progressive variation in one or both of them. That these new varieties are produced by mutations in specific factors is indicated by the independence of character changes in series of suc- cessively produced forms that differ in several characters; for example, the appearance of dwarf uni-pinnate forms as sports of dwarf multi- pinnate forms. Various series derived from bostoniensis show progressive degrees of reduction in size of frond. When a dwarf tri- or quadri- pinnate plant throws a uni-pinnate sport the latter retains the dwarf size of its parent. Again the difference between Piersoni and superbissima, its sport, consists of the deeper color and twisted, irregular shape of the latter. When it in turn produced viridissima the new uni-pinnate 316 GENETICS IN RELATION TO AGRICULTURE form retained the other distinctive characters of its parent. Finally, as Boshnakian points out, similar sports have been secured among sexually produced progeny in other species of Nephrolepis. Thus it appears that these interesting and valuable ornamentals owe their origin to altera- tions in specific genetic factors, i.e., to factor mutations in vegetative reproduction. We have found that new varieties of cultivated plants may be arti- ficially produced in either of two ways, viz., by the discovery and pre- servation of mutations or by hybridization. Factor mutations occur in both sexually and asexually reproduced plants and frequently produce new forms of immediate economic value. Sometimes, however, the original mutation may be merely a starting point indicating the line along which selection must work. There is always the possibihty that subsequent mutations in the same direction, even though they be minute, will be taken advantage of by the breeder. In the creation of new varieties for special purposes hybridization must usually be employed. The success of breeders in combining the desirable qualities of several species in the best modern varieties of the rose suggests untold possi- bilities in this field of plant breeding. 1 CHAPTER XVII THE COMPOSITION OF PLANT POPULATIONS Before taking up in detail the various methods of plant breeding and considering their effectiveness it is well to enquire as to the nature of the populations with which we are required to deal. By a population in this connection we ordinarily mean a variety as that word is used in the trade, although populations as found in cultivation may be made up of mixtures of varieties. Usually, however, within an established variety, that is, a strain or race bred to a given type until it reproduces that type with a fair degree of accuracy, the variations are of minor consequence and not always readily detectable. But they may be due not only to modifications consequent upon slight differences surrounding the development of individuals in a population; they may also be germinal, that is, they may arise either from Mendelian recombination of germinal differences or by actual new germinal changes. We desire to know, therefore, what sorts of populations exhibit germinal diversity, what kinds of germinal diversity they exhibit, and how the germinal diversity may be related to other characteristics of the populations. Reproduction in Plants. — ^In seed plants the important factor which determines the character of the population is the kind of pollination which normally takes place. In the following classification most of our important agricultural crop plants are listed roughly with respect to this factor. A. Plants normally self-fertilized. (a) Flowers hermaphrodite, but the floral mechanism such as practi- cally to preclude cross-pollination. Examples: wheat, oats, barley, rice, beans, peas, and most of the other legumes. (6) Flowers hermaphrodite, but the floral mechanism favorable to a low percentage of cross-fertiHzation. Examples : cotton, tobacco, tomato, flax, and other plants having a similar floral structure. B. Plants normally cross-fertilized. (a) Flowers hermaphrodite, self-fertile, but with floral devices favor- able to cross-fertilization. Examples: rye, sugar beet. (6) Flowers hermaphrodite, but self-fertiUzation precluded on account of self -sterility of the plants. Example: sunflower, red clover. (c) Monoecious plants, self-fertile, but the floral mechanism such as to favor cross-fertiUzation. Examples: maize, watermelon, squash, pumpkin, cucumber, and cantaloupe. 317 318 GENETICS IN RELATION TO AGRICULTURE (d) Dioecious plants. Flowers of different sexes on different plants, thus insuring cross-fertilization. Examples: hemp, hops, asparagus and date palm. Another class having hermaphrodite and uni-sexual flowers on the same plant is termed polygamous. The sunflower might be classified here, because its marginal ray flowers are pistillate only. Certain species of Compositae have the marginal flowers pistillate, through complete sup- pression of the anthers as in the sunflower itself, and the disk flowers are hermaphrodite, but the pistil always aborts, so that in effect they are really monoecious plants. In some cases, however, they are known to be completely self-sterile, so that cross-fertilization must always take place in seed formation. The above classification requires numerous qualifications. For ex- ample, it has been our purpose to list under Class Aa those plants which are so generally self-fertilized that it is not necessary to protect them to insure self-fertilization, but there are some species and varieties among them which sometimes exhibit a significant amount of cross-fertilization. The cultivated varieties of wheat are very rarely cross-fertilized, but the wild wheat of Palestine has a floral mechanism especially designed for cross-fertilization. Some varieties of rice, also, are cross-fertilized often enough in mixed plantings to make it impossible to assume self-fertili- zation in a given selection. In peas and beans, perhaps, the proportion of crossing is greater than in the cereals mentioned above, and in some cases it is absolutely necessary to protect them from insect activities. Thus Pearl and Surface in breeding investigations with Yellow Eye beans found it necessary to enclose selected plants in large muslin cages in order to exclude bumble bees, which were found to be effective enough agents of cross-pollination in open fields to disturb results greatly. On the other hand, however, Pearl and Surface in extensive investigations in oat breeding report not a single case of natural crossing. Also Rimpau, who carried on extensive investigations with nineteen varieties of oats over a period of six years, observed only five cases of spontaneous hy- bridization. Furthermore in most of the commonly cultivated varieties of wheat, barley, and rice natural crossing is so rare a phenomenon as to be worthy of special note in any observed case. We recall also Johannsen's pure line investigations with Princess beans which would have been impossible had natural crossing occurred among them in any significant amount. Among plants having hermaphrodite flowers which are usually self- fertilized there is also vast difference in the relative proportions of self- and cross-fertilization. In cotton, Balls has found it necessary to allow for about 5 per cent, of natural crossing. In tobacco self-fertilization is the rule, but it is not sufficiently assured to obviate the necessity for THE COMPOSITION OF PLANT POPULATIONS 319 protection in gathering pure seed. Especially is this true in sub-tropical regions where humming })ir(ls are prevalent for they find tobacco flowers a splendid source of sustenance and unquestionably often effect cross- fertilization between plants. Moreover these remarks concerning tobacco, although they apply to the commercial varieties, do not indicate the true state of affairs in all species of Nicotiana, for a few species are completely self-sterile. Thus in N. alata grandiflora some individuals are actually completely self-sterile and others exhibit no bar whatever to seJf-fertiUzation. It is especially important, therefore, in dealing with plants in this class to determine these data for the particular species and varieties and the special conditions attending the experiments. Under Ba we have included rye in spite of general statements as to its self-fertility. This classification appears to be justifiable in view of reports of von Riimker and Leidner on results of inbreeding rye. The diflftculties in the self-fertilization of rye appear to be technical ones, rather than physiological, consequently reports as to its self-sterility must be in error. This is of interest in connection with the next following class which includes plants which are self-sterile. We have already mentioned the case of Nicotiana alata grandiflora in a given population of which both self-fertile and self-sterile individuals may be found. Other complications arise from contradictory reports as to self -sterility in some species belonging in these two groups. Thus there are reports that flowers on a given plant are sterile with their own pollen, but exhibit a certain degree of fertility when polhnated from some other flowers on the same plant. In effect such relations give results which are equivalent to self-fertility, but in some breeding operations it is important to know the exact relations, because it may be necessary to take advantage of them in special cases. It is probable that in general any difference which may be found in the fertilizing power of pollen derived from different flowers on a given plant are non-essential, and dependent upon some such factor as relative maturity of pollen with respect to the receptive period of the stigma. Among plants which are self-sterile are included a large number of the horticultural varieties which are normally propagated by means of clonal multiplication, but in which suitable pollination is necessary for fruit-setting or for the fullest abundance of fruit-setting. Orchard planting methods provide for this by mixing varieties which are known to act as efficient interpollinating agents. It is important to note that something more than a mere mixing of varieties is necessary; for the best results accurate knowledge should have been gained beforehand of the particular varieties which are most effective when planted together. Self-sterility in improved tree and bush fruits is a not unimportant con- sideration in practical horticultural operations. It is, also, of interest 320 GENETICS IN RELATION TO AGRICULTURE to note in passing that there is a possibihty in particular cases of dis- covering and overcoming the bars to self-fertiHty which are normally operative in such cases. Populations of Plants Normally Self-fertilized. — Continued self- fertilization in a population normally results in the automatic elimina- tion from it of all heterozygous individuals. The operation of this principle can be seen very clearly by considering the simplest case, a heterozygote for one pair of factors self-fertilized through a number of generations. Thus we see from Table XL VI that the general expression in this case for the percentage of heterozygotes after n generations of 1 inbreeding is 2"" If we set this value equal to 1 per cent., we get 2" = 100, n = 6.64 + . Accordingly beginning with a population made up entirely of individuals heterozygous for one pair of Table XLVI. — Proportioxs of Dif- ferent Genotypes and Percentages OF Heterozygotes in a Population OF Self-fertilized Plants Generation AA Aa aa Percentage of heterozygotes 2 100.0 1 1 2 1 50.0 2 3 2 3 25.0 3 7 2 7 12.5 4 15 2 15 6.25 5 31 2 31 3.125 n 2" - 1 2 2" -1 1 2" factors, it would take only seven generations of inbreeding to reduce the proportion of hetero- z^'gotes within the population below 1 per cent. As a limiting value such a population would of course consist of 50 per cent. A A and 50 per cent. aa. Jennings and others have given generalized formulae for determining the percentage of heterozygotes where any num- ber, m, of pairs of heterozygous factors is involved in the original population. Thus starting out with a single plant having m pairs of heterozj^gous factors, or a population consisting wholly of such plants, the value for h, the proportion of heterozj^gous individuals, is given by the expression: This expression is very useful for determining the degree of homo- geneity which a hybrid population may be expected to exhibit after a given number of generations of self-fertilization. Thus assuming that there are 10 pairs of factors in a given cross, what proportion of hetero- zygotes will there be after five generations of sowing? The formula is THE COMPOSITION OF PLANT POPILATIONS 321 Solving wc obtain h = 0.27; in other words, tlic clumces arc only about one in four that a plant selected from a population of this kind will be heterozygous. If there are 100 pairs of factors and ten generations of self-fertihzation only 9 per cent, of the population will be heterozygous. Thus we see how powerful is the tendency of self-fertilization to reduce the population to a homozj^gous condition. The number of homozygous genotypes to, which the population will be reduced, it should be remembered, is given by the expression, 2"', in which )n again is the number of pairs of heterozygous factors. If there are 10 pairs of heterozygous factors in the original individual, then the population will ultimately be reduced to 1024 different homozygous genotypes; if there are 100 pairs of such factors, the num])er of different kinds of genotypes is approximately 1,267,666 X 10-^. We should always remember in working with formula such as these that the}^ are only valid for conditions postulated in the premises. For the above formulae the following conditions are assumed: roughly equal viability of all genotypes, absence of any natural selection, and independent segregation of factors. Obviously none of these condi- tions is fulfilled in any even moderately complex population. We have already considered many examples of different viability in diverse geno- types, of which the many different Drosophila mutants provide the most conspicuous examples. Similarly natural selection of necessity enters in whenever any differences whatever exist in the ability of different geno- types to survive and reproduce themselves under a given set of condi- tions. In addition to these two obvious difficulties the universal occurrence of linkage also profoundly disturbs the mathematical rela- tions whenever any considerable number of factors is concerned in a given cross. It would be a very rare occurrence for even ten different pairs of factors to exhibit independent assortment in any plant species, impossible in a species like wheat which has but eight pairs of chromosomes. The biological significance of this mathematical discussion is merely this: that it demonstrates that populations in which self-fertilization is an invariable condition in seed formation must consist entirely of pure lines, if left undisturbed for a very few generations. Mathematic- ally the limiting condition is one in which all possible pure lines exist in constant proportions in the population, but biologically the limiting •condition is one in which the population is composed only of the most vigorous and productive pure lines. Populations as Affected by Crossing. — When a certain amount of natural crossing occurs the relations above described are somewhat disturbed. The population, of course, tends to reach an equilibrium, and for all practical purposes does reach one very soon, but the mathe- 21 322 GENETICS IN RELATION TO AGRICULTURE matical relations are much more complex than those given above. We may consider a simple case, however, and show the relations in that case. If we start out with a population consisting of equal numbers AA and aa forms, and assume that a given percentage of crossing occurs, then an equilibrium will be reached when the number of homozygotes produced by the heterozygotes in the population is equal to the number of hetero- zygotes produced by spontaneous crossing. Thus, if we assume 10 per cent, of spontaneous crossing in such a population, in the first gen- eration of the 10 per cent, of AA which cross with other plants, half will be fertilized by other AA plants and half by aa. The latter will give heterozygotes, consequently the proportions of different genotypes produced by the A A plants will be 0.95^ A : 0.05 Aa. Similarly aa plants produce 0.05Aa: 0.95aa, so that in the first generation the ratio is 0.95AA : O.lOAa : 0.95aa. Now in the next following generation if we assume that random mating occurs among the 10 per cent, of plants which cross with other plants, then one-third of the plants in each genotype will mate with the same genotype, one-third with one of the other two geno- types, and one-third with the remaining genotype. That is, of the 0.95AA one-tenth or 0.095 cross, as follows: ^AA X AA ^ 0.32AA, }iAA X aa =^ 0.032Aa and }iAA X Aa ^ 0.016AA : O.OlGAa. Simi- larly, of the 0.95aa, 0.095 cross: }-iaa X aa = 0.032aa, }-iaa X AA = 0.032Aa and ^aa X Aa = O.OlGAa : 0.016aa. Also of the O.lOAa, one-tenth or 0.01 cross: }iAa X AA = 0.0016AA: O.OOlGAa, ^Aa X aa = 0.0016Aa : O.OOlGaa and }4Aa X Aa = 0.0008AA : O.OOlGAa: O.OOOSaa. Summating like genotypes we have 0.05AA : O.lOAa : 0.05aa. The 90 per cent, of AA and aa plants which are self-fertilized produce 0.855AA and 0.855aa respectively, while the 0.09Aa plants which are self -fertilized produce 0.0225AA : 0.045Aa : 0.0225aa. Combining these with the results of cross-fertilization we have the ratio for the second generation, 0.928AA : 0.146Aa : 0.928aa. Now the ratio of the proportion of homozygotes to the population in the first generation is of course 0.95 and in the second generation it becomes, 0.928 + 0.928 0.928 + 0.146 + 0.928 0.927. The composition of the third, fourth and fifth generations and the ratio of the proportion of homozygotes to total population for each are shown in Table XL VII. It is evident that, under the conditions assumed in . this case, the rate of change in the ratio of homozygotes to the total population becomes very gradual after the first three generations, so that for practical purposes the population has reached a state of equilibrium in the fourth generation. In this generation the ratio of heterozygous dominants to the sum of the heterozygous and homozygous THE COMPOSITION OF PLANT POPULATIONS 323 Generation AA Aa aa Ratio x/tf* 1 0.95 0.10 0.95 0.95 2 0.928 0.146 0.928 0.927 3 0.919 0.167 0.919 0.917 4 0.915 0.175 0.915 0.913 5 0.914 0.179 0.914 0.911 dominants is O.IG +• In this or later generations, therefore, tlie chances of selecting at random a heterozygous dominant, assuming dominance to be complete, are about one in six. Table XL VII, shows the composition of the population with ref- erence to a single pair of factors, A and a, in the first five generations when there is 10 per cent, of spontaneous crossing, assuming (1) that be- fore crossing began there were equal numbers of AA and aa plants; (2) that among the 10 per cent, of plants which cross random mating occurs: (3) equal fertility and viability in all individuals. Starting again with a population of ^A and aa forms we find that, assuming 20 per cent, of crossing in this instance, other conditions being the same, the ratio of homozygotes to the whole population in the first four generations is as fol- lows: 0.90, 0.86, 0.845 and Table XLVII.— Composition of Population 0.837; while the ratio of heterozygous dominants to the total dominants in the fourth generation is 0.27. Hence, in this and later generations the chance of selecting a heterozygous dominant is about one in four. Again, with 50 per cent, of crossing the ratio of homozygotes to the whole population in the first four genera- tions is 0.50, 0.625, 0.649, 0.662; and the ratio of heterozygous dominants to the total dominants in the fourth generation is 0.50 + , so that the chance of selecting a heterozygous dominant is one in two. In the same way the theoretical expectation for any particular amount of crossing may be calculated. It must be borne in mind, of course, that we have made no allowance for greater relative vigor and productivity in the heterozygous plants. However, the method illustrated may be utilized in working out similar problems in which the genetic relations are disturbed by such conditions as difference in viability or fecundity as well as for various amounts of crossing. This brief consideration merely suggests the possibilities of mathe- matical analysis of the composition of populations under assumed condi- tions. It must be clear, however, that such analysis as applied to a given set of conditions would be of very great value in conducting breeding investigations. But it should be remembered that reliable conclusions regarding any particular case cannot be derived from such analysis unless the more important controlling agencies at least have been so carefully investigated that their combined influence can be duly esti- X = proportion of homozygotes in the popula- tion; y = value of total population. 324 GENETICS IN RELATION TO AGRICULTURE mated. On the other hand, the general principles derived from the mathematical study of the composition of populations are of universal application. These principles may be summarized as follows: 1. (a) Continued self-fertilization tends to eliminate all heterozygotes from the population. (b) The number of homozygous genotypes to which a self-fertilized population will be reduced depends upon the number of pairs of factors involved. (c) Such a population after a few generations will consist entirely of pure lines. 2. (a) With a given amount of natural crossing in the absence of any disturbing effects there will be an approximation toward a definite pro- portion of heterozygotes in the population. (b) Such a population approaches very nearly a condition of equilib- rium within a few generations. (c) Under the influence of disturbing elements the proportion of heterozygotes may be increased or decreased, but the condition of equilibrium will be rapidly approached if the disturbing elements remain fairly constant. CHAPTER XVIII SELECTION The oldest and most generally used means of plant improvement must continue to be the basic method in systematic plant breeding. Although selection is universally recognized as an effective method of breeding, yet all too long the prevailing ideas among empirical breeders regarding the way in which selection effects improvement and the reasons why selection sometimes fails in securing the end desired have been exceed- ingly vague. The confusion of thought concerning this matter which still exists among both scientists and laymen is largely due to a lack of clear understanding concerning the nature of variation. The variations upon which selection can be used effectively owe their origin either to mutations or to recombinations of genetic factors. On account of the differences in the composition of populations in various species of plants the effects of selection differ greatly in different crops. In order to employ selection most economically the plant breeder should understand the nature of the population with which he is working and the genetic prin- ciples underlying effective selection. It is our purpose in this chapter to set forth the principles of selection in both allogamous and autogamous species. Selection Methods in Maize Breeding. — The maize plant is highly variable and many different varieties and strains have been produced by selection. In most of the states where corn is grown extensively the experiment stations have published bulletins on corn improvement and the subject is discussed in more or less detail in various works on plant breeding. We shall merely consider here certain methods of maize selection in order to illustrate the principles involved and to compare them with methods used in other crop plants. Inbreeding in Maize. — Self-fertilization in maize results in marked reduction in vigor and hence in size of plant and production of seed. This was first discovered by Shull, who applied the pure line method in corn breeding, and from his results inferred that a field of maize consists of a collection of genetically distinct biotypes which may be isolated by in- breeding. East soon corroborated Shull's discovery and later East and Hayes summarized the results of inbreeding a naturally cross-fertilized plant substantially as follows: 1. There is partial loss of power of development, causing reduction in 325 326 GENETICS IN RELATION TO AGRICULTURE the rapidity and amount of cell division. This phenomenon continues only to a certain point and is in no sense an actual degeneration. 2. There is an isolation of biotypes differing in morphological char- acters accompanying the loss of vigor. 3. The hereditary differences between these biotypes is often indi- cated by regression away from instead of toward the mean of the general population. 4. As these biotypes become more constant in their characters the loss of vigor ceases to be noticeable. 5. Normal biotypes with such hereditary characters that they may be called degenerate strains are sometimes, though rarely, isolated. 6. It is possible that pure strains may be isolated that are so lacking in vigor that the mechanism of cell division does not properly perform its function, and abnormalities are thereby produced. Thus we know that any commercial variety of corn is a mixture of different genotypes and that inbreeding tends to isolate pure genotypes, i.e., inbred strains tend to become homozygous. Thus it is evident that the cross-bred progeny of two different inbred strains will be heterozy- gous for many factors. That cross-bred maize frequently displays greater vigor than either parent was first demonstrated by Beal of Michigan in 1878. But it was not until Shull and East demonstrated the existence of genotypes in maize that the genetic significance of this phenomenon became evident. The actual cause of the increased vigor has been ex- plained in various ways. Both Shull and East held that decrease in vigor in inbred strains is due to reduction in the number of heterozygous factor combinations and that increase in vigor in Fi hybrids is the result of increase in the number of such combinations. The general occurrence of decrease in vigor upon inbreeding naturally cross-bred species and of increase in vigor upon crossing closely related forms led them to conclude that heterozygosis is the cause of increased physiological vigor in Fi hybrids. Other explanations of this phenomenon have been offered, one of which was that of Keeble and Pellew, to the effect that it "may be due to the meeting in the zygote of dominant growth factors of more than one allelomorphic pair, one (or more) provided by the gametes of one parent, the other (or others) by the gametes of the other parent." East and Hayes reject this hypothesis on the grounds that this increase in vigor "is too universal a phenomenon among crosses to have any such explanation. Furthermore, such interpretation would not fitly explain the fact that all maize varieties lose vigor when inbred." But there is good evidence that all maize varieties do not lose vigor to the same extent when inbred and that certain genotypes produce much more vigorous Fi hybrids when crossed than other genotypes. As was stated in Chapter XII, D. F. Jones has explained this increased vigor SELECTION 327 in F, hybrids in tornis of dominance and linkage (p. 231, 2). The fact that different genotypes give diverse results when crossed is of immense practical significance. The Ear-to-row Method. — This has been the method of commercial corn improvement for many years and it is well illustrated by the IlUnois corn breeding experiments, which have been going on continuously for over 20 years. The original purpose of the experiments was to produce new strains which would be more valuable as a source of feed for hvestock. It was found that there was considerable variation in the relative amounts of protein and carbohydrates in the grains of different ears. Accordingly selection was begun with the object of increasing the protein and reducing the starch content of the grains; also of decreasing protein and increasing starch. As oil was worth three times as much as starch per unit of weight, selection for higher oil content was also begun, A low oil strain was started for comparison and such corn was soon found to be desirable for the production of pork and beef of high quality. The work was begun by Hopkins who picked out 163 ears of a local strain known as Burr's White, made a chemical analysis of a few grains from each ear, and on that basis sorted them into four classes, viz., high and low protein and high and low oil. The strains were grown in isolated plots from the beginning. After 9 years of selection it was found to be necessary to prevent inbreeding. Accordingly in the tenth and succeed- ing years about 24 ears were selected for each plot and one row was planted from each ear, then the even numbered rows were detasseled. Subsequent selections were made from the detasseled rows, the first consideration always being high yield. Usually 20 ears were taken from each of the six higher yielding rows, or 120 ears for each plot. These were tested by chemical analyses and the most extreme variants in the desired directions were selected for the next planting. The results in general have been more regular in the high and low oil series than in the high and low protein strains. In the latter there seems to have been no very decided effect of selection after the first 10 years. Similarly there has been no continuous advance in the low oil strain since the seventeenth year of selection, but in the high strain the per cent, of oil has continued to increase slightly. The progressive effects of selection in the four series are graphically illustrated in Figs. 133 and 134. That the striking results depicted in these graphs were not caused by environ- mental conditions was proved by planting mixed plots with two grains of ''high" and two of "low" corn in each hill so arranged that the resulting plants could be identified. This test, according to L. H. Smith, was made for three successive years, and subsequent analyses showed that under these conditions the different strains maintained their distinguishing chemical characters. 328 GENETICS IN RELATION TO AGRICULTURE The Illinois Station experiments have included selection for many- other characters of the corn plant in more recent years. One of the most striking results was obtained by selecting for height of ear on the plant. Data on which to base selection were secured by measuring several hundred stalks in the oil and protein plots, noting height of ear above the ground, total height of stalk, apparent number of internodes below the 'B6 '97 -as '99 '00 '01 '02 '03 '04 '05 'OC '07 'OS '09 '10 '11 '12 '13 'H '15 Year of Selection Fig. 133. — Two graphs representing the effects of selection for high and low oil content in the Illinois Station corn experiments. {Data from Castle.) ear and number of internodes above the ear. Fig. 135 shows the result of selecting for high and low ears during five generations. Similar results were obtained from selection in the case of position of ear at maturity and total yield. The striking results of these carefully conducted experiments have been citfed by various authors as evidence par excellence for the most Fig. 134.- '96 '97 "98 '99 '00 '01 '02 '03 '01 '05 '00 '07 '03 '09 '10 'U '12 '13 '11 '15 Year of Selection -Two graphs representing the effects of selection for high and low protein eon- tent in the Illinois Station corn experiments. {Data from Castle.) diverse conceptions of the rdle which selection plays in evolution and breeding. Thus the earlier allusions of Hopkins and Smith, the discus- sion in E. Davenport's text on breeding, and the recent treatment by Castle all seem to attribute a peculiar creative power to selection which meets with a certain "response" on the part of the plant. This is in line with the Darwinian idea that all fluctuating variations are heritable and that the continuous selection of minor fluctuations in a certain direction is always effective in shifting the type. SELECTIONS 329 The futility of attempting to generalize regarding the effects of selec- tion in plants must be obvious from what we now know about the com- position of plant populations. With the application of Johannsen's genotype conception in analyzing the composition of a field of maize the problem of explaining the role of selection in the Illinois corn breeding experiments was immediately simplified. This was perceived by ShuU who pointed out that the results of these experiments might be readily explained on the ground that some hybrid combinations of genotypes have greater capacity for the production of the desired qualities than other combinations, and that the selection has gradually l)rought about Fig. 135. — Result of selecting corn for high and low ears during 5 generations. The white tape marks the position of the ears on the front row of plants in both plots. the segregation of those genotype combinations which had the highest capacity for the production of the desired quality. Meanwhile Surface had made an illuminating analysis of the data from the first 10 years of selection as reported by Smith. This treatment is so valuable as to warrant its examination in some detail. At the time the selections were made a careful record of the pedigree of each ear was kept. These pedigrees are of course for the maternal side only since self-pollination was not practised. From these data Surface prepared a pedigree chart for each of the four strains. The chart for the high- protein strain is reproduced in Tables XL VIII and XLIX. As stated above 24 ears containing the highest per cent, of protein were selected for the 163 ears analyzed in 1896. These were given registry numbers from 101 to 124 inclusive as shown in column one of the two tables. For convenience we may refer to these ears as the first generation of high- 330 GENETICS IN RELATION TO AGRICULTURE Table XL VIII. — Pedigree Chart of High-protein Corn — Part I. {After Surface.) Generation Number 1 2 3 4 5 6 7 8 9 10 11 101 102- 215- 320- 410- 502 103- 208- 314.. . 424 409 104- 214..^ f 316- 310 421 105 306. . ■ 401- 514 106.. {III. 315- 319.. 405 418 417 414 416 107.. {^^r 301 108.. {,206- 321.. ■ 415 406 109 407 - 510 110 311.. 420 111 312 411. . { 506 - 604 513 212.. 313 404 [614 112. . 309.. 402 - 408 422 515.. 606 [603 205- 317.. 412.. { 503 509 - 613 - 705 - - 822 113 (210 i^-^- • \220- 324 114- 204- 303 115- 224- 304 116- 202 117 118- 218- 302 119 (221- i^y.. (201- 307 - 305 403 - 511 120- 203- 308 SELECTION 331 Table XLIX. — Pedigree Chart of High-protein Corn — Part II. {After Surface.) Generation Number 1 2 3 4 5 6 7 8 9 10 11 '706.. r808. 813 /901 I 908 609 . . < [801 ^512.. [ 702 '912. f 1007 1002 1 1014 . 1019 U08 r612- 711 .814. J 1 |'423 601- 710.. 810 ] 821 925 1 920. f 1001 f Vm 1020 -' ^'^ 1013 ^ ""'"^ 505 f 916 121(207 -323... 413. 501 r 719 fiO'? J 721 i 712 ^ 811. '819 902 ' 905 .914. fl009 .J, J 1016 02 1 ITl 1 1107 I i'J^i ^ 1119 906 ] [ 924 . . ' 713.. , 806. 802 [1017 .,. J 1024 Jjlf 1 1012 -' ^^ 005 1120 ^ ^"""^ [ 1108 507. . 704 820. r903 917 [ 910 508 607 . . 1 714 ' nil r 1010 j 1106 1 1123 [ 1118 716- 807 /911 \ 922.. 1 717.. J 818.. 1003 1015 r ,,^r [ 804 f915 ( 1105 605 / '^07 605.. ^ 720 '812. . 805 919 [ 923 1022, 1112 1117 f 708 815 709.. f 907 I fill J 803.. i 913 ' 1011 on . . < 703 817 t 904 f 918.. 1018 f {1^9 1006. ^ }J?^ [ WZ.i ^ JJ21 122 I 701 1 809. . \ [ 921 123 .c,, / 216 -318 ^^''•- \ 209- 322-419- 504- 610 ^"^18- 816- 909 332 GENETICS IN RELATION TO AGRICULTURE protein corn. The next season 4 sound ears were analyzed from each of the twenty-four rows. From these 96 ears the 24 again having the highest per cent, of protein were selected for planting. The distribution of these selected ears among the 24 original ears is shown in column two of the tables. For example, it is seen that ear No. 124 produced 2 ears, Nos. 216 and 209, which were among the first 24 as regards protein content. Ear No. 123 on the other hand failed to produce any ear (so far as the ears analyzed showed) sufficiently rich in protein to be in- cluded among the first twenty-four. Thus 8 of the original ears fail to be represented in the second generation, while 8 other ears contributed 2 ears each for planting the following year. Exactly the same selection was practised in the second year and the resulting selected ears are shown in the third column of the tables. Of the 16 original ears rep- resented in the second generation only one. No. 116, was dropped out in the third generation but in the next generation there is a significant dropping out of some of the original lines, so that in the fourth genera- tion only 9 of the original 24 ears are represented by progeny. Five of the original lines contribute 80 per cent, of this generation, while two lines, 106 and 112, contribute nearly 60 per cent. Hence at the end of the fourth generation it is clear that certain of the original lines have a much greater tendency to produce ears with a high per cent, of protein. By simply selecting on the basis of the protein content of the individual ear for 4 years 70 per cent, of the original lines have been dropped. Thus the elimination of the original lines gradually proceeds until, in the tenth and eleventh generations all of the high-protein corn is the offspring of a single ear, viz.. No. 121. It will be remembered that in the tenth year the method of detasseling alternate rows and saving seed from these only was put into effect. But this change in method could not have induced the results we have noted because line No. 121 had demon- strated its superiority over all the others as early as the seventh genera- tion. This isolation of a single line was brought about therefore simply by selecting each year those individual ears that showed the highest per cent, of protein. Starting with a protein content of 10.92 per cent., at the end of the third year (fourth generation, 1899) the protein content was only 11.46 per cent, or a gain of 0.54 per cent. But the next year (fifth generation) the protein content jumped to 12.32 or a gain of 0.86 per cent, in 1 year. Referring now to Table XLIX it is seen that it is in 1899 that a great reduction was made in the number of lines represented, for in the fifth generation only six of the original twenty-four lines remain. Furthermore it is just here that line No. 121 begins to show its superiority since 5 of the 15 ears selected in 1900 or 33)^^ per cent, come from this line. The course of events in the other three strains was similar but not SELECTION 333 Table L. — Genetic Relations Between Certain Physiological Characters of the Corn Grain. Character Dominant Recessive Moisture High Nitrogen and protein| Low Crude fat. quite so striking. In the low-protein strain only two of the twelve original lines are represented in the eleventh generation; in the high-oil strain three lines out of twenty-four are maintained throughout the 10-year period ; and in the low-oil strain only two lines out of twelve are represented in the eleventh generation. These results are exactly what would necessarily accrue in any al- logamous species under continuous selection for a given character, pro- vided the degree of expression of that character is dependent upon a number of genetic factors. That several chemical characters of the corn grain, including protein and oil (fat), are inherited in accordance with Mendelian principles was determined })y Pearl and Bartlett in 1911. In a cross between a white sweet corn and a yellow starchy corn determina- tions were made by direct analysis of the percentage content of the grains of the pure parent races and the Fi and Fi progeny in respect to nine chemical constituents. These are hsted in Table L, which also indicates the dominant and recessive conditions of these characters in the cross studied. This evidence^ although worked out quite independ- ently, supplements Surface's analysis of the lUinois data in a remark- able way. Although there are technical obstacles to a clear cut de- termination of the factor relations' involved, yet there is no question whatever that these characters of high and low protein and oil are con- ditioned by unit factors. A priori there is no objection to assuming the existence of several factors which affect the percentage of protein, for example, and that the original ear, 121, of the superior Hne in the high protein strain represented a genotype rich in high protein factors. Similarly in the other strains, continual ear-to-row selection has gradually eliminated all genotypes except the one, two or three as the case may be of highest or lowest factor combinations. Thus we see that selection has created nothing in the course of these justly famous experiments; it has served merely as a means of isolatmg particular combinations of factors which condition oil and protein pro- duction in the corn plant. Moreover, this sorting process has not been Ash Crude fiber. Pentosans . . Sucrose . Dextrose . Starch. . . Low (incomplete dominance) Low Low Low (incomplete dominance) Low (incomplete dominance) Low High Low High High High High High High High Low 334 GENETICS IN RELATION TO AGRICULTURE entirely regular or continuous. The saltations or jumps revealed by- Surface's analysis were directly consequent upon lump elimination of a number of mediocre lines. These results, therefore, are in entire har- mony with the known nature of allogamous populations. This conclu- sion is further corroborated by the recent report of Reitz and Smith on the statistical study of indirect effects of selection for high and low pro- tein and oil. These authors state: ''It is found that four distinct types of corn as regards length, circumfer- ence, weight of ears, and number of rows of kernels on ears are so well estab- lished that we may assign orders of values to the means of these characters that persist with but a few exceptions in such changes of environment as have been experienced in 11 years of planting, from 1905 to 1915. "While a few slight but progressive changes have been noted, the selections for chemical composition from 1905 to 1915 have not changed decidedly the dif- ferences in mean values of these characters. In fact, we are unable to assert with any high degree of probability that the strains differ more or less with respect to these characters during the second half of the period 1905 to 1915 than during the first half." The italics are ours. It is of especial significance that careful biometrical study has failed to reveal any progressive change as a result of continued selection in these strains of corn. For the results of these experiments have been cited as evidence par excellence by Castle in support of his hypothesis of factor variability. The ear-to-row method has been modified in various ways but it still forms the basis of most systems of commercial corn breeding. A popular feature of systematic corn improvement is the use of score cards in judging. A special development of the score card method of selection is the use of selection index numbers as advocated by Pearl and Surface. In this plan arbitrary values are assigned to various characters of the corn ear, for example, such as absolute size of the ear, average percentage depth of the grains, etc. The idea is to combine in a single numerical expression the values of a series of variable characters with regard to all of which the breeder wishes to practice selection at the same time. The index numbers of different varieties are not directly comparable but for a given variety they may be useful as an adjunct of the score card method. However, their use requires more attention to details and hence greater expense than most breeders can afford to give. Their use in plant breeding will probably be limited to experiment stations (see Chapter XXXI). The danger of continued ear-to-row selection or "narrow breeding" within a variety was pointed out in 1909 by Colhns, who emphasized the importance of "broad breeding" in such crops as exhibit loss of vigor when closely inbred. About the same time Williams inaugurated SELECTION 335 the remnant system of corn breeding at the Ohio Experiment Station and the plan was adopted hy the Ohio Corn Improvement Association. The plan calls for an ear-to-row test plot each year in which ears are carefully tested for productiveness. Only half of the grains on each car are planted in the test plot, the remainder being retained until the following year under the term "remnant." The car-to-row test plot need not be isolated as no seed is saved from it. The next year the remnants of a few, usually four, of the highest yielding ears are planted in an isolated breeding plot, and the stalks from all of the ears planted in this patch, except those of the highest yielding ear, are detasseled. Seed ears are selected from the detasseled rows and grown the next ITf^^^lHi^^^^^HBI^^^^^Hil^H ^^V^^T^H ^^'^^ m^^M^^KW^^^^ \^M ^m^^ ^^Hil^l K wmMKmWm) nl n^^H ^S/mk ^H ^B <- "^■1 ^^1 - ^B 1 ■ I il^H HH Hi 1^1 .1 rfiB^^^I ■V 11 - ^H H jj^l »p 3|^^^^| MiWM ■ r ^H 1 r^fl H't^^H I'iH Jl^^l ^mJM ilH ^m aH ^^kuM^I H ^^^^^1 ^^^^^^1 Fig. 136. — Delta Farm White Dent, a superior strain of maize adapted to the bottom lands of the interior valleys of California. It is the result of 30 years of continuous selection of seed in the field before harvesting. The original material consisted of a mixture of all the types of corn commonly grown at that time. A convincing demonstration of the prac- tical value of seed selection as a general agricultural practice. year in a multiplying plot to supply seed for general planting. After this method is under way on any farm, there is maintained on the farm each year a small isolated breeding plot, a multiplying plot, and an ear-to-row test plot. This method successfully excludes from the breed- ing plot all individuals except those whose producing power has been found to be very high. At the same time it provides for the intercrossing of these most productive strains, and by continuing the tests from year to year the work will "tend toward the selection of the best producing ears for all or average seasons." According to Hartley, "The choice of a high yielding variety is important; the choice of high yielding ears is even more important." The remnant system combines this result with the advantages attendant upon intercrossing of distinct strains. Rye, clover, beets, timothy and other grasses are suited to this method of breeding. It was with rye that Rimpau first employed the system that later came to be known as the German method of broad breeding. 336 GENETICS IN RELATION TO AGRICULTURE An interesting illustration of what can be accomplished in maize merely by mass selection, when a definite ideal is maintained and seed is selected in the field before harvesting, is found in the Delta Farm White Corn shown in Fig. 136. Selection Methods in Breeding Close-pollinated Plants. — The suc- cessful methods of breeding wheat have been reviewed in preceding chapters. Compared with the methods required for corn the work of isolating genotypes in wheat is relatively simple. Most commercial I l2 Fig. 137. — Typical heads from seven pure lines of Defiance wheat. Nos. 1 and 2 do not yield one grain per spikelet on the average; Nos. 6 and 7 yield from 4 to 7 grains per spikelet. Note tendency to club type in No. 6. varieties of wheat are a mixture of pure lines which can be isolated by single plant selections. In Fig. 137 is shown a typical head from each of 7 different pure lines isolated by selecting single plants from a plot of Defiance wheat. Nos. 1 and 2 did not have an average of one grain per spikelet while Nos. 6 and 7 bore from 4 to 7 grains per spikelet. If Nos. 6 and 7 prove to be superior in other characters also, they need only to be multiplied in order to yield greatly improved strains of the Defiance variety. It was by this method that Roberts in 1906 isolated a pure line of Turkey wheat that appears very promising for the Great Plains Region. It is worthy of note that this superior pure line was the 135th SELECTION 337 single head selection made by Roberts in 1906. Altogether he made 557 selections from nearly 200 different varieties; but nearly 415 of these were discarded within 2 j'oars. The Plant-to -row Method. — Single plant selections are usually grown in garden rows, each row from a different plant. Final selection of the individual plants should be preceded by field observations, noting habit, vigor, disease resistance, season of l)looni, time of maturity, productivity, etc. Each of these plants must be harvested separately and careful Fig. 138.- -Spreading and tTcct pure lines of Gypsy wheat, 1907. A.E.S.) {After Williams, Ohio records should be made concerning yield and other important characters. It is on the basis of the field observations and the data from the harvested plants that a further selection must be made. From each of the plants finally selected sufficient seed is taken for a row of about 25 plants. The rows should be evenly placed and plants should be equidistant in the row. By subjecting these rows to severe selection the future work may be considerably reduced. Hence careful notes should be taken throughout the season and at harvesting time. Of several hundred rows only a few may be found good enough to be continued. The third year the 22 338 GENETICS IN RELATION TO AGRICULTURE Fig. 139. — The erect pure line of Gypsy wheat in 1909. (After Williams, Ohio A. E. S.) Fig. 140. — The same pure line (on the right) now known as the Gladden variety, as grown in 1915. (After Williams, Ohio A. E. S.) ^ SELECTION 339 seed from the selected rows is sown in small multiplying plots. At the Maine Station these plots are 1-2000 acre in area and usually in duplicate (see Fig. 141). These plots are subjected to still further selection and only the best retained. The next step is to sow each selected pure hne in one or more field plots. At Maine 1-40 acre plots are used and each line is tested in duplicate or quadruplicate plots for several years and only those that are superior in some respects at least to commercial varieties are retained. At the Ohio Experiment Station according to Williams, "In following the pure line method of selection, decided dif- ferences in winter resistance, stiffness of straw, yield of grain and bread- FiG. 141. — Planting board usecLin pure line work with small grains at the Maine Exper- iment Station. It provides a plot ^-2000 acre in area with the plants nearly as close together as when sown in the field. {After Surface and Zinn, Maine A. E. S.) making qualities have been found in the progeny of individual heads selected from ordinary varieties of wheat." In Fig. 138 are shown two ■very distinct pure lines of Gypsy wheat as they appeared in 1907. In Figs. 139 and 140 the same pure line appears as grown in 1909 and 1915 respectively. This selection, has been introduced under the name of Gladden. Ineffectiveness of Continued Selection Within Pure Lines. — Con- vinced of their failure to make any progress as a result of continued selection within pure lines, some experiment stations have abandoned this line of work. Hutcheson has reported on the results of 13 years of continuous selection in six pure lines that were isolated by Hays in 1901. These pure lines represent five of the sub-species of common wheat, Triticum vulgare. In brief the method consisted of selecting each year the best 100 grains from each of five or more best plants in each line. This seed was planted at regular distances in centgener plots the following year, each centgener representing a single plant selection. Hutcheson 340 GENETICS IN RELATION TO AGRICULTURE says, "the indications are that from a practical breeder's standpoint permanent improvement in pure Hnes in small grains, if possible, is certainly not rapid or apt to be very marked." He also suggests that much more rapid progress could be made by isolating pure hnes from mixed populations and combining the desirable characters of these lines by hybridization. Other crops in which the method of selecting pure lines is applicable are oats, barley, peas and beans. Notable improvement has been made in oats by this method at the Svalof, Cornell University and Maine Experiment Stations. The general method of procedure at the Maine Station is indicated by Surface and Zinn in their bulletin on pure line varieties of oats (see p. 371). The pure lines finally retained came from only three varieties, viz., Banner, Irish Victor and Imported Scotch. It is noteworthy and consistent with Mendelian principles that the physiological characters which result in higher yield are not necessarily associated with morphological characters in the plant or grain. Similar results with winter-resistant barleys have been reported by Spragg of the Michigan Station. The practical importance of the selection of pure lines as one phase of a complete system of breeding as practised with autogamous species is given further attention in Chapter XXI. Selection which is to result in the isolation of the most superior genotypes must begin with individual plants. In dioecious and self- sterile plants this method is inapplicable. Here the breeder must begin with phenotypically similar individuals and continue inbreeding of simi- lar plants for several generations in order to isolate approximately uniform strains. The earlier improvement of the sugar beet was accomplished by mass selection. But in recent years the producers of commercial seed have introduced a system of line selection. According to Briem, reliable seeds cannot be obtained by selection in the lump, nor from a single generation of mother beets followed by the cultivation of seed roots. An individual selection must be made the characteristics of which are assured by testing for three generations. That is to say, since the beet is a biennial 6 years are required to obtain seed of guaranteed quality for the seed roots and another 2 years must pass before the market product is ready. Briem's opinion is in harmony with Pritchard's con- clusion that continuous selection is not an efficient method of sugar beet improvement and that the improvement of the past is the result of isolating mutations (see p. 369). In emphasizing the importance of finding the best genotypes within a chosen species or variety the usefulness of mass selection should not be overlooked. It is frequently the first or only practicable step to take in purifying a commercial variety. The so-called "running-out" of varie- ties can be prevented by reasonable care to avoid mixing seed and by occa- SELECTION 341 sional mass selection from the field. Seed selection of this sort is of the greatest practical value to agriculture and it is applicable to most sorts of field and garden crops. The Practical Importance of Keeping Varieties Pure. — Many farmers do not regard purity of varieties as a matter of great concern, but con- tinue to use impure seed from year to year. Since the main object of breeding work is to produce new and better varieties, and since a true variety differs definitely from all other varieties, it is of great impor- tance that its purity and hence its identity be maintained. The need for care in this regard is of course much greater in naturally cross-fertilized species than in self-fertilized forms, yet even in the latter the mixing of varieties may detract greatly from the market value of the crop. It is not impossible for an impure variety or a mixture of varieties to give good returns for a year or two or even longer. When one considers, however, the rapidity with which the number of distinct strains may be increased by occasional crossing the danger of such practice will be realized. For this reason all agencies supervising the collection of stock seed of com- mercial varieties of corn, sorghum, cotton, etc., should exercise every possible precaution against mixing varieties or collecting seed that may have been crossed with other varieties. As Newman points out, however, there are certain circumstances under which the planting of mixed sorts may have their advantages. Thus a variety may contain strains which differ from each other for example chiefly in their response to different soil conditions. Were a variety of such composition sown in a field in which the soil is exceedingly variable it is possible that a better average would be maintained than from an absolutely pure sort which demands more exact conditions. Yet even here the practice marked of careful mass selection in the field would doubtless result in marked improvement. In general, however, the diflficulty of knowing the real nature of the strains which compose a mixed variety makes it unsafe to depend upon the possible virtues of maintaining the most advantageous mixture. Proved sorts of general adaptability offer much greater promise. CHAPTER XIX HYBRIDIZATION The usual purpose of the plant breeder who resorts to hybridiza- tion is to secure new and better combinations of characters among the progenies resulting from his crosses. Improvement of a given species may consist merely in the eUmination of undesirable characters or of the production of entirely new combinations of characters already existing within the species. In this work the apphcation of the Mendehan principle of segregation and recombination is of the greatest prac- tical value. By concentrating his attention on only a few important characters at a time the breeder can sometimes secure the desired combinations in F2. But at the same time one who is informed in regard to modern genetical principles will be prepared for possible disappointment in meeting an early realization of his aim. Moreover, he will understand how to select in F2 and later generations for further testing. He will realize that a specific character difference in his parental forms may be conditioned by more than one factor difference; also that some specific factors display considerable variabihty in expression; and that hnkage, crossing-over, multiple factors and multiple allelomorphs may play a role in conditioning or preventing the particular character combination for which he is striving. Furthermore, the ideal sometimes demanded of the breeder involves character complexes which include all the functions of the plant. As has been shown already the comparative difficulty between different cases of this sort depends directly upon the number of chromosomes possessed by the species in question. Finally, demands are sometimes made for the "creation" of characters which are unknown in available phenotypes and for which there is no genotypic representation within the species. In such cases recourse may be had perhaps to species hybridization. But those who are famihar with the results of species hybridization will be prepared for complete disappoint- ment from the first. It is not the motive of these remarks to discourage intending hybridizers, but merely to warn against the anticipation of success in all cases simply because of the generahty of Mendehan princi- ples. Hybridization, even of varieties, in order to be generally successful must be intelhgently performed and in the long run the experimentahst who is the most thoroughly informed concerning his plants will stand the best chance of securing the improved forms he desires. Each species 342 HYBRIDIZATION 343 has its own morphological and physiological peculiarities and general methods will need to be modified to some extent in almost every case. General Method. — -Some results of value have come from promis- cuous crossing of varieties and species that appeared to give promise of desirable combinations. Considerable hybridization has been done in this way especially in establishments where large collections are main- tained and by seedmen and nurserymen who have undertaken such work as a side issue. Some of the most important results of such work have been the accidental discovery of unforeseen possibihtics or hmitations in crossing. But many important results have come from carefully planned and executed experiments and the demands of modern agri- culture necessitate systematic procedure in the employment of hyl)ridi- zation in plant improvement. Such procedure includes six steps. 1. Choice of Parents. — This involves two important matters: first, decision regarding the object to be attained which imphes thorough famiharity with existing conditions affecting crop production; second, comparative study of existing varieties or of species that may yield the desired result. 2. Culture of Parent Plants. — Hybridization is painstaking work and when carried on extensively it is time-consuming and, therefore, expensive. While it is sometimes necessary to use certain plants, especially shrubs and trees, wherever they happen to be growing, yet it is always advisable to concentrate materials so far as possible and to grow them under protec- tion in the breeding garden or greenhouse. Arrangement of the details of culture should include consideration of the optimum conditions for normal fruitfulness of the intended mother plants. These plants in some cases must be kept under observation and prepared for crossing by reducing vegetative growth and restricting blooming and the setting of fruit. 3. Protection of Pollen.- — Flowers on intended male parents should be guarded in order to prevent contamination with pollen of other plants. 4. Castration of Hermaphrodite Flowers. — This must be accomplished before anthesis and is usually done shortly before the flower opens in order to avoid needless mutilation. But in some close-polHnated species it is necessary to emasculate very young buds. The operation consists of removal of the stamens and can usually be accompUshed easily by using a pair of fine pointed forceps or scissors. The castrated flower is then protected with some sort of covering until ready for poUination. In moncEcious plants it is necessary to guard the young pistillate flowers which are to be pollinated. 5. Pollination. — The transfer of pollen from guarded flowers of the male parent to the prepared flowers of the mother plant should be accomplished before or just at the time the stigma becomes receptive. 344 GENETICS IN RELATION TO AGRICULTURE In many species this receptive condition of the stigma is evidenced by the secretion of a viscid fluid on the stigmatic surface. It has been thought that premature polhnation wrought disastrous effects on the resulting progeny, but evidence is conflicting on this point. Certain it is that in some species, for example, wheat, no untoward results appear from pollination at the time of castration. Plants with small, entomophilous flowers such as clover and alfalfa may be hybridized by enclosing the insects in a cage surrounding the intended mother plant or plants. 6. Protection of Pollinated Flowers and Developing Seed. — The most commonly used device is the paper bag tied with a string or lead wire or fastened securely with a copper wire label on which the necessary data are written. In many cases ordinary manila bags of suitable size are entirely satisfactory. Where wasps give trouble by cutting holes the use of bags made of ramie fiber will be found more satisfactory since these bags are made with a glossy surface, but even these will give way under the attack of wasps in course of time. Bags made of thin paper which has been treated with oil or paraffine are best for withstanding insect attacks and for use on dehcate plants. Many special devices, such as glass or celluloid cylinders plugged with cotton and firmly supported, are used upon occasion. Method of Hybridizing Maize. — The technique with this plant is simple, but when working among plants growing in close proximity to each other considerable care is necessary in order to prevent accidental crossing. For protection of the pollen manilla or ramie bags, size No. 8, are tied over the top of the plant just as the staminate inflorescence (tassel) is beginning to appear. The female flowers to be crossed must be covered before any of the stigmas (silks) have protruded through the tip of the ear and become exposed. The most satisfactory covering for this purpose is a strong paper bag about the size of the bags used for the tassels. It has been found economical of time to use bags which are folded so that the center line of the bottom is exposed (not "square bottom" bags) and to slit the bottom, fold over once and fasten with a wire clip before covering the ear, with minimum danger of introducing foreign pollen. This device makes it possible to examine the develop- ment of the stigmas. After stripping off the leaf subtending the young ear to be covered the bag is pulled down over the ear as far as possible and tied securely to the stem of the plant. When the stigmas are well de- veloped and while they are still fresh the bag containing the tassel is removed from the intended male parent and carried to the plant which is to be polHnated. A hole is torn in one corner of the bag, the top of the bag covering the ear to be poUinated is then opened, the pollen is dusted over the stigmas and the bag enclosing the ear is closed immediately thereafter and securely fastened. Full data concerning the cross are recorded on a HYBRIDIZATION 345 wooden la])el with copper wires which is attached to the ear. If it is de- sired to make a second polUnation the bag containing pollen maybe tied to the plant alongside the ear and the same process repeated one or two days later. By close observation of the developing stigmas and pollinat- ing at the most propitious time well developed ears can be secured from a single polhnation. Fig. 142 illustrates the principal features above described. Fig. 142. — Hybridization of maize. Right, plant just previous to anthesis with leaves subtending ears stripped off; left, the same plant with bag enclosing tassel and cylinders covering ears. Method of Hybridizing Wheat. — ^This plant has numerous hermaph- rodite flowers arranged in a branched spike (Fig. 143, upper left). Each spikelet bears two rows of bracts or glumes (Fig. 143, 2). The lowest two bracts are sterile but each of the next four usually subtends a flower while at the top of the spikelet are two or three rudimentary flowers. Each flower consists of an ovary with two much branched stigmas (Fig. 143, 12, 13) and three stamens which are shown in cross-section in Fig. 346 GENETICS IN RELATION TO AGRICULTURE Fig. 143. — Details of wheat inflorescence. The smaller spike is Fife and at its left is shown a Blue Stem spike. In the lower right- hand corner is a spike from which small late flowers have been removed preparatory to crossing. At 2, spikelet, natural size, with a few joints of the rachis;/and g are flowerless glumes; k, florets bearing seeds; r, rudimentary florets. 3, a single flower closed just after flowering, X3. 4A, longitudinal diagram before flowering, X2.5; anthers marked a; ovary, o; stigma, s; filament, /. 4JB, diagram of floret just after flowering, X3, showing how anthers are held within the envelope. 5, transverse diagrammatic section, or floral plan, as is made by cutting across ^A at x, XQ;/g, flowering glume or lemma; p, palea; a, anthers; s, stigma. 6, flowerless glume; 7, flowering glume or lemma; 8, palea; all slightly reduced. 9, lodicule, X4, shown also at L in 4B. 10, cross-section of anther, X26; showing the pollen sacs and the central mass of tissue to which they are attached. HYBRIDIZATION 347 143, 5. The essential organs are completely enclosed by two bracts, the floral glume or lemma, which bears an awn in bearded varieties, and the smaller palea. The lemma and the palea open for a short time during Fig. 144. — Hybridizing wheat. Note position of operator and his equipment, con- sisting of a box containing strips of paper and pins for covering the wheat heads and tags for labeling, a flask of alcohol for sterilizing the hands and instruments, a pair of forceps and a scalpel. Pollinated heads which have been wrapped and labeled are shown at the left. anthesis, but as a rule some pollen is shed upon the stigma before the flower opens. The flowers remain open only a short time in cool climates 11, pollen grains, round and smooth, 55 micro-millimeters in diameter. 12, ovary and stigma just prior to flowering; 13, at the time of flowering; and 14, shortly after flowering. 15, 16 and 17, the mature seed; a, the ventral side; h, the dorsal side; c, the germ or chit; s, the stem end of the germs; r, the root end; e, outer layers or bran; d, the incurved surface of bran on the ventral side of the seed. The white portions of 16 and 17 are the floury interior consisting of cells containing the gluten and starch from which white flour is made. (^After Hays and Boss, Minn. A. E. S.) 348 GENETICS IN RELATION TO AGRICULTURE but a number of natural hybrids in wheat have been reported, especially in sub-tropical countries. In the immature flower the anthers are short and closely packed around the pistil. Just before anthesis the filaments lengthen sufficiently to allow the anthers to protrude when the flower opens. Castration can be accomphshed without difficulty by choosing flowers nearly ready to open and removing the later flowers on the upper portion of the spike as shown in Fig. 143, lower right. With a pair of fine forceps the lemma and palea are forced apart and the anthers care- fully removed. From a head of the desired male parent anthers just ready to burst are then removed and an anther is placed in each of the castrated flowers (Fig. 144). Two pairs of forceps should be used, one Fig. 145. — Sexual columns of alfalfa flowers (enlarged 7 diameters), showing differ- ent stages of development: A and B, anthers just before dehiscence: C and D, anthers dehisced; E and F, after treatment with water jet previous to artificial pollination. {After Oliver.) for castrating and the other for pollinating; or, if the same instrument is used, it should be steriHzed by dipping in alcohol between each operation. After polhnation the spike is bagged or wrapped with several thicknesses of cheesecloth and labeled with a paper string tag. The use of the cloth and light weight tag is to be preferred because most grain plants will support this extra weight without staking (see Fig. 144). A Method of Hybridizing Alfalfa. — An ingenious method of crossing this and similar small flowered species has been worked out by Oliver. The essential points are as follows : First, have pollen from male parent at hand ready to be applied to the prepared stigmas. This is accom- plished by taking a flower from a raceme of the male parent, securing the banner between the thumb and forefinger and pressing a pin against the suture of the keel, beginning at the base and gradually drawing it upward. When this is done carefully the stamens and pistil come out gently without disturbing the masses of slightly adhesive pollen (see Fig. 145, C, D). Now with the aid of self-closing forceps sever the sexual organs from HYBRIDIZATION 349 the flower and lay aside ready for application to the stigmas of the flowers which are to be depollinated. Second, select a raceme in which the terminal buds arc about to expand and cut away all the buds and flowers except three or four near the center of the raceme. The flowers should not be mutilated in any way and should be handled as little as possible. In these flowers the stamens will have dehisced perhaps a day or two previously but the pollen cannot reach the stigma until the flower is tripped. When the tripping is uncontrolled the sexual column (pistil and stamens) flies upward and strikes the banner with considerable force Fig. 146. — Flowers of alfalfa (enlarged 4 diameters) showing method of depollinating and crossing: A, untripped and unpoUinated; B, tripped and self-pollinated; C, tripped against a pin to prevent self-pollination and permit depoUination; D, after depollination with water jet; E, after artificial pollination; F, after withdrawal of pin the stigma presses against the surface of the banner. (After Oliver.) and pollen grains are imbedded on the receptive stigmatic surface. It is necessary therefore to trip the flower gently and to prevent the stigma from touching the banner which is accomplished by inserting a short pin between the sexual column and the banner (see Fig. 146, C). Third, depollination is accomplished by the use of a fine jet of water from a dental chip blower; "the jet may be of sufficient force to remove even the empty anthers without injury to the stigma." Then remove the water adhering to the flower with a piece of clean, soft blotting paper. Fourth, apply the waiting pollen to the depollinated stigma and gently remove the pin allowing the stigma to press against the banner (Fig. 146, F). ''The operation is performed in much less time than it takes to describe it and the operator is rewarded by a fairly high percentage of success- 350 GENETICS IN RELATION TO AGRICULTURE ful crosses." The various implements mentioned above are shown in Fig. 147. Some of the Difficulties Attending Hybridization. — (a) Different Seasons of Maturity. — This is a common obstacle to the crossing of different forms. When it involves merely growing periods of unequal length the difficulty can be overcome easily by planting at such times that the various forms will flower simultaneously. When this is not feasible it becomes necessary to resort to some method of preserving the pollen. It has been found that pollen of certain species will retain vitality for weeks or even months if it is kept very dry. Miss Kellerman fefl I^^^^^K^^^H ^^H Fig. 147. — Implements used in castrating and depollinating hermaphrodite flowers. Right, self-closing forceps, ordinary forceps, scissors and scalpel. Left, chip blowers and syringes. {Courtesy U. S. Department of Agriculture.) reports that the most effective method tried by the Bureau of Plant Industry was as follows : anthers were placed in dried vacuum glass tubes, e.g., tube filled with anthers 1-2 inches, cotton 3-^ inch, exhausted to about 0.5 mm. pressure in the presence of sulfuric acid, the tube then sealed. As far as practicable the pollen was kept at a temperature of 10°C. A simpler and very useful method is to make a double container by fitting a small vial inside a larger one and partially filling the space between the two with anhydrous calcium chloride or sulfuric acid, filling in the upper portion with absorbent cotton and tightly corking the larger vial. The anthers or pollen grains are placed in the inner vial after it has been thoroughly sterilized and allowed to dry. (6) Failure of Fertilization. — This may be due to many causes ranging from simple morphological maladjustments to complex physiological relations amounting to antagonism. Probably a very frequent cause of unsuccessful crosses is failure of the pollen to germinate. When repeated HYBRIDIZATION 351 failures indicate that this may be the difficulty it will l)e worth while to try the application of a film of water or weak sugar solution to the surface of the stigma before pollination. By the aid of this simple device crosses have been secured between certain species of beans which had been repeatedly attempted without success. In this connection it may be well to give a word of warning. While it is always advisable to ascertain what one's predecessors have accomplished or failed to accomplish, the hybridizer should remember that both plants and local conditions are variable, and what may have been impossible at one place may be possible at another. Or the adoption of simple devices such as the water film on the stigma may be the determining factor. Much p(;rscverance is sometimes necessary. (c) Susce-ptihility to Mutilation. — Some plants are much more sensitive to mutilation than alfalfa. It appears that some are suceptible to merely removing the anthers from the ends of the filaments. In such cases it is necessary to resort to special methods for protecting the stigma from self-pollination. The details will depend upon the structure of the flower and whether it is protandrous or protogynous. Conditions favorable for hybridization may be summarized as follows : ideal conditions for flowering and fruiting; receptive stigmas; viable pollen ; morphological and physiological compatibility between pollen and pistil.; resistance of flowers to manipulations. Species hybridization is generally more apt to be attended by diffi- culties than is the crossing of varieties, although certain varieties of the same species have been found mutually incompatible in crossing. In general crosses are most successful when made between closely related species. The reason for this is clear when the genotypic differences between distinct species are considered as differences between homol- ogous factors, i.e., factors which condition similar characters as was explained in Chapter XII. It is possible that in very closely related species the factors conditioning similar morphological and physiological characters are themselves similar, if not in a specific sense at least in terms of the whole reaction system. The new combinations of these similar systems of factors which would be formed in Fi hybrids, would be compatible with the vital functioning of the zygote including the produc- tion of viable gametes. In widely separated forms, on the other hand, the reaction systems must be very different, thus causing corresponding reduction in the chances of favorable combinations among the hybrid zygotes. While it is impossible to judge with certainty of the possibilities of species crosses by somatic resemblances and differences, yet the taxonomic relationships of forms it is proposed to hybridize serve as a general guide in forming such estimates. No hybrids between different plant families are known and few authentic cases of intergeneric crosses 352 GENETICS IN RELATION TO AGRICULTURE have been reported. While many first generation crosses between different species are more vigorous than either parent, others are known to be exceedingly weak. Unless repeating crosses which have already been made, the hybridizer of species is exploring the unknown and there is always the possibility that his results may be of interest to science as well as of practical value. The Svalof Methad of Creating Populations. — Progress in plant im- provement by means of hybridization experiments will always be limited by the available supply of experts as well as by facilities and time. Any method, therefore, that enables the breeder to secure desirable new combinations of parental characters without the enormous amount of detail involved in a system of pedigree culture, is worthy of serious consideration. Such a method was devised by Nilsson-Ehle and has been used at Svalof with success. According to Newman, "two known sorts are crossed and the whole progeny from all second and succeeding gen- erations is sown together en masse. The object of this plan is to allow the severe conditions of winter and early spring to either destroy or expose the weaknesses of as many of the more delicate combinations as possible. In the latter case the breeder is given an opportunity to assist nature in her work of elimination by practising a form of mass-selection. While there is thus effected in a very simple manner a gradual weeding out of a great mass of unfit combinations, the progeny of a crossing at the same time gradually assumes the character of an ordinary mixed population, the different combnations becoming automatically constant as time passes. The advantages of working with constant forms will be appreciated by all breeders as will also the fact that through the above arrangement the number of combinations which may arise through the repeated segregation of inconstant forms in each succeeding generation will have increased immensely. . . . While the above system requires a considerable length of time before any definite results can be reached, yet it requires very little work until the time comes to make selections. Numerous crossings of this kind may therefore be carried forward with the regular work and thus provide a constant source of new material." CHAPTER XX UTILIZATION OF HYBRIDS IN PLANT BREEDING Although the special uses to which plant hybrids may be put are very numerous, they fall into two categories, viz., first, the production of new desirable combinations and, second, the production of increased vigor in the first hybrid generation. The first category includes all phases of the usual purpose of crossing plants, which was briefly discussed in the preceding chapter. The new character combinations desired may be exclusively morphological or physiological or, as is more often the case, they may represent combinations of both kinds of characters involving many factors. In the simpler cases involving only a few pairs of inde- pendent factors the breeder who is familiar with the Mendelian princi- ples of heredity can easily compute the number of Fo individuals that he must grow in order to secure the desired combination. Even in the most complex cases knowledge of the principles of genetics will be of practical value in helping the breeder to understand his results in Fi, F2 and later generations and in guiding his selection of F2 individuals for further test- ing. These principles are discussed in Chapters V to X. It is the purpose of this chapter to present some specific results of the increased vigor so commonly observed in Fi hybrids. This increased vigor, or heterosis, as it has been termed by Shull, may manifest itself in greater size, more rapid growth, larger productivity, greater hardiness, drouth resistance, etc. The theoretical explanations of heterosis have been dis- cussed in Chapter XII. In the present chapter we shall consider only the utilization of the principle that hybridization of closely related varieties or species usually results in heterosis. As the methods used with plants grown from seed differ from those which can be used with vegetatively propagated plants, the two groups will be considered separately. Increased Production in Fi Maize Hybrids. — This phase of corn breeding has come into considerable prominence in recent years. Although it has not yet become an important factor in corn growing, it presents interesting and important possibilities in the way of increased production. The most significant results have been obtained by growing Fi hybrids between species, sub-species, commercial varieties, local strains of commercial varieties and closely inbred strains or biotypes. The earliest recorded experiments on increased production are stated by 23 353 354 GENETICS IN RELATION TO AGRICULTURE ^, Collins to be those of Beal (Michigan, 1878-1882), Ingersoll (Indiana, 1881), Sanborn (Maine, 1889), and of Morrow and Gardner (Illinois, 1892). All of these crosses were made between commercial varieties and in each case the hybrids outyielded one or both parents. Then came the work of Shull and of East (1908) with inbred strains and the crosses between them, both investigators obtaining an increase in yield in the hybrids over that of the original stock. Following this the United States Depart- ment of Agricultm^e conducted experiments on an increasingly extensive scale and included work with the most distinct types as well as commejcial varieties and inbred strains. More recently various experiment stations have conducted similar investigations. Crossing inbred strains or hiotypes produces the most striking results because the rate of increase in vigor in the Fi hybrids over the inbred strains is enormous (as much as 250 per cent, over the average of the parents) . Of course it is much greater in some cases than in others be- cause of the inherent differences between different bio types. East worked with biotypes of four different varieties and secured an average increase of 73 per cent, in all crosses. The data on inbreeding the Leaming dent variety are summarized by East and Hayes in Table LI. It will be noted that two of the strains were not grown as second inbred generations Table LI. — Effect of Inbreeding in Strains of Leaming Dent Maize Yield in bushels of shelled corn per acre and years in which grown [After East and Hayes) Parent Strain number Generations inbred variety 1 2 3 4 5 6 6 59.1 (1906) 95.2 (1908) 57.9 (1909) 80.0 (1910) 27.7 (1911) 88.0 (1905) 7 60.9 (1906) 59.3 (1907) ■ 46.0 (1908) 59.7 (1909) 63.2 (1910) 68.1 (1910) 25.4 (1911) 41.3 (1911) 9 42.3 (1906) 51.7 (1908) 35.4 (1909) 47.7 (1910) 26.0 (1911) 12 38.1 (1906) 32.8 (1907) 46.2 (1908) • 23.3 (1909) < 28.7 (1909) 16.5 (1910) 9.5 (1910) 2.0 (1911) 2.0 (1911) UTILIZATION OF HYBRIDS IN PLANT BREEDING 355 until 1908 and in that year "the general environmental conditions were much above normal. For opposite reasons, poor soil and badly distributed rainfall, the yields of 1909 are somewhat too low and the yields of 1911 are very much too low." With these facts in mind an examination of the table shows that the strains became more and more differentiated as to yield as inbreeding progressed. "The first strain, No. 6, is a re- markably good variety of corn even after five generations of inl)reeding. It yielded 80 l)ushels per acre in 1910. ... In the field, even in 1911, the plants were uniformly vigorous and healthy and were especially remarkable for their low variability. The poorest strain. No. 12, is partially sterile, never fills out at the tip of the car and can hardly Fig. 148. — Inbred strains of Learning dent corn compared with Fi and F2 hybrid gener- ations. The yields per acre were as follows: No. 9 (at the left) 47.7 bu.; No. 12, IG.Gbu.; (12 X Q)Fi, 117.5 bu.; (12 X 9) i^s, 91 .5 bu. {After Ea>it and Hayes.) exist alone. . . . When two of these inbred strains are again crossed, the Fx generation shows an immediate return to normal vigor. The plants are earlier and taller, and there is a greater total amount of dry matter per plant. For example, in 1911 the average height of all the strains of inbred Leaming dent was 84 inches while the average height of the 16 hybrid combinations was 111 inches and the height of the shortest hybrid combination was 1 foot greater than that of the tallest inbred strain." In general it seems that the combinations into which strain No. 7 was introduced were the best while those in which the poorest strain. No. 12, was used were the poorest. However, a cross between these two strains in 1911 yielded 60.2 bushels per acre. The F^ genera- tion from a number of the crosses was grown and in every case there was a decided falling off in production. This would be expected as a matter of course under conditions of random mating in F\ inasmuch as some homozygous combinations would be formed among the F^. zygotes. Fig. 148 shows types of ears and comparative yields in strain No. 9 after 4 generations of inbreeding, strain No. 12 in the fifth inbred generation, and the Fi and F^ hybrids, all grown in 1910. 356 GENETICS IN RELATION TO AGRICULTURE Theoretically the maintenance of superior near-homozygous strains and annual crossing of the best for production of Fi seed for sale to growers is a practicable method of corn breeding. This plan was first suggested by Shull. It is certainly a desirable method not only because of the high degree of heterozygosity produced on crossing such strains, but because continuous inbreeding has a similar effect to growing the plants under adverse conditions. It tends to eliminate all but the strongest individuals and is thus an effective method of selection. However, as a more prac- ticable method, East suggested that combinations of the various com- mercial varieties be tested until the most profitable combination is found. There has been considerable investigation of both methods, but it is impossible at present to say which will be used more extensively. One of the most valuable features of this method of inbreeding followed by crossing of superior strains, as compared with ordinary ear-to-row selection, is the saving in time. For example, in the production of high-yielding strains of corn which differ in chemical composition of the grains, Emerson and East point out that ear-to-row selection from open pol- linated plants will, if carried on long enough, produce a strain of the desired type. It will be sufficiently homozj^gous to insure comparative constancy as regards oil, protein or starch content. At the same time a sufficient number of factors for other minor characters will be hetero- zygous to insure a fairly vigorous strain. But, on the other hand, by self-pollination, together with the same sort of selection, several practically homozygous strains of the desired type, high oil for instance, "could almost surely have been produced in much less time." These strains would doubtless have been unlike for many other characters, so that if degree of vigor is dependent upon degree of heterozygosity, the crosses between them would doubtless have been abundantly vigorous. Or if physiological vigor is conditioned by specific factors, then crosses between some of the selected strains vvould doubtless have effected the most favorable combinations for maximum vigor. In either case the result is the same. "While a few years' time may not be an important consideration where the character in question can be determined at sight, or by mere weighing or measuring, in breeding work requiring costly chemical analysis it is extremely important that the desired re- sults be obtained in as few years and, therefore, with as few analyses as possible.'"' Method of Comparing Yields. — The importance of accuracy and fair- ness in comparing the yields of Fi hybrids with their parents has been determined by Collins. We give his conclusions verbatim: "So large a proportion of first-generation maize hybrids have been found to give increased yields and the increase is frequently of such magnitude that the utilization of this factor of productiveness becomes a practical question. It is, UTILIZATION OF HYBRIDS IN PLANT BREEDING 357 therefore, highly desii-able to uiulerstand the reasons why some crosses give favorable results and others give little or no increase over the yield of the parents. A necessary step in this direction is to develop a reliable method of measuring the effect of crossing, apart from other factors that influence yield. "The development of satisfactory methods of comparing the yield of first- generation hybrids with that of their parents has been retarded by (1) a failure to fully appreciate the importance of individual diversity in hybrids, (2) the abnormal behavior of self-pollinated maize plants, and (3) the difficulty of securing for comparison hybrids and parents with identical ancestry. It is believed that the method here described avoids these difficulties and affords more accurate means of comparing first-generation maize hybrids with their parents. "The method is illustrated by an experiment in crossing two varieties of sweet corn in which it was found that the progenj' from one hybrid ear yielded nearly double that of the other hybrid ear involved in the experiment. To have taken either ear alone would have led to entirely erroneous conclusions regarding the increase secured as a result of crossing. The increase in yield due to crossing as measured by the method here proposed was 31 per cent." Collins describes his method as follows: "To compare the behavior of two varieties, which may be called A and B, with that of a hybrid between them, two plants were selected in each variety, Ai and Ao in the one variety and B^ and B^ in the other variety. The following hand pollinations were made: Ai X A2, A2 X Bi, Bi X B-i, and B2 X Ai. The result is two hybrid ears and one cross-pollinated ear of each variety. It is believed that the mean yield produced by seed from the two pure seed ears gives a fair measure of the effects of hybridization. By making two hybrids involving all the plants used in producing the pure seed ears individual differences that affect the jdelding power of the pure seed ears are similarly represented in the hybrids. Thus, in both the parents and the hybrids the average yield represents the mean yielding power of the four parent plants, the only difference being the way in which the individuals are combined. "To secure the most accurate comparison of the jdeld of the four ears, one seed from each of the ears was planted in each hill. The different kinds were identified by their relative position in the hill. To place the seeds accurately, a board 4 inches square was provided with a small pointed peg 2 inches long at each corner. These pegs were forced into the soil at each hill, making four holes, one for each of the four kinds, only one seed being p'anted in a hole. The board was a'ways placed with two sides of the board parallel to the row. It was necessary to exercise extreme care in dropping the seeds to avoid changing the position of the kinds. The best way to obviate mistakes of this kind is to make all the holes of a row in advance and to go down the row with one kind of seed at a time. "At harvest time the seed produced by each plant was weighed and recorded separately. All hills that lacked one or more plants were excluded and the com- parison confined to hills in which all four kinds were represented. The method of handling the yields was to determine the mean yield of the four kinds in each 358 GENETICS IN RELATION TO AGRICULTURE hill and to state the yield of each of the four plants as a percentage of the mean of the hill in which it grew. The percentage standing of each kind in all the hills was then averaged to secure the final expression of the relative behavior of the four kinds. "This method of comparison is similar to the ingenious plan originated by C. H. Kyle, for use in ear-to-row breeding. Kyle's method is to plant each of the ears to be tested in a separate row and in each hill to plant one seed of a stand- ard, or check, ear with which all ears are compared. Since comparative and not absolute yields are desired in the study of hybrids and with only four kinds to compare, the introduction of a check in the present experiment would have increased the space occupied by the experiment without lessening the experi- mental error." Fig. 149. — Parents and Fi hybrid between two sub-species of Zea mays: Hall's Tyler dent (left), Brewer's flint (right) and hybrid (center). The hybrid yielded 9 per cent, more shelled corn than the dent and 20 per cent, more than the flint and proved the most productive of all varieties and crosses in the 1913 test. {After Hayes, Conn. A. E. S.) Crossing Species, Sub-species, Varieties and Local Strains.^ — Many experiments have been made to test the increase in productivity of F\ hybrids between more or less closely related forms of maize. As it is impossible to review them all, we give as an illustration Collins' summary of the results of 16 crosses made in 1908 between corns of diverse types and from widely separated localities. The classification indicated by Collins' descriptions are as follows: Zea mays indentata (starchy or dent varieties)^ — Maryland, Kansas dent, Brownsville, Chihuahua, Mexican dent, Xupha (semi-starch); Zea mays ainylacea (floury variety) — Tuscarora; Zea mays everta (pop) — Cinquantino, Algerian, Tom Thumb; Zea m.ays indurata (flint) — Guatemala red, Salvador Black; Zea hirta Bonafous — Hairy Mexican, Huamamantla, Arribefio; Unclassified — Hopi, Chinese (waxy endosperm), Quezaltenango Black, Quarentano. The yields of the 16 crosses and of their parents are given in Table LII. UTIUZATION OF HYBRIDS IN PLANT HREEDlNd 359 Table LII. — Yields of 1G Maize Crosses Compahed with Parental Yields. {After Collins.) Name of liyhrid Ahz, Maryland dent by Hopi Ah>, Tuscarora by Cinquantino Dhi, Kansas dent by Chinese Dhi, Chinese by Chihuahua Dhz, Hopi by Chinese Dhi, Chinese by Xupha Brownsville by Chinese Hopi by Algerian pop Tom Thumb by Quezaltenango black. Brownsville by Guatemala red Guatemala red by Salvador black . . Quarentano by Brownsville Huamamantla by Hairy Mexican . . Arribeno by Hairy Mexican Hairy Mexican by Chinese Mexican dent by Tom Thumb Average percentage of increase of hy- brids over average parents . Dh,, Eh I, Ghi, Khn, Khi2, Mhn, Mhu, Mhi6, Mhn, Mho,, Yield of female parent, pouads 1.19 0.53 0.99 0.39 0.74 0.39 0.77 0.74 0.10 0.77 0.31 0.27 0.40 0.39 0.18 0.52 Yield of male parent, pound 0.74 0.24 0.39 0.69 0.39 0.63 0.39 0.34 0.10 0.31 0.27 0.77 0.18 0.18 0.39 0.10 Average yield of parents, pound 0.965 0.385 0.690 0.540 0.565 0.510 0.580 0.540 0.100 0.540 0.290 0.520 0.290 0.285 0.285 0.310 Yield of hybrid, pounds 1.25 0.75 1.09 0.95 1.28 0.54 1.16 0.91 0.42 0.49 0.33 0.48 0.31 0.47 0.61 0.54 PercentaRe of increase of hybrid over average of parents, per cent. 29 95 58 76 126 6 100 69 (a) -9 14 -8 7 65 114 (a) 53 (a) Where the yield of either parent fell as low as 0.10 pound per plant the percent- age of increase of the hybrid is omitted. In dealing with these small quantities it is believed that percentages would be misleading. The superior qualities of first-generation hj'brids in maize as set forth by ColHns may be summarized as follows: (1) Increased yield. (2) Uniformity equal to that of the parents. (3) Quality intermediate between parents (but Hayes' data indicate complete dominance of low protein over high protein). (4) Increased immunity from disease. (5) Extension of the industry into new territory. Especially strong evidence for this is found in several of the crosses between diverse tj^pes. "Almost without regard to the nature of the parents the hybrids remained dark green and vigorous when nearly all the pure strains were giving evidence of the lack of moisture by their curved leaves and yellow color." (6) Less localisation of highly bred strains. The importance of local adjustment in highly bred strains is the chief reason for the disappointment which sweet corn growers experience when they purchase carefully selected strains from other localities. "First-generation hybrids are to a great extent independent of this delicate adjustment to local conditions." (7) Increased utilization of the work of experienced 360 GENETICS IN RELATION TO AGRICULTURE breeders. (8) Stimulus to the work of improvement through the possi- biUty of protecting new productions. More recently Jones and Hayes have made extensive experiments in crossing commercial varieties of corn upon which they report as follows: "Fifty first generation corn crosses have been compared with their parents. Eighty-eight per cent, yielded more than the average and of these 66 per cent, yielded more than either parent. "In time of ripening the first generation crosses were on the average interme- diate when compared with their parents. Thus in crosses between varieties differing widely in time of ripening the first generation crosses not only yielded more than the late parent but matured considerably earlier. This increase in the rate of growth is considered to be fully as important under Connecticut conditions as any increase in yield. "The highest yielding parents gave the highest yielding crosses as would be expected, but a rather unexpected result was obtained in that there was ap- parently no relation between the yield of the parents and the increase in the yield of the cross. High average yielding parents gave as large increases, when stated in per cent., as low yielding parents. "There was a tendency for the crosses whose parents differed in their ability to yield to give the greatest increase. This is also shown by the fact that the dent X flint crosses gave greater increases in growth than the flint x flint crosses. "These facts bear out the assumption that hybrid vigor is not the result of an indefinite physiological stimulation but merely the result of the bringing to- gether of greatest number of favorable growth factors. Crosses between va- rieties of diverse type therefore possess a greater total number of favorable growth factors than crosses between similar varieties and hence give larger in- creases when crossed." The immediate effect of crossing upon size of the grain and hence on yield should not be confused with the increased production of hybrid plants. There is a popular belief that by planting two varieties in alternate rows the yield will be increased. That this idea is supported by scientific evidence was indicated by the earlier work of Correns, Carrier, and Roberts, but it remained for Collins and Kempton to secure the proof of this important fact. These investigators used the ingenious method of pollinating various white seeded varieties with a mixture of their own pollen and pollen from some variety having colored seeds. By taking advantage of the phenomenon of xenia they were able to make direct comparison of the selfed and the hybrid grains from the same ears. The possible invalidity of their results due to more rapid develop- ment of hybrid grains and consequent repression of selfed grains was removed through the fortunate discovery of an ear that had been twice pollinated, first with its own pollen and a week later with pollen from a colored variety. "All the white kernels were on the lower portion of the ear, all the colored were on the upper portion. Obviously the hybrid UTILIZATION OF HYBRIDS IN PLANT BREEDING 361 seed could have no advantage in this case. The ear produced 212 white, or pure, seeds and 161 that were yellow, or hybrid. The average weight of the pure seed was 283 gm. per 1000 kernels. The average weight of the hybrid seed was 292.5 gm. per 1000, a difference of 9.5 ± 1.06 gm., or 3.4 per cent." In the experiment itself eleven ears, involving five different varieties, were crossed, giving a total of 1,658 hybrid seeds to be compared with 3,513 selfed or pure seeds. In every instance the size of the seed was materially increased by the foreign pollen, the increase ranging from 2.8 to 21.1 per cent. The fact that the pericarp of the mother plant is not strictly a part of the seed but is of purely maternal origin might seem difficult to harmonize with the jiesults, but Collins and Kempton point out that the necessary increase in size of the pericarp would be comparatively slight and to seek any explanation may be superfluous. The practical value of this evidence is great. As the authors state, "the results afford additional reason for the use of first generation hybrid seed; but even where hybrid seed is not to be used, the planting of two varieties in alternate rows may be found to increase the yields sufficiently to warrant the additional trouble." And further, "as the increased size is evidently a manifestation of vigor, it may be considered as a factor of adaptation, like the vigor of the first-generation hybrid plants. It would seem especially desirable to take advantage of this method of increasing yield in regions which do not produce their own seed corn." Centralized Seed Corn Production. — Carefully selected strains of maize are liable to prove disappointing when grown under conditions different from those obtaining at the locations where they are produced. But the work of intensive selection requires considerable skill and ex- perience and the farmer can seldom attend to it properly. He should obtain his selected seed corn from a local breeder if possible. The fact that 7^1 hybrids in maize are comparatively resistant to local and seasonal conditions which prove detrimental to pure strains indicates that such hybrids may be produced at central points in quite a large territory. When it is known which combination of varieties, or of pure strains of a single variety, is best adapted in certain localities, pure seed of these varieties or strains may be maintained and the crosses made under expert supervision at a central seed farm. On the other hand, a farmer who wishes to produce his own hybrid seed need not hesitate on account of increased cost of production. Collins has shown that even though the cost of raising hybrid seed be double that of ordinary seed, yet "where increases ranging from 5 to 50 per cent, may be expected there are few farm operations that yield such large returns." A Method of Producing Hybrid Corn Seed.— A grower intending to produce his own hybrid seed each year might do well by beginning 362 GENETICS IN RELATION TO AGRICULTURE with a series of trials with varieties in alternate rows. After determining which varieties are best adapted to the local conditions and give the best results when crossed he will be ready to adopt a simple system of hybrid seed production somewhat like the following. These directions have been sent out by the United States Department of Agriculture to cooperative experimenters. Various other plans could be devised. Experiments as outlined below involve the use of two varieties and two separate plots. Varieties may be designated as No. 1 and No. 2, the plots as A and B. The plots should be sufficiently separated to prevent cross-poUination between them. It should be kept in mind that the increased yield can be expected only for the one year immediately following that in which the cross is made. Plot A is planted with alternate rows of No. 1 and No. 2. The rows planted with No. 2 are to have all plants detasseled. The crop of No. 1 and No. 2 is to be saved separately. Plot B is planted entirely with variety No. 2 and has alternate rows detasseled The crop from the tasseled and detasseled rows is to be saved separately. At harvesting there will be the following lots of seed : 1. Plot A. Variety No. 1, field-pollinated. 2. Plot A. Hybrid between No. 1 and No. 2. 3. Plot B. Variety No. 2, field-pollinated. 4. Plot B. Variety No. 2, cross-pollinated. The yields in the year the cross is made should show the comparative value of the two varieties and the effect, if any, of detasseling on the immediate yield. A comparison of the yield from these four lots of seed the following year should show the yield of the first-generation hybrid as compared with the pure varieties and to what extent the increase, if any, is due to the elimination of self -pollinated seed. If plot B cannot be provided, seed of variety No. 2 should be held for planting the following year in comparison with variety No. 1 and the hybrid seed. Application in Other Annual Crop Plants. — The increased vigor due to heterozygosis has not yet been utilized in a practical way in annual crops other than corn. Melons and other cucurbits are monoecious, easily crossed under proper conditions and within a single species, and large quantities of seed are produced. Very little is known concerning the value of Fi hybrids between varieties as compared with parents, but Hayes and Jones report preliminary experiments which indicate that first generation cucumber crosses may frequently be expected to exceed the higher yielding parent in yield. Only one out of four different crosses failed to exceed the average of the parents in any character by an appreciable amount. In tomato growing Wellington has shown that crossed seed is worth its production as based on the increased value of a single crop without ref- erence to origin of new varieties. He states that, while desirable results have been obtained by crossing plants indiscriminately, "better results would undoubtedly have been obtained if high-yielding mothers had been selected for one or two generations previous to the first crossing. " Toma- toes are normally self -fertilized and, the high-yielding strains or pure Unes UTILIZATION OF HYBRIDS IN PLANT BREEDING 3G3 having been isolated, they can be maintained and the crosses may be repeated from time to time. This is a very important consideration for the grower who wishes to put the same grade of protkict on the market from year to year. "As tomato seed remains fertile from 3 to 7 years, a grower does not need to make his crosses oftener than once in 3 years. The seedsman, as well as the farmer, can profitably raise Fi generation seed, provided a guarantee is not given for more than one generation, for the buyer, to maintain his quality of product, will have to purchase seed every year.'' Welhngton thinks the best results with tomatoes can be obtained by keeping within a species and crossing distinct varieties or strains. Dominant characters, that will certainly appear in the fruits of Fi plants if present in either parent are rough or irregular shape, dark red color as contrasted with pink or yellow and pink as contrasted with yellow. Size and season of ripening in Fi will be intermediate between the parental characters. Jones and Hayes report results of similar experiments which corroljo- rate Wellington's conclusions. Of two different crosses one (Stone x Dwarf Champion) gave an appreciable increase in both size and number of fruits and the total yield was thereby increased. It even exceeded the better parent by 15 per cent. Moreover, the increase above the latter parent was uniform throughout the four years of the test. The other cross (Lorillard x Best of All) exceeded slightly the better parent in average weight of fruit but it did not excel in total yield. "These re- sults show that not all combinations of tomato varieties give the vigor usually derived from crossing, but when a desirable combination is found it can be counted on to give the increase in yield every time the cross is made. Vigor due to crossing as measured by increased yield was not appreciably greater in crosses between artificially selfed strains than in crosses between ordinary commercial varieties. These results are in agreement with the fact that the tomato is naturally almost completely self-fertilized. The cross of Stone x Dwarf Champion which gave a significant increase in yield also showed a hastening of the time of pro- duction. It not only gave a 15 per cent, larger yield than the later parental variety but was earlier in its time of production than the earlier parent. Hence its value to market gardeners was increased." Similarly, in tobacco, Selby and Houser claim that the culture of first- generation hybrids will prove both profitable and practicable. Since the added cost of producing hybrid seed should not exceed 50 cents per acre and the crossing need not be repeated oftener than once in 3 j-ears, the financial consideration is neghgible. In regard to uniformity of crop they find that Fi hybrids between pure varieties or fixed hybrids show no essential difference in uniformity from the parent varieties and for com- mercial purposes only such parents should be used. As for yield their 364 GENETICS IN RELATION TO AGRICULTURE results in 1909 showed the average of the hybrids to be about 185 pounds more per acre than that of their parents. The maximum increase obtained was 492 pounds per acre. By selecting seed from the highest- yielding F2 plants it is possible to produce even higher yields in F3 and Fi. But such high-yielding selections are not fixed and under con- ditions of commercial culture the yield and uniformity would undoubt- edly decrease rapidly. It appears that the growing of Fi hybrids offers the one chance of commercial production of the highest possible yields combined with uniformity in size and shape of leaf. The matter of quality of cured leaf is more difficult of solution since this is a complex character and is easily affected by environmental conditions. Until further investigations have been made it seems that the only safe pro- cedure is to choose as parents only varieties or strains that produce leaf of high quality. Application in Vegetatively Propagated Plants. — In this class of plants the stimulus due to heterozygosis has been extensively utilized, but this has been the result of the method of propagation rather than the conscious use of the principle. In potatoes and strawberries, for example, the commercial varieties are all hybrids. The crosses having been made, the best plant of the first generation became the source of a new variety. There are many opportunities for further application of this principle in the bush and tree fruits, not only for vigor but for excellence of quahty as well; also in asparagus, rhubarb, hemp, hops, pineapples, sugar cane, etc. It is thought by some horticulturists that the greatest possible im- provement in fruits can only be secured by preparing for hybridization by several generations of inbreeding. Thus Jones, proceeding on the assumption that increase of vigor in hybrids is due to heterozygosis, recommends the general adoption of inbreeding in order to secure homo- zygous strains which can then be utilized in the production of the most vigorous Fi hybrids. But it is to be remembered that only a portion of the homozygous strains could be expected to produce superior Fi hy- brids. It is, therefore, a serious question whether this method would be as economical in the long run as the crossing of existing varieties. The use of Fi hybrids as rootstocks for vines, tree fruits and nuts is of recognized importance. The Royal and Paradox Walnuts, which were named by Burbank from specimens which he produced, furnish a striking illustration. The Royal type of hybrid is produced by crossing the Black Walnut of eastern states {Juglans nigra) with the California Black Walnut (J. calif ornica) ; while the Paradox type comes from crossing the walnut of commerce (/. regia) with either of the above named black walnuts. Hybrid seedlings commonly appear in the seed beds planted with seed from trees standing near trees of other species. As they are UTILIZATION OF HYBRIDS IN PLANT BREEDING 365 easily distinguished by their larger size while still quite small, all that the nurseryman has to do is to select the hybrids for budding or grafting. Sterility or partial sterility is frequently associated with increased vigor in first-generation hybrids between species. The large flowers and luxuriant growth of some of the sterile tobacco hybrids render them promising subjects for use as ornamentals. Partial sterility, when mani- fested by a lessened production of seed, may not be accompanied by any decrease in yield of fruits. In such cases therefore it is a positive advantage if the plant can b(> propagated vegetatively. Rapid-growing timber and ornamental trees of a number of dilfcrent species have been produced by crossing distinct forms. Henry men- tions the following valuable trees which, on account of their vigor, botanical characters and non-occurrence in the wild state, are presumably first-generation hybrids: black Italian poplar, London plane, Huntingdon elm, cricket-bat willow and the common lime {Tilia vulgaris). Accord- ing to Henry the pioneer work on hybridization of trees was done by Klotzsch at Berhn in 1845. He crossed two species each of pine, oak, elm and alder. He "claimed that by hybridization, both the rapidity of growth and the durability of timber of forest trees could be augmented considerably; but no further experiments were made, and his pioneer work fell into oblivion. " The art of breeding trees was renewed by Burbank's work with the walnuts about 1890. Henry reports results with Fi hybrids in Populus, Fraxinus, Alnus, Ulmus and Larix. He points out that one of his most vigorous hybrids (Popidus generosa) was "derived from two parents so little related that they are placed in two distinct sections of the genus." At the same time, "a cross between two races of the common alder shows considerable vigor, though the parents are so closely allied that they can only be distinguished by the most trivial characters." Thus it appears that prediction as to the outcome of species crosses in trees is quite as impossible as in other classes of plants. There is great need for further experimentation. In planting wind-polhnated species provision can easily be made for natural hybridization by mixing groups of different species. It has been found that the quality of the timber in rapid growing Fy hybrids is equal or superior to that of the parents. Increased resistance of F\ hybrid plants to insect pests and diseases is doubtless often merely another manifestation of their increased vigor. But in this connection it is to be remembered that disease resistance is generally a heritable character, so that in a particular instance its ap- pearance in F\, will depend on the factorial composition of the parents and the relation of the factors in inheritance. CHAPTER XXI MUTATIONS IN PLANT BREEDING Discontinuous heritable variations have appeared very frequently in cultivated plants under conditions such that they could not be attributed to hybridization. The selection of these variations has produced new varieties in the same way that the early color varieties of the sweet pea arose. By means of breeding experiments many such variations have been proved to follow the Mendehan principles of inheritance. The general conformity of varietal crosses with the Mendelian principles is sufficient reason for asserting that the vast majority of cultivated varieties arise either directly or indirectly through factor mutations. A few have originated through chromosome aberrations, but, so far as is known, none which are of importance to agriculture. It is especially clear that in self-fertilized species the production of new varieties from single plant selections is made possible by the occurrence of factor mutations. The successes of Le Couteur, Shirreff and Hallet, and the achievements of Vilmorin, Nilsson, Hays and Johannsen with individual plant selections find their explanation in the existence of genetically diverse forms within the species or varieties with which they worked. These pure lines must have originated through changes in specific factors. Similarly with the genotypes of maize and other cross-fertilized species, although new combinations of factors are continually arising through natural inter- crossing, yet entirely new factors can arise only by means of changes in existing factors. These factor mutations do not necessarily induce pro- found somatic changes, and slight morphological variations may hardly be distinguished from modifications due to environment. When physio- logical characters alone are affected, as is sometimes the case, the most careful tests of many individuals may be required to discover a desirable mutation. But once a mutation arises, the normal range of fluctuation in the character or characters affected is different from that of the parent form, and new material has been provided for man's selection if he desires to use it and can isolate it. When these facts are realized the funda- mental importance of mutations to breeding will be appreciated. Bud mutations, especially when strikingly different from the parental type, have long been known. Bailey states that Carriere in 1856 enu- merated over 150 bud-varieties or sports of commercial importance in France and he estimated that no fewer than 300 named horticultural 366 MUTATIONS IN PLANT liREEDING 307 varieties grown in this country in 1895 had a similar origin. There is no reason to suppose the number has decreased and it is probably larger. There is good evidence (see Chapter XIV) to show that bud sports arise through factor mutations and that they occur in as great diversity as do seed sports. Sometimes striking morphological or substantive changes are produced but probably the somatic effect is often slight and hence not easily detected (Chapter XXIII). Mutations in Crop Plants. — Johannsen has reported two mutations in his pure linos of beans. The careful statistical analysis of his successive pure line families revealed the first mutant in 1903 and Johannsen thinks it appeared as a bud sport. It was characterized by its large size and relatively narrow shape. As it was constant from the first it must have originated in homozygous condition. The second mutant bore seeds which were relatively broad in shape. It could be traced back to 1907 when it existed in heterozygous condition. Later it was obtained in pure line. Very recently mutations of great commercial value occurred in the Florida Velvet Bean, Mucuna utilis. The old variety was limited to Florida and the Gulf Coast on account of lateness. About 1 ,000,000 acres wei'e grown in 1915. We are informed by Piper that early varieties originated by mutation at at least three different places, the first in 1906. These resulted iii the crop being adapted to the entire cotton belt and in a very rapid increase in acreage since 1915. In 1916 about 2,650,000 acres were grown and in 1917, about 6,000,000 acres. Hayes describes a number of mutations in tobacco which is normally self-fertilized. The first was found in a homozygous strain of the Connecticut Cuban shade variety of commercial tobacco {N. tahacum). This strain bears from 14 to 25 leaves per plant, the mean number for 1910 and again for 1914 being 19.9 leaves. In 1912 the Windsor Tobacco Growers Corporation grew about 100 acres of this strain and during the clearing of the field three plants were found that had not yet bloomed and which bore a number of uncut leaves. One of these was transplanted to the greenhouse of the Connecticut Agri- cultural Experiment Station. It produced 72 leaves on the main stem and blossomed about January .first. All the seedlings grown from this plant came true to the new type which differs from the parent strain "in having leaves of a somewhat lighter green shade, in a partial absence of basal suckers, and in a practically indeterminate growth " (see Fig. 150) . The quality of leaf seems as good as the Cuban and an increased yield per acre of approximately 90 per cent, has been obtained, but it is yet too soon to know how satisfactorily the new variety will conform to trade requirements. Several similar mutants have been found in plan- tations of the Connecticut Havana variet5^ This variety has been grown in Connecticut for over 50 years and is uniform in habit of growth. 368 GENETICS IN RELATION TO AGRICULTURE On one of the farms the same mutation has recurred several times. Hayes beheves that these mutations cannot be explained as the result of accidental crosses. For in the large series of crosses that have been Fig. 150. — The Stewart Cuban variety of tobacco, a very promising mutation. Plants from seed sown under glass in December and transplanted to the open in May were twelve to fourteen feet tall in September and had produced eighty leaves per plant. {From the Journal of Heredity.) made in the Connecticut station in no case have new forms exhibiting this tendency to indeterminate growth been obtained. Nilsson-Ehle discovered that in pure lines of oats occasional grains appear that are aberrant either in color or in morphological characters. The variations tested by him either bred true at once or after one or two generations practically all of the progeny would breed true for the MUTATIONS IN PLANT BREEDING 369 new characters. Surface and Zinn consider this sufficient evidence to make it "ahnost certain that similarly inherited variations may occur in respect to phj^siological characters such as yield." That they were justified in making this inference is shown by the success of their ex- periments on pure line selections for yield. Other self-fertilized crop plants in which mutations have been reported are barley, wheat, tomato and potato. In maize the sudden appearance of new characters in established vari- eties or strains has been reported by a number of investigators. The remarkable diversity between inbred strains as discovered by Shull and by East indicates the extent to which germinal variations occui' in this plant. Each author obtained one strain which was so nearly sterile as to be in danger of complete extinction while other strains appeared to be capable of maintaining fairly good annual yields indefinitely. Abundant evidence of the occurrence of factor mutations in maize is also found in the numerous pairs of contrasted characters which are inherited in Mcndelian fashion. Besides those referred to in earlier chapters we may mention multiple stems (suckers) as dominant over single stems (no suckers), normal stature (tall) as dominant (usually) over dwarf stature, normal green leaves as dominant to striped leaves, presence and absence of aerial roots, hairy and glabrous stems, branched and unbranched tassel, normal ear and branched ear, normal anthers and fasciated anthers, presence of normal reproductive organs and absence of the same (barrenness). Most of these allelomorphs behave as unit characters. Various quantitative differences such as stature, ear-length and number of rows to the ear are either conditioned by several pairs of factors or by single pairs of factors which are subject to a wide range of variability in expression. Con- stitutional vigor and productivity are doubtless conditioned by the in- teraction of very many factors and a mutation in a single one would alter the end result. In short the inherent individuahty of corn plants, which makes possible the successful appHcation of selection methods, must be referred to factor mutations. Sugar beet improvement, particularly increase in sugar content of the roots, depends directly upon the occurrence and selection of mutations, according to Pritchard. As a result of statistical investiga- tions of variation, correlation, inheritance and selection in the sugar beet, he concludes that although sugar beet improvement has been ac- complished, continuous selection is not necessarily the determinmg factor in attaining the present high sugar content of the best varieties. His statistics show that the best roots transmit no better qualities than do the mediocre roots because the differences are merely "fluctua- tions" (modifications). The real differences between sugar beet families are usually very slight and are greatly exceeded by their "fluctuations." 24 370 GENETICS IN RELATION TO AGRICULTURE Both the best and the poorest families transmit average qualities, so that continuous selection is not an efficient means of improvement. The isolation of mutants, on the other hand, is thought by Pritchard to offer promise of improvement, but if the mutation method is to be used, it is deemed essential that more efficient experimental methods be devised to reduce the effects of soil differences and thus make it possible to distinguish real differences more clearly (see Chapter XXV). Other normally cross-fertilized crop plants in which mutations are known to have occurred are cotton, hemp, rye and the sunflower. The Search for Mutations.- — It has long been thought that the two most effective methods of inducing heritable variations in plants are hybridization and change of environment. Regarding the importance of the first there is of course no question, and there is evidence that very radical changes of environment such as Tower applied to his beetles and White to his tomatoes may induce germinal variation. But the idea that mere change of location from warmer to cooler climates or from poorer to richer soils, or vice versa, is very effective in " breaking the type" finds very little to support it. This notion that culture induces germinal variation doubtless finds its explanation in the fact that sooner or later after a plant is subjected to intensive culture and close observation new heritable varia- tions appear. But why conclude that these variations are induced by culture? During the first season of garden cultivation of a species of tarweed two mutations were discovered. One was a change in the color of the stamens, the other was petallody in the ligulate flowers. It seems very probable that these variations would have occurred had the plants been growing in the wild. They were found because the plants were closely inspected. But is their not fair evidence that cultivation of the same species in different regions gives rise to different mutations? There is danger of befogging the issue by this question unless we distinguish clearly between the origin of mutations and the origin of varieties. To consider only one of many possible illustrations : the native sorghums of South Africa, the Sudan, Egypt, Arabia and Persia, India, and Manchuria are a diverse lot of forms; yet sorghum undoubtedly originated in Africa and spread thence to the various regions where it now exists as distinct varieties. It must be admitted that different varieties have developed in different regions, but does this necessarily indicate that geo- graphical differences actually caused the original germinal alterations which resulted in the different varieties of sorghum? Such a conclusion seems unwarranted in view of what is actually known concerning the occurrence of mutations under both natural and artificial conditions. Moreover, it is not improbable that the progenitors of existing varieties of sorghum all originated in Africa, although geographical differences may have been the determining factor in the survival of those mutations MUTATIONS IN PLANT BREEDING 371 which gave rise to existing varieties. Factor mutations conform in their manner of occurrence with de Vries' mutation theory; they arise suddenly, they occur in all directions, they are heritable, and some of them are advantageous to the species and are preserved by natural selection. When so preserved they give rise to new forms or races, and when fostered by man they make possible new horticultural varieties of l)lants or new breeds of animals. But as yet we have no ground for asserting that factor mutations are caused by geographical differences or by any specific elements of the environment. From his study of varia- tion in tobacco Hayes reached the conclusion that while environment is of great importance in breeding tobacco as well as in growing the commer- cial product, yet change of environment "does not cause a breaking up of type, and whatever variations occur due to environment appear alike in all plants of a particular type." Thus it appears that mutations arise quite independently of conditions of culture, and it is probable that they ai-e somewhat more frequent than has generally been supposed. It is certain that mutations which are undesirable for agricultural purposes are quite as apt to occur as are desirable ones. For this reason neglect of seed selec- tion has caused the "running out" of many good varieties. The breeder who would improve the best existing varieties which are adapted to a given location must either resort to hybridization between the varieties or else search for the most desirable biotypes within each variety. Even though hybridization is clearly necessary from the first, it may well be preceded by a systematic search for the best forms within the varieties to be crossed. One of the most successful attempts to take advantage of the desirable mutations which had occurred within cultivated varieties was carried out by Surface and Zinn at the Maine Experiment Station in their experi- ments on oat breeding. Oats being self-fertilized, they assumed that any new characters which had originated would breed true. The two most desirable improvements in the commercial varieties of oats culti- vated in Maine are increase in yield and in strength of straw. Accord- ingly individual oat plants were selected with these two points in mind. This work began in 1910 when 460 plants were chosen from 18 different commercial varieties. Of these only 188 were selected for planting in 1911 and on the basis of the results obtained 80 were continued for test- ing in duplicate 1-2000 acre plots in 1912. Of these pure lines 34 were sufficiently promising to be continued into field tests in 1913. Thirty- one of these were again tested in 1914. In 1915 all of these pure lines were discarded except 12 and these were tested in quadruplicate plots in 1915. In each of the 3 years 1913-15 these pure lines were grown along with a number of the best commercial varieties obtainable. In 1914-15 the pure line plots alternated in the field with commercial variety plots. 372 GENETICS IN RELATION TO AGRICULTURE After correcting the yield of individual plots for differences in soil fer- tility (see Chapter XXV) it was found that the 12 pure lines averaged to yield 80.8 bushels per acre against 75.2 bushels for the 11 commercial varieties. "Only 4 of the commercial varieties gave a better yield than the poorest of the pure lines. In all cases the average yield of the pure lines selected from a given variety exceeded the yield of the parent variety." Of the 18 commercial varieties with which they started only 3 are represented among the 12 pure lines. It was found that these pure lines closely resemble their respective parents in morphological characters and concluded, therefore, that mutations in the physiological characters which result in higher yield are not necessarily associated with changes in morphological characters of the plant or grain. Even more striking results in some respects have been obtained by Clark by means of head-to-row selections of Ghirka spring wheat in North Dakota. Starting with 300 individual plant selections in 1909, in spite of the destruction of all the cultures by hail in 1912, after 5 years' work two pure lines were found, one of which was superior to unselected Ghirka in all characters except crude protein content and the other in all characters except volume of the baked loaf. In all such work the importance of beginning with large numbers must be emphasized. Other things being equal as the number of selected in- dividuals increases the chances of locating the desired variants increase. The same holds true when attempting to locate aberrant individuals by inspection of young seedlings, nursery stock, etc. Propagation of Mutations. — The question of how to preserve and prop- pagate a desirable mutation becomes a problem only in plants which are normally cross-fertilized. In self-fertilized species the new form exists as a pure line and need only be isolated. Similarly in plants that are propagated vegetatively usually there is no difficulty in multiplying a new variet3^ This method has been applied to such crop plants as alfalfa with great success. But in a cross-fertilized plant in which vegetative propagation is impracticable the method of procedure will depend upon circumstances. If in a given species the plants are self- fertile the appearance of only a single plant of a new form makes it possible to test its genetic constitution and if it is a mutation to multiply it. If it is heterozygous for the factor conditioning the new character or char- acters it will of course be necessary to select the best individuals from the next generation. This situation will confront the breeder only in the case of mutant factors which are dominant or partially dominant when in the heterozygous condition. When the mutant factor shows partial dominance in a heterozygote but segregates as a Mendelian recessive there will be no difficulty in establishing a pure strain, but should it segregate as a Mendelian dominant it becomes necessary to test a number MUTATIONS IN PLANT BREEDING 373 of the seedlings exhibiting the new characters. In cross-fertiHzcd species, in which individual plants are self-sterile, where a mutation appears in only one plant, several successive crosses may be necessary in order to produce a strain which breeds true for the new type. It must be crossed back on the parental form to begin with. If the change from the parental type is conditioned by a single factor the number of hybrid generations to be raised will depend on whether that factor segregates as a dominant or a recessive. In the latter case a true breeding strain should be obtained in the second generation but in the former it will require three or more hybrid generations depending on the extent to which the new characters depend upon environmental conditions for their expression. CHAPTER XXII GRAFT-HYBRIDS AND OTHER CHIMERAS A graft-hybrid, as its name implies, is a shoot or plant which is pro- duced by grafting one kind of plant upon another and whose characters are intermediate between the characters of the two components. None of the so-called graft-hybrids are really hybrids at all; they are merely mixtures of tissues from two kinds of plants which can live in unison. But each kind of tissue is distinct in its every cell, i.e., there has been no fusion of cells or blending of germ plasm as in the case of sexually produced hybrids. Some difference of opinion still exists regarding a single case which will be referred to again, but the above statement certainly applies to all other graft-hybrids that have been investigated. As Buder says, a graft-hybrid is nothing else than a special form of graft-symbiosis. Thus all graft-hj'brids are chimeras (p. 271). A conception of how chi- meras originate naturally may be gained by learning how graft-hybrids have been produced experimentally. Tomato -nightshade Graft-hybrids. — Some members of the night- shade family are easily grafted even though they belong in different genera. Thus it is possible to double work nightshade on tobacco on tomato. Reciprocal grafts of tomato and potato are easily made and Heuer has grafted tomato on egg plant and tomato on bittersweet. In recent years Winkler has produced four different chimeras and another form which he considers a true hybrid by grafting tomato on nightshade or vice versa. The four Solanum chimeras are shown in Fig. 151. Wink- ler's method is to graft on a scion by one of the ordinary methods and soon after it has united with the stock to cut it off taking pains to make the cut pass through the united tissues of stock and scion as shown in Fig. 152. Most of the adventitious buds pushed out are either night- shade or tomato. But occasionally a bud will be formed on or near the line of union. In such cases either one of two combinations of the graft- components may result depending on the relations of the two kinds of callous tissue. If the two masses meet as in a of Fig. 153, the young shoot will consist of sectors of nightshade and tomato, but should one of the cell-masses grow over the other tissue producing the condition shown in 6, the young shoot in this case will be composed entirely of tomato inside but will have an envelop of nightshade cells surrounding it. The 374 GRAFT-in URJlJ.S A.\JJ OTIIKk CHIMERAS 375 first type is termed sectorial and the second periclinal. A third type has been recognized \)y Winkler in which the vegetative cone is a mosaic of unlike cells. This type he named hyperchimera. Coit has reported the case of a Valencia orange tree whi(;h from its consistent instability appears to have been propagated from a mixed bud and hence belongs in this class of chimeras. All of Winkler's Solanum chimeras are periclinal and the degree of lesemblance of such a graft-symbiont to the parent whose tissue comprises the inner portion of the shoot seems to depend upon the number of layers of cells from the other parent which envelop it. Thus the form luhujense A li Fig. 151. — Winkler's Solanum chimeras. Produced by grafting tomato on nightshade (and vice versa). From left to right, Solanum Gaertnerianum, S. Koelreuterianum, S. proteus, and S. luhigense. The second resembles the tomato parent most closely; it has a tomato body with nightshade epidermis (a single layer of cells;. The fourth is most like the nightshade parent and it has a nightshade body with tomato epidermis. The first has a tomato body covered with 2 layers of nightshade cells and the third, a nightshade }>ody covered with two layers of tomato cells. Such combinations are called periclinal chimeras. {From Journal of Heredity.) (Fig. 151, IJ) wliich closely resembles the nightshade has all the inner portion of ;S. nigrum, with just an epidermal layer, one cell thick, of S. lycfjper.sicum. But proteus (Fig. 151, C), whose leaves are much more like tomato leaves, has a double layer of tomato cells overlying the nightshade body. Similarly, Koelreuterianum (Fig. 151, B) is really a tomato with nightshade epidermis, while Gcertnerianum has a tomato body covered with two layers of nightshade cells. These graft-hj'brids were discovered only after much patient work in the course of which Winkler made 208 grafts which produced more than 3000 shoots. All four were propagated from cuttings and by this method they have been obtained and grown by the New York Botanical Garden. It has not been possible to compare these forms with true sexual hybrids because no one has yet succeeded in crossing the tomato and the 376 GENETICS IN RELATION TO AGRICULTURE nightshade. But sufficient evidence of their mixed composition is found in the character of their progeny and from the study of chromosome numbers made by Winkler. In tuhigense the fruits are almost identical Fig. 152. — Diagrams showing methods of grafting used in producing the tomato- nightshade chimeras and some of the results; shaded portions represent scion tissue, un- shaded, stock tissue, a, Splice graft; b, cleft or wedge graft; c, saddle graft; d, sectorial chimera (shaded portion, nightshade; unshaded portion, tomato tissue); e, chimera leaf, part nightshade, part tomato; /, nightshade; g, periclinal chimera, Solanum. tuhigense; h, Tomato. {After Winkler from White.) with those of the nightshade and the seeds produce only pure nightshade plants. The reason for this is clear when it is remembered that this form is really nightshade with a single epidermal layer of tomato cells, since GRAFT-HYBRIDS AND OTHER CHIMERAS 377 the germ cells arise from the sub-epidermal layer. The seedlings of Gcertnerianum are also pure nightshade because, although this form consists of tomato tissue within, it is enveloped by two layers of night- shade cells. Similarly with proteiis, which is a nightshade except for its two outer layers of tomato cells, the fruits resemble tomatoes and from the seed pure tomatoes have been raised. The other form, Koslreuterianum, fails to produce fruits. The chromosome numbers are 24 for the tomato and 72 for the nightshade. If a fusion of nuclei involving diploid numbers had occurred the cells of the supposed hybrids should contain 96 chromo- somes, but the only counts obtained by Winkler in the four chimeras were 24 and 72. Thus it appears that in each graft-symbiont the two kinds of tissue maintain their identity. Yet there is a combined effect on the morphological characters. The physiological interactions too are such as to cause reduced vigor. This effect is least noticeable in Kcelreuterianum which is sterile. The fifth new form, which Winkler Fig. 153. — Diagram to show formation of adventitious buds arising at the point of union of the two graft-components A and B. a, represents a sectorial combination; b, a periclinal combination. (After Buder.) claims is a true graft-hybrid, was named Darwinianuni. It appeared on one of the shoots from a decapitated graft. Winkler claims the chromo- some number of this form is 48 and that certain if not all of the tissues of this plant are composed of cells derived from the actual fusion of tomato with nightshade cells which involved nuclear fusion. If this is actually the case S. Darwinianum is a hybrid in the strict sense and the only one known to have been produced by vegetative means. However, Baur points out that Winkler bases his claim for this number on the ground that he found 24 chromosomes in the pollen mother-cells which arise from the sub-epidermal layer. Baur thinks that Winkler's interpretation is unwarranted. He believes it much more probable that S. Darwin- ianum is a periclinal chimera with a nightshade epidermis, then a sub- epidermal layer of tomato cells and the adjoining inner tissues of night- shade. Then the number found, 24, is the diploid number of the tomato and the reduction division, according to this explanation, is omitted in the tomato pollen mother-cells which, in this chimera, are bounded on both sides by nightshade tissues; or else it occurs at an unusually late stage in development. ''I cannot admit," says Baur, "that the exist- ence of real graft-hybrids in the strictest sense of the word is proven." 378 GENETICS IN RELATION TO AGRICULTURE Since the true nature of Winkler's chimeras has been made clear a number of historical cases of graft-hybrids have been investigated. The results of this work have been summarized by Buder whose list appears in Table LIII. Most of these cases have been fully discussed in other works. Typical leaves of the two types of Cratsegomespilus and of the two parents, the whitethorn and the medlar, are illustrated in Fig. 154. Table LIII. — List of the Most Important Periclinal Chimeras Produced BY Grafting. {Adapted from Buder.) Name and origin Used in grafting As stock As scion Composition tiahurnumiCytisus) Adami , arose spontaneously in 1826 from an unsuccessful graft. The Cratsegomespili of Bron- vaux, originated spontaneous- ly many decades ago at Bron- vaux in Metz from places where stock and scion had overgrown on grafts nearly a century old. (a) Cr. Asnieresi (resembling whitethorn) (b) Cr. Dardari (resembling medlar) The Cratsegomespili of La- grange, apparently complete analogues of the two forms from Bronvaux. The pear-quince "hybrid" of Fr^re Henri, originated about 1903 in Rennes. The peach-almond graft hybrid of Daniel and Delpon, arose spontaneously in 1908 at Mas- Grenier (Tarn and Garonne).' The tomato-nightshade "hy- brids" of Winkler, produced experimentally in 1907-9. (rt) S. tubigense (b) S. proteus Laburnum vulgare (Shower of Gold). Cratmgus monogy- na (Whitethorn). Cydonia (Quince). Amygdaius com- rminis (Almond). Solanum lycoper- sicum (Tomato). Cytisus purpureus (Purple Broom). Mespilus german- ica (Medlar). Pyrus (Pear). Amygdaius persica (Peach). Solanum nigrum (Black nightshade) (c) S. Kcelreuteriatium. (d) S. Gwrtnerianum. . . (e) S. Darwinianum. . . The tomato-eggplant and to- mato-bittersweet "hybrids" of Heuer, produced experiment- ally in 1910. Form I Form II. The populus "hybrid" of Baur, produced experimentally in 1911. Solanum, lycoper- sicum (Tomato). Solannm lycoper- sicum (Tomato) Populus canaden- Solanum melongena (Egg plant). S. dtdcamara (Bittersweet). P. trichocarpa. According to Buder only one outer layer of C. purpureus, all within being L. vulgare. According to Baur and H. Mayer both forms have a Cratsegus body which is cov- ered by a Mespilus mantel: In (a) of one layer of cells. In (6) of two layers of cells. Probably consists of pear tis- sue within a layer of quince cells. Evidently a mixture of sec- torial and periclinal chim- According to Winkler: Outer layers Inner tissue 1 of S. lycopers. S. nigrum 1 and 2 S. lyco- S. nigrum persicum 1 of S. nigrum S. lycopers. 1 and 2 S. nigrum S. lycopers. ".\n actual hybrid" (but see text). Probably only the epiderrnis is egg plant, tomato within. Epidermis of tomato, inner portion bittersweet. Only the epidermis of P. tri- chocarpa, within P. canaden- In addition to the above there are the Bizarria as they have been termed. These are periclinal chimeras (some of them perhaps also sectorial chimeras) between different species of Citrus: Pomeranze, Citrone, Cedrate, Limette. The earliest record of these dates from Florence, 1644. They aroused in- terest in their day because of the manifold sectorial and periclinal chimera combinations in their fruits. Although most of these forms are now forgotten, several are still in cultivation, but they have not received close study (but see Coit, "Citrus Fruits"). Baur's Investigation of a Natural Chimera. — The key to the ex- planation of Winkler's artificially produced chimeras was furnished by Baur's discovery of the difference between the white-edged and solid GRAFT-llYHRIDS AND OTIIEU CIIIMERAH :37i) green vajieties of geranium {Pelargonium zonale). From his stud}- of seedlings of the white-edged variety he had come to reahze that the color of the leaves on a seedling depends entirely upon the nature of the cells composing the vegetative cone or plumule. This led him to examine a b c d Fig. 154. — Leaves of the CratjEgomespili of Broiivaux and their components. ' a, Mespilus germanica (Medlar); d, Crataegus monogyna (Whitethorn); b, Cratoegoniespilus Dardari, with two outer layers of medlar cells; c, Cratwgomespilus Asnieresi, with one outer layer of medlar cells. (After Buder.) the cells in white-edged and green leaves and he found that in a white-edged leaf there is an extra layer of colorless cells in addition to the true epidermis (see Figs. 155, 156 and 157). He concluded that a plant bearing all white- edged leaves must have a complete peripheral layer of the colorless cells just below the epidermis as shown in Fig. 157, a, and that a plant differing Fig. 155.— Leaves from periclinal chimeras of the white-edged geranium ; a, from a plant with two white peripheral cell-layers; b, from a plant with only one epidermal layer of colorless cells. {After Baur.) in this respect from a normal green plant should be considered a peri- clinal chimera. He had observed sectorial chimeras among his geranium seedlings and found that occasionally a plant having some of its leaves entirely green and some of them entirely white would produce a shoot bearing white-edged leaves. He found that such shoots arose near the 380 GENETICS IN RELATION TO AGRICULTURE Fig. 156. — Sections through edges of the two leaves shown in Fig. 155. Green tissue indicated by stippling. For much enlarged views of the portions enclosed by the small rectangles see Fig. 157. (After Baur.) Fig. 157. — Microscopical views of those portions of the cross-sections in Fig. 156 represented by the small rectangles. Colorless chromatophores are indicated in outline, green in black. {After Baur.) A B Fig. 158. — A, Diagram of longitudinal section through a young shoot of a plant bearing white-edged leaves — a periclinal chimera. B, Cross-section of stem of a sectorial chimera. A bud pushing out at a would produce a sectorial chimera, while one arising at h would form a periclinal chimera. {After Baur.) GRAFT-HYBRIDS AND OTHER CHIMERAS 381 l)ouiitlaiy line between the green and the white tissue of the stem. Hence he conchided that in order to have a perichnal chimera arise from a sectorial chimera the relation between the two kinds of tissue would have to be as shown at b in Fig. 158. A bud pushing out at such a point would have an envelope of colorless cells in addition to the colorless epidermis. Without doubt this structural principle explains the origin of all natural chimeras. But this principle holds only when there are groups of normally homogeneous tissues in the same stem or bud which have come to differ with respect to one or more characters. In graft-hybrids such diverse tissues come from different plants. The question of the natural origin within the same plant of morphological and physiological differ- ences causing somatic heterogeneity where homogeneity is the ordinary condition is a problem of far greater fundamental importance. It has already been shown that probably all such diversities within single individuals arise as factor mutations (Chap. XIV). Other Natural Chimeras. — Sectorial chimeras caused by mutations in color factors are the most common natural chimeras. They occur very frequently in citrous fruits, especially in the orange and lemon (see Fig. 161). Other chimeras in these fruits are caused by factor dif- ferences affecting thickness and texture of rind and frequently associated with these are differences in color and flavor of the pulp. Color chimeras are also fairly common in apples and pears and they have been found in grapes, olives and tomatoes, as well as in gladiolus, poppies, sunflowers, dahlias, and doubtless many other flowers. Many valuable variegated forms of ornamental shrubs are mixtures of sectorial and periclinal chi- meras with normal green vegetative parts. Some striking examples are the Variegated Black Elderberry {Samhucus nigra variegata), the Variegated Deeringia {Deerimjia celosioides) and the variegated forms of the Japanese Spindle Tree or Strawberry Bush {Euonymus japonicus) . Variegated foliage which is caused by factor mutations causing complete or partial chlorophyll reduction are also fairly common among herbaceous plants. Two Categories of Variegation. — The variegated plants mentioned above, like the white-edged geranium, can be propagated asexually and it is known that in the geranium, snapdragon, four-o'-clock, maize and other plants the variegated character can be transmitted to sexually produced offspring. However, certain variegated plants cannot transmit variegation through the seed although it is transmissible by means of vegetative propagation. Baur has shown that in the latter class, variegation results from a pathological condition and by double working susceptible and immune stocks he determined that it must be caused by a toxin produced by the diseased cells. The long familiar cases of "graft infection" among the Malvaceae are thus explained. It seems that all cases of "infectious chlorosis" in this family can be traced back 382 GENETICS IN RELATION TO AGRICULTURE to a single variegated specimen of Ahutilon striatum Dicks., which was introduced into Europe in 1868 and named Ahutilon thomsoni. Miss Reid has shown, however, that among the flowering maples (Abutilon) the variegated forms can be grouped into two classes: "those with a mottled variegation which is infectious and those with a non-infectious variegation with the white cells at the periphery. Both types are of importance in horticulture, especially for use as bedding plants; both types are of special scientific interest." The Physiological Behavior of Graft-hybrids. — Although chromosome counts and progeny tests indicate that the cells of each graft-symbiont maintain their identity independently of the close proximity of foreign cells, yet the intermediate characters of graft-hybrids indicate that the components have a mutual influence upon each other. This influence is especially notable in the manifestation of physiological activity in- volving the whole plant. None of the Solanum graft-hybrids are as vigorous as either component under normal conditions. In fact, it is with considerable difficulty that they are maintained by means of cut- tings, except in the case of Koelreuterianum. This lack of vegetative vigor may not be characteristic of all graft-hybrids but it seems to be common to most of them. Lack of vigor exists in many natural chimeras also, especially in those involving chlorophyll reduction. In the Solanum graft-hybrids the germ cells of the two components are not equally susceptible to the effect of adjacent foreign cells. In both tuhigense and Gaertnerianum the fruits contain fertile seeds which pro- duce only pure nightshade plants. But in proteus only part of the seeds are viable and these produce tomato seedlings, while in Koelreuterianum the flowers are entirely sterile. Similarly, in Cytisus Adami the "hybrid " (intermediate) flowers are sterile. Graft-hybrids offer many interesting possibilities as a means of studying the physiology of development as influenced by the reciprocal relations between the components. This method of attack has been utilized much more extensively in the study of development in animals than in plants. Crampton, for example, grafted together the pupae of different species of moths thus producing double monsters. From the specific effects upon pigmentation iji some of the graft symbionts it was concluded that the pigmental colors in some species are derived from the haemolymph by processes of drying and decomposition which are regulated by some specific internal factor. Crampton mentions several other investigators who performed similar experiments on animals. Buder has suggested the re- ciprocal grafting of male and female plants in dioecious species as a means of investigating the physiology of sex determination in plants. This suggests the whole field of reciprocal effects between scion and stock, concerning which there has been considerable investigation in recent decades. GRAFT-HYBRIDS AND OTHER CAIMERAS 383 Modification of One Graft-symbiont by the Other. — The repressing or stimulating effects of certain scions on certain stocks is well known. Excellent examples are found in the various "dwarf" rootstocks used in the culture of the pome and citrous fruits. Besides the dwarfing effect of the stock upon the scion, there is often a reciprocal stimulating effect of the scion upon the stock which causes the latter to increase in diameter faster than the scion. A mutually stimulating effect is sometimes ob- served, as in the almond and the peach when used as graft components. The importance of selecting stock of about the same vigor, when grafted, as the scion has long been recognized by nurserymen. The fact that grafts usually exhibit a certain amount of modification according to the kind of stock used has given rise to many reputed cases of deleterious effects and extreme modifications due to grafting. A matter of considerable economic importance involving this question concerns the culture of wine grapes. After the introduction of American vines and their hybrids into the phylloxera-infested districts of France there was widespread concern over the possibility that the quality of the French wines would be injured by grafting on the new stocks. Many investigations were carried on. Although in the earlier stages of the work some very definite effects of stock on scion were reported, the evidence as a whole is considered by leading investigators as indicating merely that the stock may either increase or decrease the capacity of the scion according to the combination used. It has been concluded that where due account is taken of affinity of stock and scion, if other condi- tions are favorable, grafting has caused no deleterious effect on yield or quality. A specific, case of supposed deleterious effects attributed to the in- fluence of stock on scion was observed by Paelinck. A dark-red variety of cherry. Early Rivers, was grafted on mahaleb stock. The resulting tree bore fruit which was yellowish white in color, of smaller size and which matured 8 days later than Early Rivers. Scions from this white- fruited tree were grafted on mazzard stock which has small black fruit "to see whether the white fruits would revert to the dark color." The result as one would expect was negative. Undoubtedly this was a case of bud mutation. Other reputed extreme effects of graft-symbionts involve the supposed transfer of characters from the one to the other. Baur asserts that after reviewing the accounts of many grafting experiments he has reached the conclusion that most of the reciprocal effects between stock and scion can be explained on the basis of modification in nourishment. Moreover, where this explanation does not hold there is a more probable cause than the notion of transfer of characters. For example, in the case of Daniel's eggplant which, when grafted on tomato bore tomato-shaped 384 GENETICS IN RELATION TO AGRICULTURE fruits, Baur states that there are varieties of eggplant which occasionally bear tomato-shaped fruits even when not grafted, and that Daniel probably used such a variety. Again Daniel and Elder have reported experiments tending to show that the seedlings of scions exhibited an influence of the stock. Baur is inclined to think that accidental cross- pollination must explain such cases. But Daniel has recently reported similar results when working with different varieties of beans. In this case, however, there is the possibility that the seedlings of the scions and the seedlings used for comparison, which were from ungrafted plants, belonged to different pure lines. Thus in some such simple manner all the supposed cases of transference of morphological characters may be explained. Regarding the actual transference of the chemical constituents of the tissues from stock to scion and vice versa, the results of experiments differ with respect to different plant ingredients. Thus, according to Guignard, glucosids do not pass from one graft component to the other when the two contain different kinds of glucosids, and the glucosids present in plants are apt to differ unless the plants are closely related. In graft-symbionts whose components belong to different species, Guignard thinks that each component tends to retain its own chemical properties. On the other hand, Meyer and Schmidt found that alkaloids such as nicotine will pass from a tobacco scion into a potato stock. This is a promising field for future investigation. CHAPTER XXIII BUD SELECTION The efficacy and practicability of bud selection is a subject of con- siderable interest especially among horticulturists. During the past de- cade it has received more and more attention from investigators until now there are under way a number of comprehensive projects which, in future years, should furnish definite information concerning the more important vegetatively propagated crop plants. If it is determined that bud selec- tion is an effective method of improving certain varieties either by secur- ing increased yield or by the discovery of superior strains, its importance to horticulture will have been demonstrated. It will still remain for horticulturists to decide as to the practicability of introducing systematic bud selection in the commercial propagation of those plants in which it has been proved to be an effective method of improvement. The efficacy of bud selection depends upon the nature of bud variation. Bud Variation in Plants. — There are two kinds of bud variations, viz., modifications and mutations. Modifications are common to all plants and are easily detected even in dormant buds. On deciduous trees, for example, the buds formed during one season's growth usually show considerable variation in size. Such variations do not necessarily repre- sent inherent differences between the buds. They are usually due to differences in the particular combinations of conditions which exist during development of the buds. Phytomers exhibit fluctuating modifications in all other characters as well as size in response to the varying con- ditions of nourishment, light, temperature and other elements of the environment. These modifications are not transmissible and selection of such bud variations alone could never change the average output of an orchard or establish an improved strain. Bud mutations, on the other hand, although comparatively rare, are of general occurrence and the new characters induced by them are trans- missible. Hence in considering the efficacy and practicability of bud selection in a horticultural variety the first thing to be determined is the nature and frequency of somatic mutations in that variety. There is only one way in which this question can be answered completely and definitely and that is by extensive tests of vegetatively propagated off- spring. Such tests must be made under controlled conditions especially as regards the nature of the rootstock on which the tested scions are pro- 25 385 386 GENETICS IN RELATION TO AGRICULTURE pagated. Careful inspection may reveal a certain number of chimeras and bud sports. Both, in fact, are comparatively common in some varieties, such as the Boston Fern and the Washington Navel Orange, and they sometimes give rise to superior new biotypes. Yet inspection alone is not sufficient. A new fern sport must be propagated in order to test its constancy when multiplied vegetatively. Similarly a supposed orange sport must be propagated and the progeny must be tested in order to ascertain whether the selected phytomer is really a mutation and a desirable one at that. Having discovered a new type which originated by bud mutation, the question arises : Will there be any practical difficulty in maintaining this new form by means of vegetative propagation? Are additional somatic mutations likely to occur in sufficient number to endanger the preservation of the selected form? It will be remembered that sometimes bud mutations in the Boston Fern produce ever-sporting varieties that have little or no commercial value. Similar inconstant forms have arisen in other ornamentals. Nevertheless it has been practicable to propagate vegetatively many valuable bud sports and hybrids, in- cluding some that are highly variable. One such product of composite hybridization is the cultivated Coleus. This ornamental foliage plant is commonly used for beds and borders in summer and as a conserva- tory plant in the colder months. The foundation stock was produced in England about 1867 by hybridizing four different exotic species. Hundreds of named varieties have been produced, some having appeared as bud mutations, but the majority being seedlings. Some of the varie- ties now in cultivation are characteristically variable. In one such variety Stout has investigated the variations in leaf color pattern and leaf shape in a series of 833 plants, all descended by vegetative propagation from two similar plants. Bud Selection in Coleus. — The two plants with which Stout began had a definite pattern of leaf coloration consisting of a green mid-region and yellow border with blotches of red in the epidermis (Fig. 159). The green and yellow pigments exist in the sub-epidermal layers. The vegetative offspring from the two original plants were kept separate and the simple habit of branching in this plant made it possible to indicate the particular branch as well as the individual plant from which a cutting was taken. In this way Stout was able to trace the pedigree of any plant to its original source. During the course of the investigation 16 new color patterns were obtained. There also appeared the laciniate form of leaf which is seen in the younger leaves of the plant on the left in Fig. 159. Of the 16 new color patterns 15 arose by somatic mutations which produced bud sports either directly or, in some cases, indirectly from chimeras. The other new pattern arose solely as a fluctuating variation. BUD SELECTION 387 Fig. 159.— ^Two Colcu.s plants which descended from the same branch, which at the time it was propagated was uniform as to leaf shape and bore leaves having the same general type of color pattern, viz., green mid-region and yellow border with red blotches on the epidermis. The right hand plant resembles the original plant although it represents the fourth vegetative "generation." The left-hand plant shows the laciniate type of leaf which appeared several times as a bud mutation. (From Stout.) Yellow- Red /Blotched Green- Yellow Ked Blotched Spontaneous Yellow —5 7 --aor4auciE 5 and '8a Green-Yellow Spotted-Solid .Eed 13a Green-Yellow Spotted Spontaneous Yellow id E ■9 Greeu-Solid Red 13n ■13 Green 13a 1 14 Yellow-Green 9 Green-Solid Ked ^^ ""^^ ^ 10 Green- Yellow-Solid Eed Upper Center ll-Yellow- Green-Solid Red ICYellow. Solid Red 'l2 Green-Yellow Fig. 160. — Diagram showing derivations of color patterns in Coleus. The descriptive name of a color pattern is given only whore it first appears in a line of descent. A contin- uous line indicates origin by bud mutation and a dotted line indicates fluctuating variation. (No. 2 = original form. No. 7 = laciniate-leaved form. E = entire leaves.) (From Stout.) 388 GENETICS IN RELATION TO AGRICULTURE It consisted of absence of yellow and decrease of red in the younger leaves of a few plants. Six of the 15 patterns that arose as bud mutations also appeared more or less frequently as fluctuating variations on certain plants. The derivations of the various color patterns are shown by the diagram in Fig. 160. It will be noted that the original pattern, No. 2, is described as green-yellow-red blotched which means green center, yellow marginal border and red blotches on the epidermis. In the same way the description of pattern No. 4 is interpreted as green center, yellow- spotted marginal border and red blotches on epidermis. The frequency with which these various bud mutations occurred is shown in Table LIV. Here are indicated under ''Plants" the number of plants in which each type of change might have occurred, under "Frequency," the actual number of bud mutations that did appear, and, finally, the ratio of bud mutations to constant buds. This ratio is obtained by assuming that on the average each plant produced 200 buds. Table LIV. — Frequency op Bud Mutations Producing New Color Patterns AND Leaf-shapes in Coleus. {After Stout.) Type of change Plants Frequency Ratio Increase of yellow and decrease of green . Decrease of yellow and increase of green . Reversal of positions of green and yellow Increase of epidermal red to solid red . . . Decrease of epidermal red, complete loss Decrease of epidermal red, all cases Appearance of the laciniate character . . . Entire leaf from laciniate leaf 827 27 740 50 450 8 770 8 815 19 815 21 765 13 68 1 6,130 2,960 11,250 19,250 8,580 7,760 11,770 13,600 Stout remarks that these data indicate the tendencies of the bud variations and give a clew to the behavior of the characters in question. Thus, decrease of yellow occurred twice as often as increase of yellow, and loss of red 2.2 times as often as increase of red. Although these data indicate a tendency toward loss rather than gain of the two colors, the fact that the number of mutations involving gain is about half as large as the number involving loss has considerable interest. It has been generally considered that mutations involving addition of a character are exceedingly rare. While this may be the case in many pure species, it would appear from the above evidence that among the progeny of species hybrids such mutations may be relatively frequent. The manner of appearance of these bud mutations was typical of somatic factor mutations. Stout says, for example, "the loss of yellow, loss of green, and gain and loss of red all occurred in single branches and in sections of branches. Frequently two quite different changes occurred BUD SELECTION 389 on the same plant." After citing a case of modification in degree of pigmentation in Coleus by the use of artificial light, Stout declares: ''In marked contrast to these results it may be noted that the bud variations that I have reported give more marked changes than those induced by Flammarion and that these appear suddenly and in a sector of a bud in a manner that suggests internal readjustments rather than external environmental influence." Stout tested the seed progeny of two of his plants, obtaining in all 45 plants from selfed seed. As would be expected in such a case of composite ancestry, there was a wide range of variation in size as well as shape of leaves and in leaf coloration every gradation between pure yellow plants that died within a few weeks and pure green. No better evidence could be offered that these highly variable characters are actually conditioned by specific factors which segregate and recombine in sexual reproduction. The history of Coleus as reviewed by Stout also contains some interest- ing facts about the origin of the characters under discussion. The four original species that furnished the foundation stock, from which the modern Coleus has been developed, contained no yellow coloration whatsoever. They were characterized by green leaves overlaid with different shades of red, purple or chocolate. The first appearance of yellow occurred in a leaf sport, i.e., in one-half of a single leaf the green was exchanged for a decided yellow tint. "The bud at its base was propagated and gave the new variety." However, it does not appear that this sport was used in hybridization work. "Yellow coloration also appeared as a new or spontaneous development" among the second lot of hybrids raised at the gardens of the Royal Horticultural Society about 1869. Again in regard to the laciniate-leaf character, as early as 1856 a variety of Coleus hlumei (which was the first species introduced into Europe and was originally described in 1826) was described as being ''somewhat more richly colored but differing chiefly in having the leaves deeply and doubly lobed." While there is no record of the origin of this variety, it is certain that it appeared before C. hlumei had been hybridized with any other species. It is very probable, therefore, that it arose as a bud mutation. Thus it appears that two of the three characters whose presence, absence or partial development figure conspicuously in Stout's "bud variations" actually originated by factor mutations during the early horticultural history of this plant. Regarding the efficacy of selection in maintaining the new forms that arose by bud mutation, Stout's results show clearly that even in such a highly variable plant as a horticultural variety of Coleus, bud selection is very effective. Stout describes his methods as follows: "The series of plants considered under any type jiattern are in large measure a selected stock. When cuttings were made for the perpetuation of the pattern 390 GENETICS IN RELATION TO AGRICULTURE in a new generation, they were made from the plants most typical and constant for the pattern concerned. When a bud variation appeared, if the conditions were favorable, the parts possessing it were allowed to develop until there were several branches from which cuttings could be taken simultaneously. In such cases the selection of branches for the new type was a simple matter, as it depended on the talcing of branches sharply distinct from the main part of the plant. . . . When further cuttings were made for a new generation to perpetuate the type they were made from plants most uniform and constant (determined from the records) for the pattern in question. Usually but three cuttings were taken from a plant and these were taken from branches most uni- form and clearly conforming to the type." The relative numbers of "constant" plants and of plants showing either clear-cut bud mutations or ''fluctuations" are of considerable interest in connection with this matter of maintaining strains. These proportions are given in t\vQ tabulated summary of the main clones that were derived from one of the two original plants Table LV. Clones 11 Table LV. — General Summaky of Clones Derived prom Plant No. 1 Clone Total number of plants Plants constant Per cent, of plants constant Number of bud variations Ratio of frequency 11 12 13 14 117 111 211 192 138 155 91 34 132 87 75 80 54 29 62 45 54 51 59 85 49 21 4 18 31 4 860 1,830 6,900 1,720 590 1,700 and 12 were derived from two branches that had the same color pattern, yet it seems that they possessed "quite different potentialities for con- stancy and for bud variations. Even more marked differences than these developed among the various subclones. A study of pedigrees shows that in all patterns and in all main clones there were certain lines of progeny much more constant than many others. These could not be detected by any other than a pedigree method." These facts have a definite bearing on the maintenance of yegetatively propagated varieties or strains. If similar diversity as regards degree of variability exists in other cultivated plants, as it undoubtedly does, careful bud selection must be of prime importance in keeping varieties true to type or at least in preventing deterioration through the accumulation of undesirable bud mutations. In Coleus at least it is certain that bud selection is effective and necessary in maintaining strains true to type. If it is equally potent in its effect on other vegetatively propagated plants, bud selection should be given far more attention by nurserymen than it has generally received. BUD SELECTION 391 Bud Selection in Horticultural Practice. — The probable importance of bud selection to American pomology was recognized by Munson in 1906. He advocated its practice in the following words: "Select through successive generations buds, that is cuttings or scions, from branches which bear fruit most nearly approaching the ideal." Two years later Webber presented the subject of ''clonal or bud variation" to the American Breeder's Association and also recommended bud selection. At this time, however, there was but little evidence that could be cited as proof of the practical value of bud selection. Working with violets Galloway and Dorsett were able to produce tlisease-resistant and florif- erous strains. The Boston Fern sports were beginning to attract at- tention. But it was still generally supposed that bud sports were of comparatively slight importance as far as pomological practice was concerned. About this time two of the pioneer citrus growers of Cali- fornia, E. A. Chase and J. P. Englehart, became interested in the numerous variations among their orange and lemon trees and the latter began to experiment with bud selection. He first recognized and propagated the Golden Nugget Navel Orange, a sport from the Washington Navel. He soon convinced himself that many of the variations in fruit characters which he observed could be perpetuated by budding. In 1910 Coit emphasized the fact that through the unintentional propagation of undesirable sports a gradually increasing proportion of the trees in the citrous orchards of California were developing into drones or worthless types and that the only way to prevent this was greater care in choosing the buds used in propagating nursery stock. Meanwhile, Shamel had begun an extensive series of experiments on the improvement of citrous fruits through bud selection. During his preliminary investigation in 1909 Shamel found what he took to be distinct types of Washington Navel Oranges and the observations and experiences of certain growers seemed to indicate that trees producing small annual yields and poor quality of fruit could be top-worked with scions from trees known to be annual high producers of superior fruit to the very great advantage of the citrous fruit growers. Accordingly an elaborate system was de- vised for recording the performance of individual orange, lemon and pomelo trees. Performance Records as a Basis for Bud Selection. — This study of the performance of individual citrous trees has emphasized the fact that there may be inherent differences with respect to quality and yield be- tween different trees of the same commercial variety. Already these tests have been supplemented by demonstrations in top-working low- producing trees with scions from fruitful individuals. As a result of the systematic campaign which the government representatives have con- ducted throughout the citrous fruit districts of California and Florida, 392 GENETICS IN RELATION TO AGRICULTURE many of the growers are keeping records of the annual yields of part or all of the trees in their orchards. Bud Mutations in Citrus. — What has been accomplished through bud selection in citrous fruits has been made possible by the relatively high frequency with which bud mutations occur. A dozen distinct types of Washington Navel orange are now known to occur more or less fre- quently in California orchards (Fig. 161): This fact is of especial sig- nificance in the light of the history of the Washington Navel orange which, as it originally existed in Southern California, consisted of only a few trees Fig. 161. — Fruits of the Washington Navel orange (1), and four forms that have originated from it by bud mutation; (2), Thomson Navel; (3), Yellow Navel; (4), Corru- gated; (5), Ribbed. {Courtesy U.S. Department of Agriculture.) (possibly only two) which were propagated from navel orange trees that were introduced from Brazil by the U. S. Department of Agriculture. According to Coit the evidence from early California horticultural literature indicates that the Washington Navel variety was recognized as a distinct and at least fairly uniform type of orange. That a strong tendency to mutation characterizes this variety is evidenced by the frequent origin of new forms or reappearance of old ones as bud sports. In some cases the aberrant type differs not alone in fruit characters but also in habit of growth or leaf-shape and frequently in yield. In fact high yield is said to be correlated with superior fruit at least in some types. Similarly in the Eureka lemon the so-called "shade-tree type" makes BUD SELECTION 393 rank growth and low yields, and such trees have been successfully top- worked with scions from fruitful types. Deciduous Tree Fruits. — The efficacy and practicability of bud se- lection in other species than the citrous fruits is not yet determined. But there is considerable evidence that in certain varieties at least it is possible to find distinct types which remain constant when propagated vegetatively. In the apple and peach bud sports are known and they may be more frequent than has been supposed. Dorsey mentions four varieties of apple which originated in this manner and reports the dis- covery of another sport. It is possible that certain varieties have a greater tendency to sport than others. The Baldwin apple may be such a variety. It is claimed by some authors that the many variations occur- ring in this variety are purely environmental, while others assert that they have propagated such variations successfully. But in the apple and most other deciduous fruits there are plenty of good varieties which are adapted to conditions in the regions at present devoted to fruit growing. Here, as with citrous fruits, new varieties are not needed so much as profitable orchards. Will it pay to keep performance records as a basis for bud selection in deciduous fruits? That is the critical point, and it is not yet settled. Both favorable and adverse evidence has been presented. The Virginia Station kept a record for four years of the yields of 1245 trees in the same apple orchard (variety or varieties and age not known). Of these 375 yielded an average of four barrels to the tree and produced 60 per cent, of the crop, while 215 trees averaged less than one barrel per tree and were kept at a loss. The Dominion Experiment Station, Ottawa, Canada, has kept records of yields of different trees in the same orchard covering a period of 18 years. The most productive tree of McMahon- White yielded 1250 gal., and the least productive, 882 gal. Of Patten Greening the most productive tree yielded 974 gal., and the least produc- tive 586 gal., while in the case of Mcintosh Red one tree yielded 1219 gal., and another 670 gal. Clearly there are wide differences in the produc- tivity of individual apple trees. Much of this variability in production is probably due to soil differences. It is claimed by Powell, however, that one cause of the marked decrease in the number of apple trees in New York State is the absence of profits due to low-producing trees. The true condition can be determined only by keeping performance records on an extensive scale. Even though many healthy but low-producing trees may be found, there still remains the question whether or not it will pay to top-work these low-producers with scions from high-producers. It has not yet been determined whether any of the wide differences in the productivity in individual apple trees are due to bud mutations. Next to yield, uniformity of fruit is perhaps the most important commercial consideration. But there are marked differences in varieties 394 GENETICS IN RELATION TO AGRICULTURE in respect to uniformity and in the extent to which this feature is trans- mitted to vegetative offspring. The experience of Tyson Brothers in Adams County, Pa., illustrates this point. They propagated 8000 trees with scions from two old trees of York Imperial apples which had been noticed because of their productivity and the uniformity in shape of their fruit. Unfortunately no scions were taken from average or poor trees, and hence there is no basis for comparing productiveness in the young orchard. Furthermore it is possible that these trees will exhibit less variation in form of fruits with increased age. But, as yet, so far as uniformity of fruit is concerned, the experiment seems to have been decidedly inconclusive. Progress with a similar experiment on the Ben Davis variety has been reported by Whitten. Scions were taken from an exceptionally poor tree and from another tree which produced the largest and best apples of its kind on the station grounds. Examination of the third year's crop showed no perceptible difference in size, color, grade or quality of the fruit from the two lots of trees. In fact the average yield per tree was somewhat higher in the lot propagated from the poor tree than in the lot propagated from the superior individual. There appeared to be as much variation between individual trees in each plot as between the two plots. "Pedigreed" Nursery Stock. — In response to the growing interest in bud selection many nurserymen have taken advantage of the idea of value which is commonly associated with pedigree. The more conscientious ones have selected their scions from trees which they believe to be supe- rior, but a certificate of source is not a pedigree. This term, it must be admitted, has been used in scientific investigations of vegetatively propagated plants, where careful records were kept for a relatively large number of asexual "generations," as in Stout's work on Coleus. But no nursery stock now on the market is entitled to be known as pedi- greed and even though such stock may be produced in future years, the danger from misrepresentation, either intentional or unintentional, will be as great as ever. Coit has suggested that stock propagated from tested trees be known as recorded stock and recommends a simple plan by means of which Deputy County Horticultural Commissioners may officially seal and record each tree when it is budded. Bud Selection in the Potato. — No one doubts the occurrence of bud mutations in the potato. Numerous instances of the origin of new varieties as bud sports are on record. Yet there is considerable difference of opinion regarding the relative frequency of bud mutations in this species. Numerous investigations have been made on the improvement of the potato by means of tuber and hill selections. The most important papers have been reviewed by East and, more recently, by Stuart. East observed over 700 named commercial varieties during a period of 3 BUD SELECTION 395 or 4 years and found 12 bud mutations. Changes were noted in the color, shape and habit of growth of the tubers and in the depth of the eyes. But as for the bearing of bud mutations on origin of new varieties East reached the conckision that, while isolated cases of improvement might be due to selection of bud mutations, yet comparatively few (probably less than 0.5 per cent.) of our present varieties arose in this manner. This evidence on the origin of varieties has led East to adopt the view that probably all bud mutations are so exceedingly rare in the potato that few, if any, cases of "running-out" or "degeneration" in varieties are to be explained on this basis. He beheves the principal factor in such dete- rioration is disease, and that in numerous experiments on potatoes, in Fig. 162. — Variation in yield between tuber-units from the same hill. Above, the progeny of two tubers from hill selection No. 35; below, that from hill selection No. 4. (After Stuart.) which it is shown that successive selections have raised the average yield over that of the unselected tubers, the results are entirely due to the elimination of diseased tubers. While the ehmination of diseased tubers or of tubers that were weak- ened by disease in the leaves or stem does undoubtedly explain the success of many selection experiments it may not account for all of them. Tests of individual tubers of almost any commercial variety apparently reveal inherent differences in the tubers. Although the plant is very susceptible to environmental conditions and some tuber characters such as shape and size are very unstable, yet sometimes the product of two closely similar tubers which came from the same hill when grown under closely similar conditions will differ widely (see Fig. 162) . The most satisfactory method of testing individual tubers is the tuber-unit method which was introduced by Webber. Each tuber which is to be tested is cut lengthwise into four equal pieces which arc planted at equal distances from each other. The four hills thus comprise a tuber-unit. 396 GENETICS IN RELATION TO AGRICULTURE Stuart in 1911 conducted a tuber-unit experiment with some 150 standard commercial varieties of potatoes. "The seed used was grown in Burhngton, Vt., in 1910, on land which had not grown a cultivated crop of any kind for at least 35 years. In addition to this the seed was selected from the most promising hills at the time the crop was harvested. The tubers as a whole were remarkably uniform in size and there could, therefore, have been little difference in the size of the seed pieces used. Any variation, therefore, which occurred between the flBHy^Ml 4li Mill Fig. 163.^Strong and weak tuber-units of the Gold Coin variety of potatoes. Nos. 1 and 2 represent strong and weak tuber-units in 1911; Nos. 3 and 4 represent yields from tuber-units 1 and 2; Nos. 5 and 6 represent yields in 1912 from 5 tuber-units of Nos. 3 and 4. {After Stuart.) plants of the various tubers which were planted would seem to be due to some inherent tendency in the tuber itself. The remarkable dissimilarity between the growing plants of the individual units of a variety planted contiguously in the row was so surprising that some three dozen units were photographed and when these were harvested the tubers were also photographed (see Fig. 163). It was found that the divergency in yield was just as great as in the size and vigor of the plants. In 1912 five units were planted from both strong and weak plants, and it was found in practically every instance that the low-yielding 1911 plants gave poor germination, a feeble vine growth and a still lower yield than in 1911." There have been many experiments similar to the one above de- scribed and they certainly indicate that a certain proportion (from 5 to BUD SELECTION 397 10 per cent, according to Stuart) of weak, diseased or unproductive plants are to be found in all unselected varieties. It is equally certain that in the vast majority of cases in ordinary field practice these are unrecognized and the resultant effect upon yield is unnoted. Even though lack of vigor and low yield is entirely due to disease and hill or tuber selection does nothing but eliminate these undesirables, it will be well worth doing. But other characters, as well as yield and vigor, should be kept in mind in attempting to produce an improved strain of potatoes. The ideal market tuber is of medium size, round or oblong in outline and somewhat flattened. The eye should be shallow. The eating quality should not be overlooked, but varieties and tastes differ greatly in this respect. In addition to these, adaptability to local conditions and disease resistance should receive attention. "The selection of a large number of high-yielding hills which are then thrown together for mass planting the ensuing year is not likely to result in any marked improvement except by the elimination of the diseased or unproductive plants. The only certain method of securing a superior strain is to plant each selection separately. . . . Every progressive farmer should have his selection plot, in which to grow his yearly selections; and, in addition, he should have his in- crease plot, where the promising selections may be increased for the field-crop planting" (Stuart). Certified Seed Potatoes. — ^The certification of seed potatoes based on official inspections during the growing season and after harvesting has been adopted in some states. According to Milward the summer in- spection considers stand, vigor of vine, specific and non-specific diseases and varietal purity; and bin inspection looks after conformity to type, diseases, market condition, quality and yield. In view of the increasing importance attached to disease in the degeneration of potato varieties some such system of inspection and certification should be adopted in every state where potatoes are extensively grown. But it must be borne in mind that complete protection against failure or loss is by no means assured even under a system of seed certification. Stewart has recently reported several instances of sudden degeneration of prohfic strains, at the Cornell University Experiment Station, through the invasion of some obscure disease of which there are a number that infest the potato. In some cases only the larger tubers in a hill are alTected while the smaller tubers are apparently healthy. Stewart's conclusions follow: "(1) Neither normal foliage nor high yield is a guaranty of productivity in the progeny of the following season. Degeneration may occur quite suddenly. (2) It is unsafe to select seed potatoes from fields containing many degenerate plants. Even the normal plants from such fields are liable to produce worthless progeny. (3) Mosaic threatens to become an important factor in the production 398 GENETICS IN RELATION TO AGRICULTURE of seed potatoes. It is transmitted through the seed. (4) It is doubtful if any method of seed selection will prevent the "running out" of seed potatoes under certain conditions." Oth€r Crops in Which Bud Selection May Apply, — It is claimed that many desirable varieties of roses, carnations, chrysanthemums, violets and other plants which are cultivated for their flowers originated as bud sports. The best florists are very critical regarding the character- istics of their stock and sports are soon discovered. The importance of propagating from typical, healthy plants is generally appreciated. Many of the roses used for forcing winter blooms produce two types of shoots which are known to the horticulturist as blind and flowering wood. For some years the Bureau of Plant Industry conducted ex- periments on the selection of buds from the two types of shoots, but it became apparent that the diversity among individual plants in regard to their flowering habits, whether propagated from blind or from flowering wood, was greater than the diversity between the progeny of flowering wood plants as compared with the progeny of blind wood plants. As a result of fertilizer experiments with the variety, My Maryland, Blake inferred that there was a real basis for production of improved strains by bud selection. But he points out that it would require time and much care in selection and that the average florist can hardly attempt to do more than to note the relative vigor of his plants at various stages and propagate from the best producers that are not especially favored by particular environmental conditions. The strawberry is so important commercially and comes into bearing so soon when propagated from offsets that if bud selection were effective in producing improved strains it would be of tremendous practical value. But the results of experiments indicate that the individual differences so frequently observed in strawberry plants are merely modifications. Whit- ten reports that bud selection of strawberry plants during a period of 15 years has given no gain in the total productiveness of the plots which originated from high-productive plants over the plots which originated from low-productive plants of the same variety. The experi- ment began by selecting six plants that yielded four times the amount of fruit of six low-producing plants all of the Aroma variety. Each succeeding year selections in the high-yielding plot were made from the highest plants and in the low-yielding plot from lowest-yielding plants. It is possible that in some varieties of strawberries bud mutations occur more frequently than in others. But in order to find a high-yielding plant whose high-producing character would be maintained among its vegetative offspring it would probably be necessary to test hundreds of individual high-producing plants. Hybridization offers much greater promise in the production of high-yielding strains of strawberries. liUl) SELECTION 399 Limitations of Bud Selection. — The efficacy of bud selection as a means of improving the type is dependent upon the occurrence of bud mutations; its practicability, upon their frequency. As a method of plant improvement bud selection will always be handicapped because recombinations of factors are possible only in sexually reproduced individuals. Moreover, it appears that in some vegetatively propagated crop plants desirable bud nuitations, which can be detected without resorting to statistical methods, are so rare that bud selection can never become a generally used method of producing new varieties, even though it may occasionally be used effectively for that purpose. On the other hand, it is highly practical to give careful attention to the selection of scions from such plants as are known to be healthy and typical of the variety. Such bud selection is a means of preventing the propagation of worthless or undesirable mutations and it should be practised by every nurseryman as a matter of course. CHAPTER XXIV BREEDING DISEASE-RESISTANT PLANTS The term, plant disease, has been restricted by some authors to those disorders and abnormaHties caused by fungous parasites only. Other authors have employed the term in a more general sense, including there- under all abnormal conditions of structure and function which are caused by the different elements of the environment. We shall use the term in this more general sense and for the purpose of this discussion it may be defined as follows. Plant diseases include all the ailments and injuries which can be traced to specific causes or agencies as well as certain func- tional disorders the causes of which are obscure or difficult of analysis. In order to discuss profitably the breeding of disease-resistant plants it is necessary to consider more fully the various categories of causes. The Causes of Plant Diseases. — In general the diseases of plants are caused either by unfavorable conditions among the inanimate elements of the environment or by the invasions of other organisms. While every case of disease must be considered as the result of interrelated causal agencies, yet it is usually possible to discover specific agents that are primarily responsible for the pathological condition. It is then possible to determine the nature of disease resistance in particular instances with more or less definiteness according to the nature of the specific causes. The most important non-living elements of the environment affecting the health and vigor of cultivated plants are the soil, the water supply and the temperature and humidity of the atmosphere. These environmental factors influence plant development in so many ways that the oppor- tunities for maladjustment between plants and their environment are practically endless. Such conditions as excess of alkali or lack of suffi- cient moisture in the soil or the combination of excessively high temper- ature and low relative humidity are typical and important illustrations of specific environmental conditions which induce disease in plants. The living organisms of chief importance in causing plant diseases are insects, fungi and bacteria. Injurious insects may be roughly classified according to their ways of feeding under two heads, viz., sucking and biting insects. The first class includes the plant lice, phylloxerans and scale insects which obtain their nourishment by sucking it from the living plant. The second class includes all moths and butterflies whose larvae devour living plants as well as beetles and other insects that obtain their 400 BREEDING DISEASE-RESISTANT PLANTS 401 food in similar fashion. Pathogenic fungi and bacteria are wholly or partially parasitic. Bacteria which cause plant diseases are those capa- ble of establishing themselves and multiplying in number within the living tissue of the host, A few of the important plant diseases caused by bacteria are "fire-blight" of pears and apples, crown gall of many fruit trees, grapes and other plants, and the black rot of the cabbage. Some fungi, such as rusts and smuts, are strictly internal obligate parasites (as distinguished from those obligate parasites which are wholly or partially epiphytic), i. e., they cannot exist outside the body of a particular host plant or plants except in the spore stage. In such cases the relation between parasite and host is symbiotic. The specific re- lations between parasites and their hosts vary from a condition of tol- erance of the parasite without serious injury to the host to one in which the destruction of the host finally ensues. Many fungi, such as the powdery mildews, are epiphytic although they derive their nourishment from the living plant tissue by means of haustoria. Between the epi- phytes on the one hand and the internal parasites on the other are many types of endophytic fungi in which various proportions of the parasite's life cycle are spent within the host plant, Thiis there are many agencies, some non-living as well as many living things, which threaten the normal development of cultivated plants. Even among the parasitic fungi themselves there are many devices for invading the host plant and many instances of specific physiological relationship between parasite and host. The Nature of Disease Resistance in Plants. — Disease resistance in a plant may be defined as the ability to develop and function normally under conditions such that other plants of the same species fail to develop or are destroyed. Resistance is always either partial or complete. The avoidance of disease by such means as precocious or delayed maturity is hardly to be considered as true disease resistance. Since there are so many agencies which may cause disease in plants it is evident that the ability to resist disease may depend on any one of many characters or it may involve every function of the plant. In either case this ability is a manifestation of the physiological individuality of the plant and hence it may be inherited. Nowhere is this more strikingly shown than in the disease resistance of certain natural species. Disease Resistance in Natural Species. — The nature of disease resist- ance in a particular instance is indicated by the nature of the cause of the disease. In the case of non-living causes resistance on the part of certain plants can be explained only as a manifestation of the inherent properties of the protoplasm. Thus the alkali resistance of salt grass, the Austrahan salt bushes, the common beet and asparagus is a heritable character. If it were not so these species could not perpetuate themselves on soils which 26 402 GENETICS IN RELATION TO AGRICULTURE are too strong in alkali content for most plants. Similarly with many plant troubles that are referred to adverse soil conditions, such as chlorosis and die back, it has been found that some species are much better able to resist such conditions than other species and within a particular species certain varieties may be more resistant than other varieties. This holds true in the case of other non-living agencies such as excess and deficiency of moisture and heat. For every plant there is a set of optimum condi- tions and these conditions are very different in different species and among varieties of the same species. For example, rice flourishes in standing water while maize requires well aerated soil. But there are thousands of varieties of rice, each one adapted to the conditions peculiar to a certain locality and there are many varieties of maize which make possible the culture of this species under conditions varying from the humid corn belt to the arid regions of northern Mexico, Bolivia and central China. Similarly in other field crops and in fruits, in various parts of the world there exist species and varieties which are adapted to certain local condi- tions that would be inimical to normal development of related species and varieties. Agricultural exploration cooperating with systematic seed and plant introduction has already made available for the plant breeder a large number of distinct forms of economic plants which in course of time may revolutionize many productive and manufacturing industries. Turning now to the phenomena of resistance to the attacks of animal or plant parasites, we find that natural species are characterized by as great diversity in this respect as was observed in the case of resistance to alkali, drouth and other physical elements of the environment. A few specific examples will serve to illustrate this general principle. The relation of different species of the grape to the phylloxera, Peritymhia vitifolice Fitch {Phylloxera vastatrix Planchon), is representative of a great number of reported instances of insect parasitism on vegetation. Also in their general aspects the phenomena of variation in phylloxera resist- ance among species of the vine are representative of the facts of disease resistance in general. Moreover, on account of the great economic importance which this particular vine disease assumed in Europe some forty years ago, and later in Calif o-rnia, there has been a large amount of investigation on the culture of grapes in phylloxera infested regions. The life cycle of this insect includes both leaf-feeding and root-feeding forms. The extent of the injury caused by the warty galls on the leaves is com- paratively insignificant. It is the root-feeding form which inflicts serious damage to susceptible vines. On the roots of such vines the character- istic symptoms are of two distinct kinds, viz., small galls or "nodosities" near the tips of young rootlets, and larger swellings or "tuberosities" occurring upon the older rootlets and roots (Fig. 164) . The root-tip galls or nodosities are commonly found even on resistant species if phylloxera are BREEDING DISEASE-RESISTANT PLANTS 403 present. The principal diiTerence between resistant and susceptible vines as regards reaction to phylloxera attacks is found in the number, size and penetration of the lesions on the larger roots. This phylloxeran is a native of eastern North America and the species of Vitis which are native to this region all exhibit some resistance to its attacks. This resistance of species native to the habitat of a disease-causing parasite is a general fact of great significance to agriculture on account of its potential value in both plant and animal breeding. Fig. 164. — Effects of phylloxera on vine roots. On left affected root tips or nodosities; in same figure incipient tuberosities are shown at a. Center, non-penetrating tuberosities on an American vine. Right, penetrating and confluent tuberosities on V. vinifera, the most serious form of the disease. {After Viala and Ravaz.) • The phylloxera was introduced into France through the importation of American vines and it soon became a most serious obstacle to the culture of the choice wine, table and raisin grapes of the Mediterranean region, all of which varieties belong to a single species, Vitis vinifera. In fact, every member of this large and valuable plant group has been found to be susceptible to phylloxera thus making impossible its culture as a direct producer, i.e., on its own roots, in a phylloxera infested region. After striving in vain to exterminate the insect in all infested areas, 404 GENETICS IN RELATION TO AGRICULTURE European vineyardists gradually adopted the only other practicable method of grape growing, viz., the grafting of vinifera varieties upon resistant roots. The problem of determining which species of Vitis were both highly resistant to phylloxera and well adapted to the soil and climatic conditions of various European localities required extensive investigations. Eighteen native American grapes have been tested as well as several Asiatic species, but the latter were all less resistant than the most susceptible American species. The American vines which have come into most prominence on account of their proven value in the reconstitution of phylloxera devastated vineyards may be listed according to relative resistance about as follows, if the maximum or absolute inimunity be taken as 20. 18-19. V. rupestris. 18. V. riparia and cordijolia. 17. V. herlandieri. 16. V. cinerea. 14-15. V. cestivalis, linsecomii and candicans. All of the above species belong to the sub-genus or section, Euvitis. Two of these, rupestris and riparia, together with certain hybrids between these and between these and vinifera, are now considered the most valuable resistant stocks. Another American species belonging to the section Muscadinia, viz., rotundifolia, has been found to have a resistance of 19 or higher inasmuch as the insect has never been observed on its roots. It is also free from the common fungous diseases of the vine, but the difficulty of propagating it from cuttings and its slight affinity for grafts of other species make it a valueless species for the reconstitution of vineyards. On the other hand, the American species, labrusca, has become of great economic importance since it is the parent of the Concord, Isabella, Niagara and many other cultivated varieties. Yet its resistance to phylloxera is ranked at 5, and when grown in Cali- fornia it is no more resistant than is calif ornica when used as a rootstock for producing vines, and the resistance of the latter is ranked at 4. Yet the labrusca derivatives are extensively grown in the northeastern states and in other northern temperate regions. This is explained by the fact that the phylloxera itself, does not thrive below a certain minimum temperature. Thus we find that resistance to phylloxera in the species of Vitis varies all the way from zero in vinifera to practically absolute resistance in rupestris, rotundifolia and certain hybrids and that the existence of highly resistant forms which are also suitable for vineyard culture has made possible the preservation of an important agricultural industry. The question of the nature of the cause of resistance to phylloxera has received rather wide attention among investigators, but it has not BREEDING DISEASE-RESISTANT PLANTS 405 yet been definitely answered. According to Ravaz, a chemist has thought to measure resistance by the amount of resinous principles in the roots; a physician by the relative duration of the roots; an anatomist by the relative thickness of the medullary rays; but all these explanations have failed to withstand investigation. Foex states that resistance was first thought to be due to great vigor, large root development and ease of production of new roots but that this was insufficient since some vines of small vigor, like Vitis monticola, are resistant while others of great vigor are susceptible. Foex, himself, traces a relation between the thickness and succulence of the bark of the root and susceptibility. There is also a theory, which originated in Italy, that resistance is due to acidity of the sap and the degree of acidity is highest in seedling plants and in clones which have recently come from seedlings, the acidity decreasing with the age of the variety. But this is contradicted by the fact that vinifera seedlings are quite as susceptible as their parents. Variability in resistance of several varieties of grape when grown in different infested localities is aecepted by Grassi as evidence of the existence of "benignant" and "malignant" races of phylloxera. But this does not explain the liigh resistance or immunity of some American species. Having in mind the fact that the phylloxera sucks its nourishment from the leaf or root by inserting its prolonged rostrum into the living tissue, it seems most probable that resistance is to be explained as absence of response to a specific stimulus. The many remarkable instances of hypertrophy in vegetative tissues due to wounds inflicted by insects can be explained satisfactorily only by assuming that the insect injects something into the wound which causes abnormal functioning of the affected parts. If this occurs in the case of phylloxera then resistance consists in failure of the wounded tissue to respond to the foreign element injected by the insect. Such failure of response might be due either to the absence of a particular substance which reacts so as to stimulate growth or to the presence of a specific anti-body which counteracts the effect of the insect's poison. The latter seems the more probable condition in view of what is now known concerning immunity in general. The complete susceptibihty of V. vinifera would then be due to absence of the anti-body. But the absolute resistance or complete immunity of V. rotundifolia may be caused by the presence of a sub- stance which is actually repellent to the insect itself. At any rate, the fact that we are dealing here with distinct natural species makes it rea- sonably certain that resistance and susceptibility to phylloxera infesta- tion are somatic expressions of genotypic diversity. Another important case of variation in disease resistance among species of the same genus is found in the relation of various chestnuts to the very destructive bark disease caused by the fungus, Endothia 406 GENETICS IN RELATION TO AGRICULTURE parasitica. The parasite is a native of eastern Asia where it is parasitic upon native species of chestnut, to which it appears to do relatively little harm. In other words these species are highly resistant to the parasite. However, when the fungus was introduced into America, pre- sumably in nursery stock some 25 years ago, it found in our native species, Castanea americana, a very susceptible host (Fig. 165). The parasite has already caused the destruction of the American species throughout the northern Appalachian region and is strongly threatening its complete extinction as a timber tree. Investigations have determined that the Euro- FiG. 165. — An advanced stage of the chestnut bark disease, caused by Endothia parasitica, a virulent pathogenic fungus from China. {From the Journal of Heredity.) pean chestnut is also susceptible to the attacks of this fungus, so that the future existence of this species is also jeopardized. The American chest- nut is one of our most valuable forest trees and its destruction will entail an enormous loss. A very promising Chinese species is known pro- visionally as C. mollissima. While it is scarcely a timber tree as compared with our native species, yet it may thrive in our climate. As the nuts are of good quality and the tree has shown marked resistance to experi- mental inoculations on plants already established in this country it is hoped that it will prove to be a successful substitute for the vanishing American species. Even the culture of the American species for com- mercial nut production in western North America will be constantly threatened. Hence it is fortunate that bTeeding experiments with the BREEDING DISEASE-RESISTANT PLANTS 407 chestnuts have ah-cady })ceii under way some 20 years. To these we shall refer again. A bacterial organism which finds similar wide diversity in the resis- tance of possible hosts is the fire-blight pathogene, Bacillus amylovorus. Being indigenous in eastern North America, this organism must have maintained itself on the native species of apples and related genera l)revious to the introduction of European apples, pears and quinces since it cannot survive long even in the dead tissues of the host. The dis- ease is spread naturally by insects that visit infected plants; it may also be carried on pruning tools. Fire-blight is the most widely de- structive of all pomaceous fruit diseases; but the pathogene manifests different degrees of virulence in different species. Its most susceptible hosts are the commercial varieties of the pear, which are all derivatives of the European species, Pyrus communis. In several regions naturally well adapted for pear growing the culture of this fruit has been abandoned on account of the destructiveness of pear-blight. Even the more resistant varieties of communis as well as certain hybrids between communis and other species, such as the Kieffer, a supposedly resistant pear, have all proven to be susceptible to the disease when grown in the humid climate of the southern States. Therefore in discussing the problem of blight- resistance in pears it must be remembered that the pathogene itself is very susceptible to environmental conditions and that a particular host which is known to be resistant under one set of conditions will not neces- sarily prove to be generally resistant. Hence the breeding of blight- resistant pears should be carried on in a region ideal for pear culture in every respect except that it is ideal for the fire-blight organism also. Such conditions exist in southern Oregon where Reimer has made a complete collection of the known species of pears and has conducted scientific tests of their resistance to blight by means of inoculations with pure cultures of the bacillus. The results to date indicate that the fol- lowing species are highly resistant: Pyrus sinensis, P. ovoidea and P. variolosa. Under P. sinensis he finds there are several distinct species which will be classified after they have fruited, but they are all resistant. The birch-leaved pear, P. betulifolia, which is used as a stock in China, proved susceptible when the inoculations were made on 1- and 2-year old trees. But it is probable that older trees will show greater resistance and the same may be said of the 16 other species in which inoculation established the disease and which might be considered as susceptible varieties. However, varying degrees of susceptibiUty were exhibited by these species. Hansen reports the birch-leaved pear as quite resistant to blight in South Dakota where it has grown for over 20 years. We have now considered one plant disease which may be considered typical of each of the three great classes of disease-causing organisms and 408 GENETICS IN RELATION TO AGRICULTURE in each we find the same diversity among natural species as regards disease resistance. It is unnecessary to multiply instances further. In all likelihood the resistance of the Chinese chestnut to Endothia parasitica and of the Chinese Sand Pear to the fire-blight bacillus is due to some specific quality of the protoplasm probably something in the nature of an antitoxin. That this quality is heritable will be seen from the results of hybridization experiments. Breeding Disease -resistant Varieties by Hybridization. — Allusion was made in Chapter XX to the fact that first generation maize hybrids are often more drouth resistant than either parent. Presumably this is merely one manifestation of heterosis. Hybridization is a very im- portant means, however, for the production of improved varieties which are better adapted to specific adverse elements of the environment. Witness the important results already secured in the production of cold- resistaiit varieties of fruits, grains and forage plants, by Hansen, Patten and Saunders and at the U. S. Agricultural Experiment Stations in Alaska. At one stage in the anti-phylloxera campaign in France and California viticulturists held definitely to the ideal of securing through hybridi- zation "a vine that, while resisting the phylloxera, the two mildews, the black rot, etc. (all of which diseases are natives, and which the American vines resist more or less well), will give without grafting a grape that has size and the quantity and quality of the Vitis vinifera." With this object in mind many crosses were made but they have produced no hybrids between vinifera and American species that can be substituted for the choice vinifera varieties. It, therefore, became necessary to util- ize resistant species and hybrids as stocks on which to graft the pro- ducing varieties. However, it is still possible that, by growing large numbers of F2 and Fs seedhngs from some of the most promising Fi hybrids, the dream of the viticulturist might be reahzed. It seems that no grape breeders have carried out extensive tests of hybrids beyond the first generation from the cross. This is not strange inasmuch as grape breeding for phylloxera resistance was at its height during the latter part of the 19th century and before the importance of testing for several generations after a cross was generally appreciated. That phylloxera re- sistance and susceptibility are conditioned by specific genotypic elements is evidenced by the results of Rasmuson who tested F2 seedlings from several crosses between certain American species and between American species and V. vinifera, as well as crosses between different varieties of vinifera. The latter, he reports, yielded only susceptible offspring while the crosses between different American species gave both resistant and susceptible offspring, the latter being in the minority. Resistance appeared to be dominant and susceptibility recessive in the progeny of BREEDING DISEASE-RESISTANT PLANTS 409 crosses between American species and vinifera. The data are not given but he believes the observed numbers of resistant and susceptible vines favor the assumption of two factors that condition immunity when either is present alone or when both are present together. Fig. 166. — a, Sandcherry, Primus besseyi; B, Wyant plum, P. americana; C, D, F2 hybrids from Sandcherry X Wyant. {After Beach and Maney, Iowa A. E. S.) Resistance to aphis in the stone fruits is thought to be a heritable character from the result of crosses made at the Iowa Experiment Station (Fig. 166). The data permit no reliable conclusions regarding the geno- 410 GENETICS IN RELATION TO AGRICULTURE typic relation of aphis resistance and susceptibility in these plants, but the indications are that these characters are conditioned by a single factor difference. Another interesting case of the inheritance of resist- ance to aphis was observed by Gernert in Fx hybrids between teosinte and corn. Both the corn root-aphis, Aphis maidiradicis, and the corn plant- aphis, A. maidis, were involved, and both the teosinte and the hybrids were completely resistant while the corn was badly infested. The desira- bility of securing aphis resistant varieties of maize will be apparent when it is realized that most of the corn growing regions of North America are infested with these insects and that the loss in reduction of yield caused by them is enormous. The work of Van Fleet on hybridizing various species of chestnuts was begun 10 years before the terrible bark disease had worked havoc Fig. 167. — In the center is a nut produced by a cross between the American bush chinquapin, Castanea -pumila, (right), and the Japanese chestnut, C. crenata, (left). Al- though intermediate in size the hybrid nut is disease resistant and of good quality. (From The Journal of Heredity.) with the chestnut trees near New York City, which is the oldest known center of infection. Hence many crosses were made with either the American or European chestnut as one parent, but in 1907 these were all destroyed by the Endothia. Fortunately however numerous controlled polhnations were made on the bush or Virginia chinquapin, Castanea pumila, using pollen of a Japanese species, C. crenata, as well as other Asiatic chestnuts. It is asserted that the Asiatic species and the chin- quapin-Asiatic hybrids are highly resistant, because few have shown any appearance of infection although surrounded by diseased trees, and that even when infection takes place the injury is quite local in character. Van Fleet adds that second generation seedlings of chinquapin-crewa^a crosses show no disease although constantly exposed to infection (Fig. 167). Thus a beginning has been made in what promises to be an import- ant branch of nut breeding, and the orchard production of commercial chestnuts has been insured against future encroachments by a deadly disease through the timely efforts of a zealous and far-sighted plant breeder. , ' BREEDING DISEASE-RESISTANT PLANTS 411 In an attempt to breed blight-resistant pears of horticultural value Hansen has produced and distributed for trial thirty-nine first generation hybrids between various commercial varieties and either the Chinese Sand Pear or the Birch-leaved Pear. Should these hybrids prove to be unsuitable as commercial varieties they may be used as foundation stock in further efforts to produce a hardy, blight-resistant variety. Although the Kieffer and the Le Conte are presumably Fi hybrids between sinensis and communis, they have not been used by Hansen because they are not hardy in the north. For lower latitudes however these two partially resistant varieties should be utilized not only by raising seedlings from them but also by an extensive series of crosses especially with other partially resistant communis derivatives of high quality such as the Seckel. The work of Reimer and of Hansen indicates that perfectly resistant stocks may be developed which are adapted for each important pear-growing region. If to this achievement may be added the creation of fairly resistant varieties of really excellent quality, the worst diffi- culties in pear production will be removed and the world's supply of this delicious fruit will be practically assured. Creating Rust-resistant Commercial Wheat by Crossbreeding. — The grain rusts are the most important of all fungous plant diseases. The annual losses they entail for the grain crops of the world must be estimated in the hundreds of millions of dollars. Although prevention of wheat rust to some extent is now possible by giving careful attention to the water and soil relations of the wheat plant and by early seeding or the planting of early varieties which sometimes escape attacks by rust, yet these diseases still remain a serious menace to the maximum production of wheat. Hence, the creation of rust-resistant varieties has become a very important problem. The diversity among varieties of wheat as regards resistance and susceptibility to rust fungi was recognized by Knight in 1815 and the desirability of creating new varieties which should be resistant to rust as well as highly productive and of good milling quality was fully realized by such breeders as Pringle, Blount and Farrer. Although they were not famihar with the Mendelian principles of seg- regation and recombination of characters, these breeders of wheat, a self-fertilized annual crop plant, were naturally led to persist in their efforts beyond the Fi generation. The work of Farrer expecially was thorough and reliable. He found that he could not secure absolute resistance to the black stem-rust, Puccinia graminis Pers., combined with good milling quality in his wheat crosses even when rigorously selected in the F2 or "wild" generation as he called it. Most of the soft bread wheats are very susceptible to rust and, when crossed with the resistant durums, poulards and spelts, they give rise to strains which are either poor bread wheats or are rust susceptible. Biffin discovered in 1903 412 GENETICS IN RELATION TO AGRICULTURE that resistance to the yellow rust, Puccinia glumarum Eriks. & Henn., in his cross between RiVet, a slightly susceptible wheat and Red King, a very susceptible variety, was recessive in the Fi generation but appeared in approximately one-fourth of his F2 population. Tests of later genera- tions proved that this character bred true. Eriksson tested Biffin's work and found only slight variations in the F2 ratio and in the intensity of the resistance. However, it appears that resistance of the wheat plant to other species of rust fungi may be inherited as a dominant char- acter. Vavilov reports that he crossed Persian wheat, Triticum vulgare var. fuliguosum Al., which alone out of 540 varieties was immune to mildew, Erisiphe graminis DC, but which was susceptible to brown rust, Puccinia triticina Eriks., with other varieties of common bread wheats and secured Fi hybrids which were immune to both diseases. Thus it is clear that the inheritance of rust resistance is dependent upon the specific relation existing between the parasite and the host. The practical aspects of breeding rust-resistant cereals is greatly complicated by the fact that resistance in a single variety of wheat, for example, is likely to vary geographically. While this is due in part to the responsiveness of the wheat plant to radical changes in environment, it is probably more often due to physiological variations in the rust fungi. The virility of a given parasite appears to vary not only with the host but with the geographical location. A striking example of this was observed by .Mackie in the behavior of Kubanka, a durum wheat of Russian origin. Although this wheat is remarkably rust resistant in the northern Great Plains region, yet when grown on the west coast of Mexico it succumbed completely to the stem rust {Puccinia graminis var. tritici) which it had resisted successfully in the Dakotas. The ex- planation of this failure of a supposedly resistant wheat is found in the existence of local physiological races of the species P. graminis. Thus Freeman and Johnson found P. graminis var. tritici, which is supposedly confined to wheat, attacking barley and rye as well. The same results were obtained with oat stem rust, P. graminis var. avence, which readily attacked barley but was less virulent on wheat and rye. The stem rust of barley was found to be most readily transferred to the other cereals. In addition to the barberry numerous wild grasses serve as hosts of the stem rusts which fact still further complicates the problem of breeding for rust resistance. Starkman and Piemeisel have investigated the rusts of about 35 species of grasses and have found six distinct biologic forms of this species of rust, one of which came from an isolated area. Among other important discoveries, they found that more than one biologic form may occur on the same host in nature, sometimes even on the same plant; that these biologic forms can be distinguished from each other morphologically as well as parasitically; that different strains of the same BREEDING DISEASE-RESISTANT PLANTS 413 biologic form sometimes differ in degree of virulence on the same host; and that all gradations in susceptibility occur among the hosts, from complete immunity to complete susceptibility to various biologic forms. Finally, it must be remembered that but little is yet known about the nature of rust resistance. That it is in no wise dependent upon morphological characters appears to be well established. Carleton has pointed out that biochemical investigations are needed in connection with this problem. The recent investigations of Wagner on hydrogen ion concentration and natural immunity in plants representing four genera including the potato resulted in the conclusion that the variation in hydrogen ion concentration in plant tissues is a phenomenon of reaction to the injection of pathogenic bacteria. Also that the course and end results are related to the susceptibility of the plant in question and to the character of the disease as acute or chronic. An investigation now in progress at the University of California (by W. W. IMackie) seems to indicate that there is positive correlation between degree of acidity as indicated by the concentration of hydrogen ions and degree of resistance to P. graminis in wheat. A similar investigation of the species of Bromus in relation to the physiological races of the Corn and Grass Mildew, Erysiphe graminis DC, as reported by Salmon would be highly desirable. Whatever the nature of the resistant quality may be, there is no question regarding its heritability. But in view of the complicated nature of the problem which we have briefly outlined it would appear to be inevitable that the utilization of resistant varieties of wheat must be confined to limited areas in which adequate tests have proven their adaptability. Inheritance of Disease Resistance in Other Plants. — The conclusions we have reached in respect to rust resistance hold good in a general way for other parasitic plant diseases. In addition to the typical cases already described brief reference may be made to other notable examples of the successful creation of disease resistant varieties by hybridization and subsequent selection. The next case, however, will be considered somewhat in detail because it serves as an excellent model in method of procedure. The ravages of a group of wilt diseases caused by closely related fungi of the genus Fusarium have been checked through the successful efforts of the United States Department of Agriculture. As reported by Orton these are the cotton wilt, Fusarium vasinfectum Atk., the cowpea wilt, F. tracheiphilum Erw. Sm., and the watermelon wilt, F. niveum Erw. Sm. It is clear that these fungi possess a high degree of adaptation to the parasitic mode of existence. Also that, while the cause of resistance-in certain varieties of the host species is not fully established, yet the resistance itself is a physiological quality. No constant mor- phological differences have been detected between immune and suscepti- 414 GENETICS IN RELATION TO AGRICULTURE ble plants; neither are there observable differences in time of germination, rate of development or period of maturity. Furthermore, the resistance is specific; varieties that resist the wilt may be susceptible to bacterial blight and vice versa. That wilt resistance is a heritable character was strikingly dem- onstrated by Orton's creation of a wilt-resistant edible watermelon, Citrullus vulgaris. All watermelons appear to be very susceptible to Fig. 168.— Parent and product