'/■ MORPHOLOGY OF ANGIOSPERMS MORPHOLOGY OF ANGIOSPERMS (MORPHOLOGY OF SPERMATOPHYTES. Part II) BY JOHN MERLE COULTER, Ph. D. HEAD OF DEPARTMENT OF BOTANY, THE UNIVERSITY OF CHICAGO AND CHARLES JOSEPH CHAMBERLAIN, Ph. D. INSTRUCTOR IN BOTANY, THE UNIVERSITY OF CHICAGO ILLUSTRATED NEW YORK D. APPLETON AND COMPANY i Qog Copyright, 1903 By D. APPLETON AND COMPANY PREFACE 1st 1901 -we published the first part of a work entitled Morphology of Spermatophytes, containing an account of the Gymnosperms. At that time it was our purpose to issue as a second part an account of the Angiosperms, which would also contain a complete index of the whole work. We have become convinced, however, that such an association of these two great groups would help to emphasize a relationship that does not exist, and that Gymnosperms and Angiosperms should be treated as independent groups, coordinate with Pteridophytes. Therefore, the present volume is issued, not as Part II of Morphology of Spermatophytes, but as an independent volume entitled Morphology of Angiosperms; and any subsequent edition of the previous volume will be entitled Morphology of (iymnosperms. This volume, as the preceding, has grown out of a course of lectures accompanied by laboratory work, given for several successive years to classes of graduate students preparing for research. It seeks to organize the vast amount of scattered material so that it may be available in compact and related form. While careful attention has been given to citations, so that the student may know the groups that have been inves- tigated, and be put in touch with the original papers, the work is in no sense a compilation. The ground has been traversed repeatedly, for several years, by various members of the botan- ical staff and by numerous students, and their results have served to check current statements, as well as to contribute no small amount of new material. Vl MORPHOLOGY of angiosperms Any one who has attempted to review the literature of the morphology of Angiosperms will appreciate the great amount of labor it involves, as well as the chaotic condition of termi- nology and citations. There is nothing more baffling than the attempt to follow the guidance of the meager, indefinite, and often incorrect citations of the standard texts. It is believed, therefore, that the attempt to reduce the numerous contribu- tions to a consistent terminology and to make the citations fairly representative of the subject as well as definite and accu- rate will be of some real service to students of morphology. The volume, therefore, seeks to give to the advanced student a continuous account of the structures involved, and to the research student the details of groups and bibliography that he needs. In every case where figures have been copied, acknowledg- ment is made and a reference is given to the original paper containing the illustration. It should be noted that much in- formation included in the legends does not appear in the text, so that in any thorough reading of the book the legends should be included. The bibliography pertaining to each subject is printed in chronological order at the end of each chapter con- taining numerous citations. At the close of the volume all of the cited bibliographv is brought together, arranged alphabet- ically by authors. It would be too large a task to include a complete bibliog- raphy of such a subject, but we have presented what may be regarded as a full representative bibliography, containing, so far as we know, all of the most important contributions. In the very nature of things, some citations may have been omitted that should have been included, but there has been no intentional neglect. Xo attempt is made to present the details of floral structure, so fully described by the earlier morphologists and taxonomi-r-, since they are easily accessible in numerous texts. Xor have we ventured to enter the old and extensive field of anatomy, although many of its details are pertinent to morphology. In PREFACE vn its later development, however, it has contributed so many important data essential in any discussion of phylogeny that we have asked Professor E. C. Jeffrey to present the general outlines of the subject in the last two chapters of this volume, a discussion which includes both Gymnosperms and Angio- sperms. It is hoped that this presentation will help to stimu- late the cultivation of an important field of research too much neglected in this country. It did not seem necessary to treat the two great groups of Angiosperms separately. They are so similar in their essential morphological features that their separate presentation would have involved a needless amount of repetition. We have also continued to regard the spore mother-cell as the end of the sporophytic generation, and its division as the beginning of the gametophyte. The reasons for this are more fully presented in the present volume than in the preceding. In the chapters upon classification we have presented the scheme elaborated by Professor Engler, believing that it is the best expression of current knowledge of relationship ap- plied to the whole group, and that it is suggestive of the most critical regions for research. This has not been pressed to the dreary details of minor groups, for these are easily accessible. It has rather been our intention to present the general ideas involved in the alliances of first rank, so that principles rather than details may be prominent. We have also thought that the special student should be somewhat familiar with the his- tory of the group, so far as known, its geographic distribution, and the current notions as to its phylogeny. The last subject may be regarded as more theoretical than profitable, but the final aim of morphology is a definite phylogeny. and advance toward it must be made by a succession of theoretical con- clusions. Johx M. Coulter. Charles J. Chamberlain. The University ok Chicago, January, 1903. CONTENTS CHAPTER PAGE I. — Introductory 1 Angiosperms and Gymnosperms contrasted. 1 — Spermatophytes not a natural group, 3 — Monocotyledons and Dicotyledons con- trasted, 4. II.— The Flower 8 Definition of a flower, 9 — Origin of floral leaves, 9 — Tendencies in the evolution of the flower, 10 — Organogeny of the flower, 16 — Dioecism, 20 — Morphology of floral members, 22 — Stamen, 23 — Carpel, 24. III. — The Microsporaxgium 27 Origin from periblem, 27 — Cauline microsporangia. 28 — Number of microsporangia, 29 — Time of formation, 30 — Development, 32 — Archesporium, 32 — Parietal layers, 34 — Tapetum, 36 — Mother-cells, 38 — Dehiscence, 41 — Line of demarcation between sporophyte and gametophyte, 41. IV. — The Megasporangium 46 Origin from periblem, 46 — Cauline ovules, 46 — Foliar ovules, 50 — Morphological nature of ovule, 51 — Time of development of megasporangia. 52 — Development of ovule, 53 — Archesporium, 57— Parietal cells, 62— Mother-cell, 66. V. — The Female Gametophyte 71 The tetrad, 71 — Number of megaspores, 76 — Reduction of chro- mosomes, 80 — The functioning megaspore, 84 — Number of em- bryo-sacs, 86 — Germination of megaspore, 87 — Variations in history, 89 — Egg-apparatus, 93 — Svnergids. 94 — Fusion of polar nuclei, 95 — Antipodal cells, 96 — Enlargement of embryo-sac, 103 — The nutritive jacket, 103 — Haustoria, 104 — The mechanism for nutrition, 108. VI. — The Male Gametophyte 121 The tetrad, 121 — Number of microspores. 125— The nuclear divi- sions of the pollen mother-cell, 126 — The microspores. 131 — Ger- mination of microspore, 132 — Division of generative cell. 135 — The male nuclei, 136. is X MORPHOLOGY OF ANGIOSPERMS CHAPTER pAGg VII. — Fertilization 143 Historical resume, 143 — Development of pollen-tube, 146 — Chala- zogainy, 14y — The pollen-tube within the embryo-sac, 151 — Dis- charge of pollen-tube, 152 — Fusion of male and female nuclei, 153 — Centrosomes, 153 — Double fertilization, 155 — Male cell and male nucleus, 160. VIII. — The Endosperm 165 Contrast between Gymnosperms and Angiosperms, 165 — The fusion nucleus, 166 — Endosperm without fusion, 166 — Endo- sperm and pollination, 167 — Division of fusion nucleus, 169 — Two methods of endosperm-formation, 171 — Function of endo- sperm, 179 — Xenia, 179 — Morphological character, 181 — Nature of triple fusion, 182. IX. — The Embryo 187 Monocotyledons, 188— Alisma type, 188 — Pistia type. 192 — Liliuin type, 193— -Orchid type, 194 — Dicotyledons, 196 — Cap- sella type, 199— Other types, 200 — Degree of development, 205 — " Pseudo-monocotyledons," 206 — Phylogeny of the cotyledon, 208 — Parthenogenesis, 210 — Polyembryony, 213. X. — Classification of Monocotyledons 227 Spiral series, 228 — Cyclic series, 234. XI. — Classification of Archichlamydeae 240 XII. — Classification of Sympetalae 252 XIII. — Geographic Distribution of Angiosperms 261 Monocotyledons, 262 — Archichlamydeae, 266 — Sympetalae, 268. XIV. — Fossil Angiosperms 272 Monocotyledons, 272 — Dicotyledons, 276. XV. — Phylogeny of Angiosperms 280 Are Angiosperms monophyletic f 280 — Relation to Gymno- sperms, 283 — Relation to Pteridophytes, 285 — Theories of alter- nation of generations, 288 — Theory of the strobilus, 288 — The mutation theory, 292. XVI. — Comparative Anatomy of the Gymnosperms and their Allies 296 Pteridophytes, 296— Cycadofilices, 300— Cycadales, 304— Ben- nettitales, 306 — Cordaitales, 307 — Ginkgoales, 307— Coniferales, 308— Gnetales, 310. XVII. — Comparative Anatomy of Angiosperms 311 Dicotyledons, 311— Monocotyledons, 314. MORPHOLOGY OF ANGIOSPERMS CHAPTER I INTRODUCTORY Theke is a very large element of uncertainty in a presenta- tion of the special morphology of Angiosperms, chiefly because of the vast amount of unstudied material, but also because of the inequality in the accuracy and definiteness of the work done. However, the general outlines seem to be fairly well established, and their filling in must long occupy morphologists. Although two very distinct groups of Angiosperms are recognized, the Monocotyledons and the Dicotyledons, their es- sential morphology is so similar that separate treatment would involve needless repetition. The chief differences between them have to do with the structure of the vegetative body of the sporophyte. A general treatment of these differences is not necessary in a book dealing with special morphology, for it belongs to elementary instruction ; while a special treatment would lead into the immense field of anatomy, which it is not the purpose of this book to present. So far as anatomical studies have a conspicuous bearing upon the phylogeny of the groat groups, they are presented by Professor E. C. Jeffrey in the last two chapters. In contrasting Angiosperms with Gymnosperms, one is im- pressed by the fact that a group of plants comprising more than one hundred thousand known species can not be presented with the same confidence and detail as can a group represented by a scant four hundred species. And yet, what have been agreed upon as the essential morphological features of these groups appear to be more uniform in Angiosperms thnn in Gymnosperms. In our treatment of the latter group, the great 1 2 MORPHOLOGY OF ANGIOSPERMS divisions were presented separately because of the diversities; but the morphological diversities among Angiosperms seem to be not so much those of groups as of habit and habitat. While it is generally agreed that the seed-bearing habit was devel- oped independently in more than one phylum, and that the Gvmnosperms and Angiosperms have probably no immediate phylogenetic relation to one another, it is of interest to note the essential contrasting features of the two great seed-bearing groups. The chief contrast in the sporophyte is that in Gvmno- sperms pollination results in bringing the pollen in contact with the ovule, while in Angiosperms the result of pollination places the pollen in contact with a receptive surface developed 1 iv the carpel. This contrast involves great differences in mor- phological structure, so great, in fact, that it is hard to imagine one of these conditions as having been derived from the other. The method of pollination might also be mentioned as a con- trasting feature, since the primitive anemophilous habit seems to be universal among the Gvmnosperms, while among Angio- sperms it prevails only among those groups that may be re- garded as primitive. There accompanies this contrast a similar erne in connection with the flower. Just how this structure may be defined is considered in the next chapter, but the char- acteristic flowers of Angiosperms have no representative among Gvmnosperms, however much the older morphology felt com- pelled to homologize them. However, the method of pollination and the flower are but corollaries to the fundamental contrast involved in the contact of the pollen with the ovule in the one case, and with the carpel in the other. A second fundamental distinction in connection with the sporophyte is to be found in the embryogeny of the two groups. In the Gvmnosperms, the free nuclear division within the fer- tilized egg, and the use of the bulk of the egg as a food re- serve in most forms are in sharp contrast with the absence of free nuclear division in the Angiosperm egg, a character ap- pearing, however, in Gnetum and Tumhoa. If the contrast between the sporophytes of Gvmnosperms and Angiosperms be pressed into anatomical details, the differ- ences are found to be quite as striking, though perhaps a little more perplexing. \ INTRODUCTORY 3 The contrast between the gametophytes of the two groups, especially the female gametophytes, is even greater than that shown by the sporophytes. The male gametophytes of Gymno- sperms when contrasted with those of heterosporous Pterid- ophytes present a much shorter history; and the gametophytic structure produced by the Gymnosperm microspore involves the formation of two or three times as many cells as are formed in the germination of the Angiosperm microspore. The female gametophytes of the two groups, however, are in the main stri- kingly different. As is well known, the female gametophytes of Gymnosperms in general, with their well-organized tissue and archegonia, are almost the exact counterparts of those of Selaginella and Isoetes; while the female gametophyte of An- giosperms remains a morphological puzzle, made still more perplexing by the discovery of the wide-spread phenomenon styled " double fertilization." It is a very significant fact, how- ever, that in spite of the difficulties of the female gametophyte of Angiosperms in the way of interpretation and of origin, it is one of the most remarkably consistent structures known to morphology, the sequence of events in its history representing an almost unvarying schedule, and supplying one of the strong- est arguments in favor of the monophyletic origin of Angio- sperms. In view of these and other differences between Angiosperms and Gymnosperms, the question is raised whether we have not been too narrow in the conception of the seed-bearing habit in compelling these two groups to remain as subdivisions of a group Spermatophytes coordinate with Pteridophytes and Bryophytes. In a certain sense, to select a single character, such as seed-bearing, as a basis for the union of two groups otherwise dissimilar is suggestive of artificial classification. Furthermore, to separate the female gametophytes of Gymno- sperms from those of the heterosporous Lycopodiales, and to associate them with those of Angiosperms, is certainly to do violence to a most important suggestion of natural relation- ships. In our judgment, therefore, the designation Sperma- tophytes should be used in a general way, as a term of con- venience rather than of classification, only less extensive in its application than " vascular plants " ; and Gymnosperms and Angiosperms should be recognized as two groups coordinate 4 MORPHOLOGY OF ANGIOSPERMS with Pteridophytes and Bryophytes. In fact, Pteridophytes and Gymnosperms together form a much more natural group than do Gymnosperms and Angiosperms ; and this fact should be emphasized by treating Gymnosperms and Angiosperms as groups of the first rank. Although it is a question whether Gymnosperms and An- giosperms should be so closely associated as to form the two subdivisions of a great group, there can be no question that Monocotyledons and Dicotyledons are naturally and intimately associated. This proposition is not affected by the question of their common origin, but is based upon their essential mor- phological features, whatever may have been their origin. The characters that separate Monocotyledons and Dicotyledons are cumulative rather than specific, and although the character of the embryo is held to be the decisive one in everv case, there is danger of using it with unnatural rigidity. When a decision between two groups is reduced to a single character, there is a suspicion either that the groups can only be separated arti- ficially or that too much stress is laid upon the character. Mon- ocotyledons and Dicotyledons are best distinguished by cer- tain tendencies that involve several characters, and if these tendencies are supported by the character of the embryo the case is clear. A brief statement of the conspicuous differences may be of service. 1. In the embryo of Monocotyledons the cotyledon is ter- minal and the stem tip lateral in origin ; while in Dicotyledons the stem tip is terminal and the cotyledons lateral in origin. This character seems to be fundamental, and at the present time is the only one that may be regarded as decisive. That the difference indicated will always be expressed in the above terms is not likely, for the nature of the cotyledon is in ques- tion, and the significance of this relation of parts has yet 'to be determined. 2. The development of the vascular bundles in the stele is very different in the two groups. This difference involves not <>nly the arrangement of the bundles, "but also the presence or absence of fascicular cambium, and is far-reaching in its re- suits upon the habit of the body. In the case of perennial stems it involves the general ability to increase in diameter, and this affects the power of branching, and this in turn deter- INTRODUCTORY 5 mines the question of an annual increase in the display of foliage, which means the working power of the body. This character can not be used as a specific test for the two groups ; nor must it be pressed in certain features alone or too rigidly. When intelligently applied, it is probably only second in im- portance to the character supplied by the embryo; but it must be remembered that these prevailing tendencies of the two groups are in some instances exchanged. 3. The characteristic foliage leaves of Monocotyledons have a closed venation, while in Dicotyledons the venation is open. This character involves many differences in detail. For ex- ample, as a result the Monocotyledon leaf is entire, while the Dicotyledon leaf, with veins ending freely in the margin, is inclined to branch more or less, this tendency expressing itself in the greatest variety of ways from simple teeth to the so- called " compound leaves." * It is also true in general that in Monocotyledons there is a sharp contrast in size between the principal veins of the leaf and the reticulating veinlets ; while in Dicotyledons the gradation is so gradual that the reticulation becomes very evident. It may be well to call attention to the fact that while the so-called " parallel '? venation may be of service in distinguishing the majority of Monocotyledons in temperate regions, as contrasted with the " pinnate " or " pal- mate " venation of Dicotyledons, it loses its significance when the tropical Monocotyledons are included. The distinctive character of closed or open venation can not be applied to all Monocotyledons and Dicotyledons, and is certainly less gen- eral in its application than the two characters already given. As a. character to be used in a cumulative way, however, it deserves prominent mention. 4. Among Monocotyledons and Dicotvledons with cvclic flowers the establishment of three as the cycle number of the former, and of five or four as the evcle number of the latter is quite distinctive. In fact, the constancy with which these numbers appear is more remarkable than the exceptions. Of necessity, this character is of comparatively limited use, but it is of service among the cyclic families, and also among those families some of whose floral parts are in cycles. The persist- * This term should he abandoned for leaves, as lias the terra ••compound flower*' for the characteristic head of Compositae. 6 .MORPHOLOGY OF ANGIOSPERMS ent tendency of the spiral groups of Monocotyledons and Dicot- yledons to express the appropriate cyclic number, when the conditions seem to favor indefinite numbers, is even more re- markable than the constant reappearance of the cyclic number in families in which it has become established. Just what has determined these numbers for the two great groups is an inter- esting but unanswered question. The problem is confused by the fact that certain plants, undoubtedly Monocotyledons or Dicotyledons by all the usual tests, have the cycle number of the other group. In addition to the distinguishing characters enumerated above, others of much less general application have been sug- gested, but it is not clear that any of them are really significant group characters. There are certain general differences in the leaves of the two groups that deserve mention, since they come as near rep- resenting group tendencies as any of the secondary characters just enumerated. Among Dicotyledons the foliage leaf is gen- erally more differentiated than among Monocotyledons, inclu- ding a petiole and often stipules. In fact stipules would be quite characteristic of Dicotyledons were they not lacking in so many, for Monocotyledons possess no such structures. Among the latter, however, there is the almost equally char- acteristic leaf-sheath from which the blade directly arises. This general distinction between the leaves of the two groups must have some unappreciated significance, and suggests that it may represent something as fundamental as do the differ- ences in the embryo and the stele. The so-called " germination " of the seed is suggestive of different tendencies in the two groups, but the data seem to be too scanty and indefinite as yet for safe generalization. So far as they do exist, they indicate a tendency in Monocotyledons to free the stem and root tips by the elongation of a portion of the cotyledon, the other portion remaining in contact with the endosperm as a digesting and absorbing organ, very sug- gestive of the " foot " of Pteridophytes ; while in Dicotyledons the tendency is to liberate the growing points and cotyledons by the elongation of the hypocotyl^ and even hypogean cotyle- dons are not related to endosperm as digestive and absorbing organs. INTRODUCTORY < It is claimed that the prophyllum * of Monocotyledons is solitary and posterior, while in Dicotyledons there are two op- posite and lateral prophylla. If such structures generally occurred, or even if this distinction were generally true when they do occur, such a character would be significant, for the prophyllnm certainly has a definite connection with the position of the successive floral parts in relation to the main axis. It has been urged also that the Monocotyledons are char- acterized by a small embryo embedded in an abundant endo- sperm, and that in Dicotyledons the tendency is to develop larger embryos at the expense of the endosperm. This involves so many and such important exceptions that it can hardly be regarded as a distinction between these two great groups. The roots of Monocotyledons are said to differ from those of Dicotyledons in that the primary roots are short-lived and there is no persistent root-svstem as in many Dicotvledons. While this may be true of Monocotyledons in general, it is also true of many Dicotvledons, and can not be used as a distinct- t t. 7 ive character. All the characters enumerated above, both those of primary and those of secondary importance, are to be considered in any general characterization of the two groups : but it must be re- membered that most of them await confirmation as essential group characters. It is of interest to note that they are all characters of the vegetative sporophyte, and that the sporangia and gametophytes of Monocotyledons and Dicotyledons have thus far given no tangible evidence of group differences. * Translated into German as Vorblatt, and into English as fore-leaf. The first leaf on a branch, but used only in connection with the bractlets of a flower cluster. CHAPTEK II THE FLOWEE The morphology of the flower of Angiosperms has an enor- mous literature, much of which is now more curious than valu- able. It is not the purpose of this book to present the numerous details and extensive terminology that were so conspicuous a feature of the older morphology, dominated as it was by the doctrine of metamorphosis. For these the student is referred to Eichler's Bluthendiagramme (1875-'78), in which may be found the most complete account of the flower of Angiosperms from this standpoint. The English student will also find an admirable short account of the same subject from the same standpoint in Gray's Structural Botany (1879). A presenta- tion that combines much of the older method of treatment with newer points of view appears in Goebel's Outlines of Classi- fication and Special Morphology of Plants (English transla- tion, 1887). Among the later important literature the follow- ing may be consulted : Goebel's Vergleichende Entwicklungs- geschichte der Pflanzenorgane in Schenck's TIandbuch dor Botanik (3 ' : 90-432. figs. 126. 1884) ; Celakovsky's Ueber den phylogenetischen Entwicklungsgang der Bliite und iiber den Ursprung der Blumenkrone, I and IT (Sitzber. Konigl. Bohm. Gesell. Wiss. 1896 and 1900) ; Familler's Biogenetische Untersuchungen iiber verkiimmerte odor umsebildete Sexualor- gane (Flora 82: 133-108. figs. 10. 189CO ; Engler and Prantl's Die Xatiirlichen Pflanzenfamilien ; Goebel's Organographie der Pflanzen (vol. ii, 1901).* These works and others like them must be consulted for the details of the structure of angio- * It should be understood that in this mention of the literature of the flower only certain important works are cited, and that only in the subse- quent chapters is there any attempt at presenting fairly complete lists of the important literature. 8 THE FLOWER 9 spermous flowers, for in this chapter only certain of the broader morphological features will be discussed. Any strict definition of a flower seems to be impossible. That the morphological precursor of the angiospermous flower was some such structure as the strobilus of Pteridophytes seems reasonably clear. In fact, the strobilus is plainly continued among the Angiosperms in spiral flowers and spirally arranged members. The appearance of distinct floral leaves associated with sporophylls, however, is characteristic of the higher An- giosperms. If a flower is essentially a sporophyll or a set of sporophylls, as the older definition insists, Pteridophytes must be included among flowering plants. If, on the other hand, a flower is characterized by floral leaves, many Angiosperms are not flowering plants. In any event, the term flower is of indefi- nite application, and is incapable of sharp definition. It is a term of convenience among Angiosperms, where it also in- cludes strobili. The attempt of the older morphology to estab- lish a definite conception for a flower, and to force all of the sporophyll-bearing structures of Seed-plants into this concep- tion was exceedinglv unfortunate. The development of floral leaves among Angiosperms seems to be connected with the evolution of entomophily, which has resulted in immense diversity in the details of floral structure, but snch details are quite foreign to the purpose of this book. The origin of floral leaves, however, is a question that must be considered. That all floral loaves are derived from sporophylls may be said to be the current view, as stated by A. P. De Candolle in 1817, and by many subsequent writers, notably Celakovsky in 1896 and 1000. Goebel, however, in his recent Organogra- phie der Pflanzen, claims that while in a large number of cases floral leaves may be derived from sporophylls, as in Nymphaea, etc., they are often derived from " bracts." For example, ho calls attention to certain anemones in which the involucre be- comes the calyx and this in turn may become petaloid. In other words, he claims a double origin for floral leaves, namely, spo- rophylls and foliage leaves, and whichever their origin the result is the same. Tt may be of interest to note that Goebel's definition of a flower, a definition originally proposed by Schleiden, is " a shoot beset with sporophylls," which of course 10 MORPHOLOGY OP ANGIOSPERMS includes certain Pteridophytes among flowering plants. It is certainly more in accord with present morphological concep- tions not to limit too rigidly the possible origin of a structure, and from this point of view it seems reasonable that floral leaves in general may have been derived from contiguous structures both above and below. It is not always easy to delimit a flower exactly from the vegetative shoot, for there are numerous illustrations of grada- tions between foliage and floral leaves ; but for all ordinary purposes four different organs are readily recognized as enter- ing into the structure of a highly developed flower. The dis- carded doctrine of metamorphosis assumed that such a flower is the type, from which all others are the modified descendants, and this concej^tion is perpetuated in terminology. The same conception dominates also in nearly all presentations of floral diversities, as it is well-nigh impossible to abandon at once all the terms of an obsolete conception and remain intelligible. It has been a very prevalent conception, therefore, that flowers of simpler structure than the assumed type are reduced forms. There are certain cases in which this seems clear, as in the relation of Lemna to the Araceae; but the vast majority of simpler flowers are better regarded as primitive than as re- duced forms. At present this is at least a valuable working hypothesis, for it coincides in general with the morphological and historical evidence concerning relationships, as well as -with the doctrine of evolution. Accepting the evidence that the simpler flowers are for the most part the more primitive forms rather than reduced ones, certain prominent tendencies in the evolution of the flower, admirably presented by Engler, may be discussed. It must be understood, however, that only general tendencies are traced, for the actual lines of descent can not be determined by our present knowledge. The naked flower with one or more free sporophylls may be regarded as the most primitive form. In fact, it is an angiospermous flower without the characteristic floral leaves. Such a flower may sometimes represent a case of reduction, lint its persistent association with plants recognized as primi- tive from other testimony is very strong evidence that it is a primitive condition of the flower. From this stage a series THE FLOWER 11 can be arranged illustrating the gradual development and dif- ferentiation of the two floral envelopes. Foliar members, whether derived from foliage leaves or sporophylls, become more and more definitely associated with the sporophylls, until they may be regarded as constituting an inconspicuous, bract- like perianth. They gradually appear in two definite sets and become more conspicuous, and sooner or later show the petaloid texture and coloration. The final stage is a completely differ- entiated calyx and corolla, with their characteristic differences. This tendency to produce a completely differentiated" calyx and corolla has resulted in its attainment by most flowers, but there are numerous cases in which even near relatives have not made the same progress in this regard. For example, the phenom- enon styled apetaly may be observed in flowers whose nearest relatives have a distinct calyx and corolla. While some cases of apetaly may be explained as the suppression of a set of floral envelopes, there are certainly eases in which it means that the two sets have never become differentiated. This indicates that progress made in a single direction can not be used as a cri- terion of relationship. In general, however, it must remain true that a flower with completely differentiated calyx and corolla, other things being equal, is of higher rank than a flower which has not attained this differentiation. Among the most primitive flowers the floral axis tends to elongate, and the members appear in indefinite numbers along a low spiral. In more highly developed flowers the growth of the axis in length is checked at a very early period, so that the spiral along which the members successively appear becomes lower and lower, until it has only a theoretical existence, pass- ing into successive cycles, which eventually become limited in number. With the appearance of definite cycles the number of members appearing in each one becomes limited, the limit in Monocotyledons being prevailingly three, and in Dicotyledons five or four. It is to be noted that the cyclic arrangement is not attained simultaneously by all the sets of a single flower. For example, in many species of Ranunculus the sepals and petals are cyclic, or approximately so, while the stamens and carpels are distinctly spiral. This tendency is so well-marked and so uniformly displayed that Engler has used it as a basis for dividing Monocotyledons into two great series, the " spiral 12 MORPHOLOGY OF AXGIOSPERMS series " comprising all those families that show the spiral tend- ency in any of the floral sets, and the " cyclic series " compris- ing all those whose flowers are completely cyclic, the former series including all the more primitive families. There is no reason why this same distinction can not be applied also in a general way to the Archichlamydeae. This gradual transition of flowers from the spiral to the cyclic condition is one of the best-marked tendencies in their evolution, and has the advan- tage of being represented by innumerable intermediate stages. All of those families which are recognized as being of the high- est rank have completely cyclic flowers, with members appearing in definite and low numbers, notablv illustrated bv the whole group Sympetalae. There is a marked tendency in flowers for the members of a single set to lose their identity and to develop en masse, a phenomenon called " coalescence " by the older morphologists, under the impression that separate members had united. This congenital union is to be distinguished from such a mechanical union as is shown by the anthers of Compositae. In the organ- ogeny of such a flower it is to be observed that in the meriste- matic zone from which a certain set is to develop, the different members first appear as separate primordia, but sooner or later the whole zone shares in the growth and, the axial growing point being checked, an annular structure arises that gradually assumes the size and form of the mature organ (Fio-. 1). It has been claimed that this is a toral uprising and that, for example, the tubular portion of "a sympetalous corolla is mor- phologically torus, but there seems no more reason for this supposition than to regard an individual petal as a toral up- rising. It is merely the difference between development from the meristematic zone at certain points and at all points. As is well known, this development of the whole zone may begin almost at once, or may be deferred until the set is nearly mature, resulting in every stage of separation in the members, from a completely tubular structure to one that is tubular only at base. Or the zone may develop for a time in two sections and later en 7nasse, resulting in the so-called bilabiate structure. Further inequalities in the time and rate of development result in various irregularities. In any event, this tendency to zonal development rather than the maintenance of separate points of THE FLOWER 13 development is persistent among flowers, the first set showing it being the carpels, resulting in syncarpy. The zonal develop- ment of the corolla, however, or sympetaly, accords with so many other characters indicating natural relationships that it has been used to designate and even to define the great group Sympetalae. This is probably pressing a single character too far, for there is evidence that the result has been to do violence to certain natural relationships, and to make certain unnatural groupings. This tendency to zonal development is found in every floral set, and those flowers that show it are certainly to be regarded as of higher rank than those that do not. Among the more primitive flowers each cycle arises sep- arately from the growing point, its members remaining separate or the whole meristematic zone entering more or less completely into the outgrowth. The insertion of each cycle is definitelv below that of the next inner cycle, resulting in an hypogynous flower (Fig. 1, A). That hypogyny is a primitive condition of the flower is a statement that does not seem to need discussion. The tendency to zonal development, however, is carried farther when a whole region arising en masse produces two or more cycles of floral members. In the simplest cases two cycles are thus produced, as is illustrated by the strong tendency of the petaliferous and staminiferous cycles to have a common origin in sympetalous flowers, resulting in the appearance of " stamens inserted on the tube of the corolla." The same tendency is shown among orchids, in which the whole region for the devel- opment of stamens and carpels arises in a single body, forming the characteristic gynostemium or " column." While these may be regarded as special tendencies of certain groups, rather than of flowers in general, there are other instances that seem to belong to the general evolution of the flower. In certain cases the region of the growing point belonging to the carpels ceases to develop, while the rest of the growing point continues to develop en masse, forming a cup or urn-like outgrowth, from the rim of which the three outer sets develop separately, form- ing the perigynous flower (Fig. 1, B). In this case the carpels arise from what seems to be a depression in the center of the torus, but which, of course, is the region of checked growth. Perigyny is chiefly displayed among families of the Archi- chlamvdeae. u MORPHOLOGY OF ANGIOSPERMS Far mure general is the tendency to epigyny, in which the checking of apical growth and the continued growth of the rest of the growing point results in an ovule-bearing cavity grad- A w B D Fig. 1. — "Diagram to illustrate the morphology of typical flowers. A, hypogynous; £, perigynous ; C, epigynous ; Z>, epigynous with prolonged ' calyx tube.' Eecep- taele is dotted; carpels are cross-line J ; 'perianth tube,1 or ' calyx tube,1 vertically lined ; sepals, petals, and stamens are unshaded, but may be distinguished by their relative positions." — After Ganong.* ually roofed over by the carpels. From the top of the ovary thus developed the four sets of floral members develop as usual, those of each -ft remaining independent, or a cycle developing * Ganoxg, W. F. The Teaching Botanist. New York. 1899. THE FLOWER 15 en masse, or two cycles (especially petaliferous and staminifer- ous) having a common origin (Fig. 1, C, D). Goebel holds (Or- ganographie) that at least in some epigynous flowers (as Pirus Mains) the carpels do nut merely roof the ovular cavity but also line it, basing the claim upon a study of the meristematic tissue ; in which case the wall of the so-called " ovary " is toral without and carpellate within. It is to be expected that numerous in- termediate stages between complete hypogyny and extreme epigyny will be displayed, as may be inferred even from the doubtful phrases employed by taxonomists to describe them. It also seems to be a safe conclusion, since epigyny is con- stantly associated with the most specialized groups of each great division, as Orchidaceae among Monocotyledons, Umbelli ferae among Archichlamydeae, and Compositae among Sympetalae, that it is a mark of higher rank than hypogyny in any evolu- tionary series. The tendency for the members of a floral set to develop unequally, resulting in zygomorphy or various forms of " irreg- ularity," is not general, and can not be applied so broadly as can the tendency to the cyclic arrangement or to epigyny. In certain groups, however, it is very pronounced as a special character, as Orchidaceae among Monocotyledons, Legumi- nosae among Archichlamydeae, and Personates among Sympet- alae. The occurrence of zygomorphy in relatively primitive as well as in highly specialized groups indicates that it is to be regarded as a special rather than a general tendency ; and yet, other things being equal, the zygomorphic flower is to be re- garded as of higher rank in any given evolutionary series than the actinomorphic flower. Diversities resulting from inequali- ties of growth are often described in terms of symmetry, a term that unfortunately has two applications in connection with the flower, for its well-known biological use by Sachs found it already used to designate a flower " in which the mem- bers of all the cycles are of the same number." In its biological sen^e a symmetrical flower is one " that can be divided into two similar halves, or the parts of which are radially disposed around a central point." The terms " monosymmetrical " and ' polysymmetrical " are logical, but not better than the older terms of Eichler, " zygomorphic " and " actinomorphic." How- ever, the phenomena of floral symmetry are not well expressed 16 MORPHOLOGY OF ANGIOSPERMS in two categories, and three have been proposed, as follows: (1) actinomorphic, in which the planes of symmetry are as numer- ous as the members of a cycle; (2) isobilateral, in which there are two planes of symmetry, but the halves produced by one plane are unlike those produced by the other (Dicentra, Cru- ciferae, etc.) : and (3) zygomorphic, in which there is only one plane of symmetry (Fig. 2). These categories are expressions of certain laws of growth, and that they are somewhat funda- c Fig. 2. — A, radial symmetry (Lilium tigrinum); B, isobilateral symmetry (Capsella Bursa-pastoris) \ C, zygomorphic symmetry (Scrophularia nodosa). mental may be inferred from the fact that they are persistent through great groups of plants. While these and other evolutionary tendencies are to be observed among flowers, it is evident that they are not neces- sarily expressed simultaneously. For example, the spiral and cyclic arrangements are associated in Ranunculus, zygomorphy is associated with polypetaly and hypogyny among the papil- ionaceous Leguminosae, epi ryjiy is associated with polypetaly among the Umbelliferae, and sympetaly and zygomorphy are associated with hypogyny among the Labiatae. It is among the Compositae that practically every evolutionary tendency mentioned finds its highest expression. Tt is only by striking an average that such characters may be used in roughly placing a family in its evolutionary position, commonly called its " relative rank." The classic memoir on the organogeny of the flower is Payer's Traite d'organogenie de la fleur (1857), but the sub- ject has not been developed since as it deserves. In the case of spiral flowers, in which the torus elongates more or less, the fV/1 c D P G t H Fig. 3. — Cnicus aruensis. Floral development: A, receptacle almost evenly convex; B, appearance of papilla? to become flowers ; C, a single papilla more advanced, show- ing beginning of corolla; D, corolla more prominent: E, stamens distinguishable; F, carpels and pappus (calyx) evident; G, carpels beginning to form cavity of ovary; //. ovule readily distinguishable; /, ovule showing megaspore mother-cell and single thick intesrument: b, bract of involucre; c, corolla; s, stamen; o, carpel; p, pappus (calyx). A-H x 50 ; / x 100. 17 18 MORPHOLOGY OP ANGIOSPERMS members appear in acropetal succession along a continuous low spiral, and just when one set of members stops and the next begins is indefinite within certain usual limits. There seems no doubt in this case that the primordia are indifferent up to a certain stage of development, and that the particular organ produced depends upon something outside of the essential con- stitution of the primordium itself. In the case of cyclic flowers, in which toral growth in length has been checked and there has been growth in diameter, the acropetal succession of members is often very much interfered with. The " disturbances " that arise in the torus by substituting growth in diameter for growth in length account not only for the breaking up of the acropetal succession, but also for the inequality of members of the same cycle, or of different regions of the cycle. It is evident that in the case of cyclic flowers organogeny must deal not only with the succession of cycles, but also with the succession of mem- bers in each cycle. Definite data in reference to these points are not so available as they should be, but a few illustrations may be cited. For the majority of cyclic flowers it seems to be assumed that the cycles appear ^in acropetal succession — namely, sepals, petals, stamens, carpels — and that the members of each cycle are practically synchronous in origin, but it is probable that this assumption is gratuitous. While theoretically it may be as- sumed that the cycles should arise in acropetal succession, the fact that they do not in many observed cases indicates that they may not in many more cases ; and the synchronous ap- pearance of the members of a single cycle is unsound as a theoretical assumption. Hofmeister * records that in Rosa, Potentilla, and Rubus the primordia of the carpels appear be- fore those of the stamens have reached the full number, and that in Hypericum calycinum the primordia of the sepals ap- pear after those of the stamens. It is also generally known that among the Compositae (Fig. 3), Dipsaceae, Valerianaceae, and Rubiaceae, in which the sepals are much reduced or modi- fied, their primordia do not appear until after those of the stamens and carpels; and that among the Cruciferae (in Cap- sella, at least) (Fig. 4) the petals are the last members to * Hofmeister, W. Allgemeine Morphologie dcr Gewachse. Leipzig. 1868. p. 462. THE FLOWER 19 appear. "\Vebb * has recently observed in Astilbe that the order of succession of floral cycles is sepals, inner stamens, car- pels, outer stamens, and petals. In this case there is an acro- Fig. 4. — Capsella Bursa-pastoris. Floral development: A, floral axis before appearance of floral organs ; B, appearance of sepals ; 0, appearance of stamens ; carpels barely distinguishable ; Z>, appearance of petals : «, sepals ; m, stamens ; c, carpels ; p, petals x 130. petal succession of certain cycles, followed by a basipetal succes- sion of the remaining ones. The remarkable case of the flower of the Primnlaeeae, noted by Pfeffer,+ is also familiar, in * Webb, J. E. A Morphological Study of the Flower and Embryo of Spi- raea. Bot. Gazette 3&: 451-4S0. figs. 87. 1902. For correction of names, see Kkiider in Bot. Gazette 34: 246. 1902. \ Pfeffer, W. Zur Bliit henentwicklung der Primulaceen und Ampelideen. Jahrb. Wiss. Bot. 8: 194-215. 1872. 20 MORPHOLOGY OF AXGIOSPERMS which the primordia of the petals appear after those of the stamens, and -each apparently from the dorsal surface of a young stamen. The conclusion that the so-called petals of this family- are not morphologically petals, but stamineal outgrowths, is unnecessary, since the phenomenon can be more logically in- terpreted as a case in which the primordia of stamen and petal have a common origin, entirely analogous to the sympetalous corolla with stamens " inserted on its tube," but in which the separate primordia of the petals have been delayed in their appearance. Such examples as those enumerated above simply serve to emphasize the desirability of a more serious and sys- tematic investigation of the whole subject. In the case of members of a single cycle, it is a question whether their primordia ever appear simultaneously, although they may appear in rapid succession. In zygomorphic flowers, however, the succession is probably always evident. For ex- ample, Goebel * cites the case of the papilionaceous Legumino- sae, in which the anterior median sepal first appears, then those to the right and left of it simultaneously, and finallv the two obliquely posterior ones; but before these last are evident the two obliquely anterior petals appear, and after them the other three in the same order as the corresponding sepals. This succession proceeds right and left from the anterior member to the posterior. In other known cases, however, as in Reseda, according to Payer, the succession is right and left from the posterior member to the anterior. It must also be noted that a meristematic zone giving rise to a set of members may add to the set later or even duplicate it, giving rise to the well-known interposition of new members or new sets. For example, it is stated that among the Gera- niaeeae, Rutaceae, and Zygophvllaceae a new cycle of five sta- mens is interposed among the five already formed ; and that in Aceraceae and Sapindaceae two to four stamens are inter- calated in the complete cycle of five previously formed. This later interposition of new sets or new members has been re- corded chiefly for stamens, and is a prolific source of inter- ference with the " symmetry " of numbers. All seed-bearing plants are necessarily dioecious since they * Goebel, 0. Outlines of Classification and Special Morphology. English translation. 1887. p. 424. THE FLOWER 21 are heterosporous. So far as there is any advantage in this habit, however, it is practically lost if stamens and carpels are present in the same flower or upon the same plant. Morpho- logically the gametophytes are unisexual, but in fact they are dependent upon the same individual. Any physiological advan- tage, therefore, that comes from the crossing of individuals must be secured by pollination or by the separation of stamens and carpels upon different individuals. It is unfortunate that the term " dioecious " has two distinct morphological applica- tions, referring to the sexual differentiation of individuals among the lower plants, and to the sporangia! differentiation of individuals among seed-bearing plants ; but from the phys- iological standpoint the distinction probably does not exist. As a consequence, the dioecious habit in effect is secured in certain seed-plants by the development of monosporangiate individuals, and it is perhaps significant that this habit not only prevails among the more primitive seed-bearing plants, but is associated in the main with wind-pollination. Among the higher Angio- sperms the effect of the dioecious habit is secured for bisporan- giate plants by means of insect-pollination. It follows from this point of view that neither the monosporangiate nor the bisporangiate habit can be regarded as in itself the more primi- tive. The former habit prevails among the more primitive families because they are necessarily anemophilous ; while the latter prevails among the higher families because insect polli- nation does not necessitate the monosporangiate habit. It should be noted that Goebel (Organographie) regards the bisporangiate condition as primitive, the monosporangiate being derived from it by reduction. This can be demonstrated in certain cases, but the monosporangiate condition is probably the primitive one in many of the more primitive angiospermous families. In any event, the monosporangiate and bisporan- giate habits are not always settled ones. For example, in the monosporangiate Amarantus retro ferns there are occasional bisporangiate flowers; while in monosporangiate and dioeciou3 willows both catkins may appear on the same individuals, and the catkins themselves may bo mixed (staminate, pistillate, and bisporangiate). It follows also that there may be monosporan- giate members in all great groups (as Ranunculaceae), or even in bisporangiate genera (as Rumex and Lychnis), for this habit 22 MORPHOLOGY OF ANGIO.S PERMS is probably not a hindrance to any form of pollination, and cer- tainly prevents self-pollination. Cross-pollination by wind or by insects, therefore, appears as an offset to the loss of any advantage originally gained by the dioecious habit; and the appearance of monosporangiate individuals in any Angiosperm group does not imply a tendency toward a more primitive or more advanced condition. For example, the monosporangiate habit of poplars is no more indicative of a primitive condition than is the monospo- rangiate habit of certain Compositae of an advanced condition. The older morphologists considered the floral members aS morphologically leaves, and presented proofs which to them seemed decisive, such as the leaf-like position and intergrading of members, and various malformations, among which are the so-called " reversions." This conclusion was controlled by the prevailing doctrine of metamorphosis, and under its guidance nothing seemed clearer than that stamens and carpels are trans- formed leaves. While sepals and petals may be regarded as often leaves more or less modified to serve as floral envelopes, and are not so different from leaves in structure and function as to deserve a separate morphological category, the same claim can not be made for stamens and carpels. They are very an- cient structures, of uncertain origin, for it is quite as likely that leaves are transformed sporophylls as that sporophylls are transformed leaves. It is a rigid morphology, however, domi- nated by the doctrine of " types," that denies to an organ so thoroughly established as the stamen of Angiosperms a mor- phological individuality. One might almost as well deny to the leaf itself a morphological individuality because it did not always exist as a distinct organ. Just how long an ore;an must maintain its independence before it can be recognized as a morphological unit is not easy to say, but stamens and carpels seem to have earned the right. To call a stamen a modified leaf is no more sound morphology than to call a sporangium derived from a single superficial cell a modified trichome. The cases of " reversion " cited are easily regarded as cases of re- placement. Lateral members frequently replace one another, but this does not mean that one is a transformation of the other. For example, in 1889 Barber* observed a Nymphaea in which * Barber, C. A. On a Change of Flowers in Tubers in Nymphaea Lotus. var. momtrosa. Annals of Botany 4: 105-116. pi. 5. 1889. THE FLOWEit 23 foliage leaves had replaced all the floral members within the calyx and the end of the axis had become much swollen. It is probable that the latter fact was responsible for the former, and that a growing axis put forth leaves, as it usually does ; but the inference that these leaves represent the replaced floral mem- bers in any morphological sense has no logical connection with the facts observed. Such cases as that of the ordinary flowers of Nymphaea, in which stamens seem to be gradually differ- entiated from petals, present no difficulty when one notes the remarkable indifference of sporangia to the nature of the mem- ber upon which they appear. Because microsporangia appear occasionally upon an axial structure it might as well be argued that stamens are transformed stems. The stamens and carpels are just as definite morphological structures as are foliage leaves, with just as distinct functions, and should be so re- garded, whatever may have been their historical origin. Stamen and leaf probably merge into one another in history, and so do stem and leaf, but all have become established as distinct organs. Further details as to the varving form and structure of sepals and petals are of no special morphological significance, and are of interest chiefly to the taxonomist and the ecologist. The stamens and carpels, however, are so intimately associated with essential morphological structures that some further de- tails in reference to them are necessary. The stamen set has been called collectively the " androe- cium," a name so objectionable to the morphologist on account of its sexual significance that it should be abandoned. The stamen is an organ devoted to the production of microsporangia, and its endless diversity of form and position is related more or less directly to the needs of pollination. The term " anther " is one of convenience, but represents a morphological complex made up of sporangia and more or less sporophyll tissue. The cooperation of sporophyll and sporangia in the dehiscence of, the latter will be included in the discussion of the microsporan- gium, as well as those various differences among anthers that have to do with the number and behavior of their sporangia. It is important to note that stamens have the power of branch- ing, and can thus multiply sporangia. Well-known cases are Callothamnas, in which the branching is like that of a pinnate 3 24 MORPHOLOGY OF ANGIOSPERMS leaf; Ricinus, in which repeated forking results in a stamen hearing very numerous sporangia; and Hypericum, in which the primordium branches, that is, produces secondary primor- dia, the common base of the tufted cluster not being recognized in the mature condition. The case of zonal development, that is, an uprising from the whole staminiferous zone, and also the case of stamen and petal or stamen and carpel regions rising en masse, have already been noted in connection with the gen- eral tendencies of the flower. The tendency of stamens and of carpels to become more or less coalescent through pressure is also well marked, as in the anthers of Compositae and Lobe- liaceae, and in some eases that have been called syncarpy. It remains to note the fact that stamens occur in all stages of abortion, especially to be observed among the Personales, from the absence of sporangia to that amount of abortion that is only short of suppression. Stamens that have lost their normal function are generally called " staminodia," but they may as- sume various forms and serve a variety of purposes. In certain cases, as notably among the Labiatae, the claim that one or more stamens have been suppressed is justified by their pres- ence in near relatives, combined with the occurrence of unoc- cupied points where stamens ordinarily appear. The carpel set has been called collectively the " gynoecium," a term that also should be dropped from morphological ter- minology on account of its implication of sexuality. The carpel is the organ most intimately related to the megasporangia, in- vesting them more or less completely, but not always producing them, and giving name to the Angiosperms. Its history is un- known, for although it is easy to imagine it derived from such open carpels as are found among Gymnosperms, no clear inter- mediate stages have been found. At all events, it is a thor- oughly established and characteristic organ. The term " ovary v for the sporangium-bearing cavity is particularly unfortunate on account of its very different application among animals. To avoid this confusion Barnes * has proposed the term " ovu- lary," but even this contains in its stem the sexual implica- tion. The style is definitely related, in its varying form and length, to the problem of pollination, and upon it the stig- * Barnes. C. R. Plant Life. 1898. p. 240. THE FLOWER £5 matic surface is developed in various ways. This surface is increased in area by the enlargement of the apex of the style, by its branching, or by being developed laterally upon the style. One of the essential features of the structure of the carpel is the provision for the progress of the pollen-tube from the receptive surface to the sporangium or even to its micropyle. A specialized and continuous nutritive tissue connects these two extremes, often confused in the sporangial chamber with the " placenta," in the style called " conducting tissue," and upon its surface the " stigma," but forming one continuous tissue system, well named the conducting tissue. It is unfor- tunate that the terminology of taxonomy has somewhat di- verted attention from the continuity of this tissue, for in it the " stigma " is an organ distinct from the style, rather than a display upon the surface, often modified to receive it, of a special tissue of the style. While the placenta is the point or line of sporangium origin, and may be said to consist of spo- rangiogenic tissue, it is probably true that much of the out- growth that stands for the placenta to many is conducting tissue. In the case of hollow styles, as in Lilium, Butomus, Agave, Erythronium, Viola, Campanula, Sarcodes, etc., the conducting tissue lines the canal as a glandular layer, or in some cases, as in Anagallis, fills up a hollow style; but in most cases the style is solid, with the conducting tissue as an axial strand. In case a single style is connected with two or more sporangial chambers, the strand of conducting tissue branches into each chamber. This suggests the possibility that the stylar canal, with its lining of conducting tissue, may represent a primitive angiospermous condition, and that the larger development of this tissue has resulted in the prevailing solid style, a view that is also suggested by the development of the style. Of course the reverse may be true, and the stvlar canal a result of the breaking down or rupture of the axial strand of conducting tissue. The strong tendency to a congenital development of carpels has been previously noted, and this justifies the use of the term " pistil " as one of convenience, although it does not stand for a morphological unit. It is applied to any organization of car- pels that appears as a single organ with one ovary, whether one or more carpels are involved. It is to be noted that the term 26 MORPHOLOGY OF ANGIOSPERMS . " ovary " also, as usually applied, has no definite morphological significance, referring to a morphologically single sporangia! chamber or to a combination of several such units, and these chambers may be of axial as well as of carpellary origin. The various ways in which the congenital carpels are related to one another in a compound pistil are of great service in taxonomy, as the particular structure of such a pistil is usually charac- teristic of great families, or even of groups of higher rank. These details of structure are too fully presented in various texts, however, to justify their repetition here. The relation of sporangia to carpels is an important subject to the morpholo- gist, and will be considered in connection with the development of the sporangia. CHAPTER III THE MICROSPOBANGIUM The microsporangia of Angiosperms are embedded struc- tures, and are derived from the outermost layer of the peri- blem. Thus far, the only recorded exceptions to- this origin are Naias flexilis, and probably Zannichellia 15 and Lilaea subulata,ls whose micro- sporangia are claimed by Campbell to be derived from the plerome (Fig. 5). The periblem origin of the spo- rangia seems to account for the fact that the archesporium is superficial in Pteridophytes and hypodermal in Spermatophytes. It also accounts for the indifference of the sporangia to the morphological nature of the or- gan upon which they appear. In general, they occur upon a lateral member that holds the same relation to the axis as do the leaves, and in this sense it may be called a leaf-like member. Such sporangia, therefore, may be called foliar, and the struc- ture that bears them a sporophyll. Fig. 5.— Naias fiexilis. A, young In certain cases, however, the sporan- stamen showing " integument » gia are derived from the periblem of the axis, and such may be called eauline. In each case the resulting and plerome origin of arche- sporium ; sporogenous cells rep- resented with nuclei : x 200. £, later stage; x 70. — After Campbell.16 o^o-an is a stamen, whether in the po- "■ m of a leaf or of an axis. The freedom with which micro- sporangia are sometimes produced may be illustrated by the 27 28 MORPHOLOGY OF ANGIOSPERMS willows, notably Salix petiolaris, in which Chamberlain 16 found microsporangia in the " placenta " of the ovary, the carpel some- times being wide open and bearing both microsporangia and Fig. H.—lSalix petiolaris. A, microsporangia in wall of ovary ; both anatropous and orthotropous ovules. B, microsporangia with long stalks within the ovary ; pollen normally developed ; ovule orthotropous. C, branching stamen, each anther with four microsporangia ; anther on right terminated by a stigma ; x 50. — After Cham- berlain.16 megasporangia, and in some cases stigmas developing on sta- mens (Fig. 6). The canline origin of microsporangia seems to have been recorded first in 1868 in the case of Casuarina, bv KaufT- mann ; 6 and then in 1869 for the species of Naias, by Magnus.7 confirmed in 1897 by Campbell.15 In 1873 Warming8 made a similar record for Cyclanthera, and was confirmed by Eng- ler 9 in 1876. Rohrbach 5 discovered canline microsporangia in Typha; Goebel11 (p. 353) states that they occur in the " unbranched stamens"; and their occurrence in T. latifolia was confirmed by SchafTner 17 in 1897. In 1897 Campbell15 added to the list Zannichellia, and in 1898 LHrtea.18 In 1900 Lotsy 30 suggested that the curious stamen of Rhopalocnemis phalloidcs (BalanophCraceae) is an axial structure. THE MICROSPORAXGIUM 29 It is reasonably assured, therefore, that eauline micro- sporangia occur in at least seven genera, both Monocotyledons and Dicotyledons being represented. Upon the whole, they seem more characteristic of the primitive members of these two groups than of the more highly specialized members, but this impression may disappear with further investigation. If the eauline origin of megasporangia be considered, the primi- tive character of this feature becomes increasingly uncertain, for eauline megasporangia are common even in the highest groups. It seems probable, therefore, that the eauline or foliar origin of sporangia among Angiospenns is not to be taken as an argument for or against the primitive character of the group in which they occur. The particular organ developing micro- sporangia was probably determined not by its morphological nature, but by what may be called its physiological relations (Fig. 6). Even among Pteridophytes, the sporangia of Lycopo- dium are foliar, and those of the nearly allied Selaginella eau- line ; and aiming Gymnosperms sporangia have both origins. It is evident, therefore, that the distinguishing morphological Fig. y.—Lilium philadelphicum. Transverse section of almost mature anther; nearly all the walls separating the uiicrosporangia have broken down; highly developed stomium i 8) and endothecium (with its nb-hke thickenings i very prominent ; x 25. — From a drawing by W. J. G. Land. structure is the sporangium rather than any member of the plant body from which it may arise. In most case- the stamen produces four microsporangia (Fig- 7), and the exceptions noted thus far are by no means 30 MORPHOLOGY OF ANGIOSPERMS numerous. Caldwell 23 lias called attention to the occurrence of what might be regarded a single microsporangium in Lemna ; it is well known that the stamens of Asclepiadaceae produce only two microsporangia ; and in Hamamelis (Shoemaker35) there is a single sporangium to each " pollen-sac." Eight mi- crosporangia had long been observed among the Mimoseae when Engler 9 reported a still larger number. Among the Orchida- ceae Guignard 10 reports eight microsporangia in the stamen of Calanthe veratrifolia ; and among the Onagraceae, as in Gaura, more than four microsporangia are suggested by the pollen-sacs (see Goebel,11 p. 369, foot-note 2). Among Loranthaceae Van Tieghem 13 says that the number of pollen- sacs is exceedingly variable, ranging from one to an indefinite number ; and the same is true of the Balanophoraceae, as re- ported by several investigators. Attention should be called to the fact, however, that the number of sporogenous masses finally developed may not necessarily determine the number of spo- rangia, for plates of sterile tissue, derived from sporogenous tissue, have been observed to divide a single mass of sporoge- nous tissue into two or more. This has been made out clearly by Caldwell 23 in the case of Lemna (Fig. 11) ; and in those cases in which more than four microsporangia are reported a detailed study of their origin is desirable. In the case of branching stamens, referred to on p. 23, the microsporangia may become very numerous. The time for the formation of microsporangia in relation to what is usually called " the growing season " has not re- ceived the attention it deserves. In 1896 Anna Smith14 re- ported that she had discovered the pollen mother-cells of Tril- lium dividing in the spring beneath frozen soil. In 1897 Chamberlain 16 called attention to the fact that the microsporan- gia of Salix glaucophylla are in the mother-cell stage in Oc- tober, and that they pass the winter in this condition. In 1898 the same investigator 20 reported that this is true of other species of Salix; that in Corylus americana (Fig. 8, B) and Alnus glutinosa the midwinter catkins contain pollen ready for shedding with the generative cell formed ; that in Populus monilifera (Fig. 8, A) the primary sporogenous cells are found in July and the mother-cell stage in October, the latter condition persisting through the winter; and that in Hepatica THE M1CR0SP0RANGIUM 31 the mother-cell stage was found in September, and fully formed pollen in the spring while the ground was still frozen. Dug- has also observed that the microsporangia of Symplo- gar carpus pass the winter in the mother-cell stage. The pollen mother-cells of Podophyllum peltatum are forming the tetrads when the young plant has reached the surface of the ground, so that in all probability the winter is passed in the mother-cell staae. Although Conrad 29 found stamens well formed in the winter buds of Quercus velutina, the tissue of the anther was still homogeneous. These records mere- ly serve as an indication of what may be expected when the subject is really investigated. It is evident that the time elapsing between the differentiation of the archesporium and pollination is often much longer than has been ordinarily supposed. It would seem probable that in gen- eral those plants whose flowers open early in the season, as most trees and the vernal herbs, develop their microsporangia before the end of the " growing season," and that the mother-cell stage is the usual win- Fio. 8. — A, Populus monilifera, probably spore mother-cell stage, Jan. 25, 1895 ; x 600. B, Cory lug americana, pollen ready for shed- ding, Dec. 7, 1897 ; x 400.— After Chamberlain.30 ter condition. In the case of such plants, therefore, the earliest stages in the history of the microsporangia must be looked for during the latter half of the growing season that pre- cedes the season of " blooming." This suggests that the natural end of a growing season for the sporophyte is the attainment of the mother-cell stage by its spo- rangia, which is really the limit of the sporophyte in the alterna- tion of generations ; and the natural beginning of the next season is the reduction division and the beginning of the gametophyte. Of course such a distinction disappears in many plants whose seasonal habits are different from those we have been consider- ing, but it suggests a natural division of growth between seasons, and even in annuals the mother-cell stage is a prolonged one. 32 MORPHOLOGY OF ANGIOSPERMS The development of the microsporangia began to be de- scribed bv Xiiii'eli 2 in 1842, and was continued bv Hofmeister 3 in 1859-'61 ; but the first detailed account from the standpoint of modern morphology is that of Warming 8 in 1873, which has been made the basis of all subsequent accounts. This was sup- plemented in 1876 by Engler,9 and since then numerous inves- tigators have added extensively to the literature of the subject. The anther at first is a homogeneous mass of small meriste- matic cells covered by an epidermis (Fig. 9). Very early it Fig. 9. — Development of the mierosporangium. A-D, Doronicum macrophyllum : A, transverse section of very young anther, showing primary sporogenous cell (a) and primary parietal cell (b) ; B, slightly older stage ; C\ longitudinal section of anther in same stage as that shown in B ; I>, later stage ; a, sporogenous cells. E< Menyanth, t trifo/iata, transverse section of a microsporangium at a still later stage showing tapetum (t) and microspore mother-cells (sni). F, Mentha aquatica, transverse sec- tion showing tapetum (t) and sporogenous cells (a). — After Warming, from Goebel's Outlines of Classification and Special Morphology. becomes faintly four-lobed in cross-section, and the differentia- tion of the vascular strand of the connective outlines the gen- eral plan of the structure. The whole hypodermal layer of cells, representing the outermost layer of the periblem, is prob- ably to be regarded as archesporial in its possibilities, and one region of it is just as likely as another, under similar condi- tions, to develop into actual archesporial cells. The favorable conditions for this development, however, are under the lobes : so that almost simultaneously with their appearance, a plate THE MICROSPORANGIUM 33 of hvpodermal cells becomes differentiated in each lobe, dis- tinguished from the adjacent cells by their larger size, their usual radial elongation, their larger nuclei, and their different reaction to stains. In cross-section this plate is a single row of cells of variable number, sometimes almost equaling in extent the contour of the lobe, as in Mentha aquatica (Warming8); sometimes consisting of four to six cells, as in Orchis maculata ( Guignard 10) ; sometimes three or four cells, as in HemerocaJ- lis fulva (Fullmer 24) ; sometimes one or two cells, as in Conval- laria majalis and Potamogeton foliosus (Wiegand25); and sometimes constantly one cell, as long known in Malvaceae and most Compositae, and recently reported in Arena fatua by Can- non.26 In longitudinal section the plate extends approxi- mately the length of the anther, being a single row of cells in case the cross-section consists of a single cell ; but in Mimo- seae the whole archesporium is reported by Eosanoff 4 as being a single cell, as is also the case in Euphorbia corollata, as re- ported by Miss Lyon.22 The general fact becomes clear, there- fore, that an exceedingly variable amount of the hvpodermal layer may become archesporium, from nearly all of it to a single cell; and further, that this amount usually varies within cer- tain limits in the same species, and that the extent of the archesporium is in no way related to the primitive or highly specialized character of plant groups. The subsequent divisions to the mother-cell stage usually follow one another rapidly (Fig. 10). Following the history of a single sporangium, the radially elongated arehesporial cells all divide equally and almost simultaneously by periclinal walls, forming an outer layer (primary parietal,* Fig. 9, A, b) and an inner layer (primary 3porogenous,f Figs. 0, A, a, and 10, * This has been commonly called the "primary tapetal layer," on the ground that the tapetum is one of its derivatives. At most only a part of the tapetum can be derived from it, and in some cases none of the tapetum is so derived. Besides, the tapetum is a physiological layer of variable morpho- logical origin. The essential morphological feature of this outer sterile layer is that it develops the wall of the embedded sporangium, and hence we have preferred to designate it as the primary parietal layer. + This is the "archesporium " of Gocbel's Outlines of Classification and of other texts. With such an application of the term the archesporium of the microsporangium of Angiosperms does not homologize with that of the mega- sporangium, and is of indefinite application among the Pteridophytes-. By 34 MORPHOLOGY OF ANGIOSPERMS A). The names used to designate these two layers indicate their subsequent history, the former producing the wall of the embedded sporangium, and the latter the sporogenous tissue. The cells of the primary parietal layer divide by periclinal walls, so that usually a definite series of concentric parietal layers appears (Fig. 8, A). Walls in other directions also ex- tend the parietal layers uniformly with the rapidly enlarging anther. The number of parietal layers is variable, but in most cases there are from three to five. Sometimes there are only two layers, as inSilphium (Merrell 2S) and in Quercus (Conrad 29) ; and among the Pontederiaceae Smith -1 has regularly found six. Even higher numbers have been reported, and Goebel 1X (p. 368) cites Agave americana as developing eight to twelve fibrous or endothecial layers. In Rhopalocnemis phalloides (Balanophoraceae) Lotsy30 has shown that the sporangia of the curious axial stamen do not organize definite parietal layers and have no method of dehiscence, although the microspores are fully matured. The outermost parietal layer usually develops very differ- entlv from the others, and has been called the " endothecium." This name was given by Purkinje 1 to designate all the layers of the dehiscing anther wrall within the epidermis, which latter he named the " exothecium." Since in most cases the outer- most parietal layer is the only one represented in Purkinje's " endothecium," the name has come to be restricted to it, which seems to us unfortunate, for it should be retained in its original application and used only in connection with the dehiscing anther-wall. It remains true, however, that the outermost pa- rietal layer generally becomes the endothecium, and in the fol- lowing account this condition will be presented. If the anther does not dehisce, the endothecial cells do not become specially modified ; but if the anther dehisces, the cells develop thicken- ing bands in various wTays, the position and extent of these banded cells being directly related to the method of dehiscence (Fig. 7). Between the outermost and innermost parietal layers there are usually one to three " middle layers," and this amount of applying the term to the first cell or group of cells differentiated from the ordinary vegetative cells to produce sporogenous tissue, it is easy of applica- tion and the homologies are definite. Fig. 10. — SilpMum integrifolium. Longitudinal sections of microsporangia ; x 520. A, single row of archesporial cells ; in two cells division into primary sporogenous and primary parietal cells has already taken place, fi, sporogenous and tapetal cells sharply differentiated. C, later stage showing spore mother-cells in synapsis. 2>, a tetrad ("only three microspores shown) formed within the spore mother-cell. — After Merrell.*8 85 36 MORPHOLOGY OF ANGIOSPERMS variation may occur in the same Avail, as noted by Coulter 1!> in Ranunculus. The cells of these layers are usually tabular, and gradually become flattened and disorganized ; but in some cases the one or two innermost middle layers become prominent as a part of the tapetum ; in others the outer ones may become a part of the endothecium ; and occasionally there is no dis- organization of parietal layers. The innermost parietal layer, as a rule, becomes part of the tapetum, which is a jacket of nourishing cells in immediate con- tact with the sporogenous tissue (Figs. 9, 10). The tapetum has D Fig. 11. — Zostera marina. A, young mierosporangium with archesporial cells shaded. -ff, later stage showing tapetum derived from sporogenous cells ; t, tapetum ; p, pollen mother-cells ; st, sterile cells, as shown by transverse wall. C, portions of the two long cells resulting from the first division of the pollen mother-cell. Z>, portion of a microspore showing the nuclear division that gives rise to the generative and tube nuclei ; there are six chromosomes. £, the filiform pollen grain. — After .Rosenberg.32 no definite morphological boundary or origin, but results from pressing into special physiological service the sterile cells, of whatever origin, contiguous to the sporogenous tissue. While one THE MICROSPORANGIUM Fig. 12. — Lemna minor. Section of microspo- rangium showing some of the spore mother- cells broken down and functioning as tape- turn ; x 1100. — After Caldwell." layer of cells is the rule, the tapetum may include two or more layers, as pointed out by Frye 33 in Asclepias. The same inves- tigator has also followed the origin of that portion of the tape- rum next the connective from the plate of cells im- mediatelv within the arche- sporium; and in a recent paper Rosenberg 32 de- scribes and figures the much elongated arehespo- rial cells of Zostera as cut- ting off isodiametric cells at each end, that divide more or less and form the tapetum on the outer and inner surfaces of the spo- rogenous mass (Fig. 11, B, t). There is evidence, therefore, that in certain cases the tapetum, or at least part of it, may be derived from sterile cells cut off from the .periphery of the sporog- enous mass. Such a probability islalso reported by Coul- ter19 in Ranunculus, and by Webb34 in Astilbe. Enough is known, at least, to lead to the conclusion that any sterile cells in contact with the sporogenous tissue assume the tape- tal function. This is a well-known fact in connection with sterile mother-cells, which in this sense are a part of the tape- tum. Among the Pontederiaceae Smith 21 . found that the tapetal cells, closely adherent to the mother-cells, are often wedged among them; and in Lemna Caldwell23 observed that the cells of the regular tapetal jacket often divide and form groups of cells projecting deep among the mother-cells, sterile mother-cells also disintegrating (Fig. 12): while in Symplo- carpus Duggar 27 reports that the tapetal cells become free and " wander " among the mother-cells. Tt seems clear, therefore, that the tapetum is a set of sterile cells that nourish the func- tioning mother-cells, and that while ordinarily it is a definite layer none of which is derived from the primary sporogenous cells, it may include a variety of morphological elements. As a rule, the complete organization of the tapetal jacket is 38 MORPHOLOGY OF ANGIOSPERMS D coincident with the mother-cell stage, but the greatest devel- opment of the tapetal cells is during the formation of tetrads. During this process they may increase greatly in size, this being associated with the disorganization of the cells of one or more of the middle layers. It is very common for the enlarged tapetal cells, filled with food material, to become binucleate (Fig. 10, C), and later even multinucleate, as in Typha (Schaff- ner17) and llepatica (Coul- ter19), in the latter genus six to thirteen nuclei hav- ing been observed in a sin- gle cell. At the end of the tetrad division the tapetal cells usually become disor- gaiiize-dv- also such of the middle layers as have not disorganized previously, and - the outermost parietal layer begins to develop the thick- enings characteristic of the endothecium. The fact that the endothecium may con- sist of additional layers of cells has already been men- tioned. During the development of the parietal layers the primary sporogenous cells either directlv or bv divi- sion produce the mother- cells. When division oc- curs, it is in every direc- tion, so that all appearance of layers is lost. Perhaps the usual case is for the primary sporogenous cells to divide two or three times, but there are sometimes more divisions, and a number of cases are known in which the primary sporogenous cells, without division, become mother-cells, as has been long known in Malva, Datura, Mentha, and Chrysanthemum, and recently reported for several species of Asclepiadaceae by Stras- Fig. 13. — A and D, Or-chis maculata : A, trans- verse section of an anther with four micro- sporangia, each showing five or six cells, each of which gives rise to a " massula" as shown in D. B, C, and £, Neottia ovata : B, a tetrad ; C, the four microspores within the common wall dividing to form tube nu- cleus and generative cell ; £, the division completed ; two of the microspores show the generative cell cut off by a lenticular wall. A x 2o\ D x 240; B, C, E x 305.— After GuiONARI).10 THE MICROSPORANGIUM 39 burger 31 and by Frye.33 The case of certain orchids, such as Orchis maculata, Calanthe veratrifolia, and Neottia ovata, in- vestigated by Guignard,10 and their allied forms, deserve special mention. Each primary sporogenons cell gives rise to a well- defined mass of mother-cells known as a massula (Fig. 13, A, D), and separated from its fellows by thicker walls. The mother-cells and their nuclei usually increase very much in size, and differ from the adjacent cells in their reaction to stains. This growth is usually accompanied by a rounding of the cells and their separation from one another, and also by a thickening of the wall; but in many Dicotyledons (Tropaeo- lum, Althaea, etc.) the mother-cells do not become isolated, and remain packed closely together in the sporangium, due probably to the tardy disorganization of the tapetum or its failure to disorganize. The time relations of the events described to those that form the history of the corresponding megasporangium are exceed- ingly variable, but the case of Astilbe, as described by Webb,34 may be taken as an illustration, especially as it probably rep- resents an average ease. The microsporangia develop rapidly, maturing in one or two weeks, and precede the megasporangia stage for stage. The anthers are rounded and enlarged before the carpellary cavity is closed over; the four microsporangia are well marked when the " placentae " are wholly undiffer- entiated; the tapetum is organized and the mother-cell stage reached when the integuments have not appeared ; during the tetrad stage the integuments appear, while the microspores are " rounded off " before the functional megaspore is determined. The most extreme cases are probably certain orchids in which pollination occurs before ovules are formed; and oaks (Con- rad29), in which pollination occurs one spring and the ovules do not develop until the next. The case of Lemna, as reported by Caldwell,23 deserves sepa- rate mention (Fig. 14). In the nascent anther a single hypo- dermal group of cells appears as an archesporium and enters upon the usual history, a wall of several layers and a broad spo- rogenous mass being formed. A plate of sterile cells gradually divides this sporogenous mass into two, each of which continues to divide as the anther increases in size. Each of these two sporogenous masses is divided by a plate of sterile cells, so that 4 40 MORPHOLOGY OF ANGIOSPERMS four distinct sporogenous groups are formed, each surrounded by its own tapetum. As a result, the mature anther seems to contain the usual four sporangia. Such a case makes the defi- nition of a sporangium difficult. If a single archesporium is the criterion, Lemna has a single sporangium ; if a group of mother-cells invested by a tapetum is the criterion, it has four sporangia. The explanation probably lies in the fact that the whole outer layer of the periblem is capable of becoming trans- Fig. 14. — Lemna minor. Development of microsporangium and sporogrenous tissue. A young stamen with sporogenous cells. £, two young stamens ; in the one at the left the sporogenous tissue is becoming divided by a sterile plate. 0, a more ad- vanced stage than B. D, a single stamen showing the four masses of sporogenous tissue well separated by sterile plates. A > 1100; £, C, D x 712. — After Cald- well.23 formed into an archesporium, and that while in ordinary cases archesporial tissue is developed in four separate regions, in Lemna the conditions favor a more general development. To divide a large sporogenous mass by sterile plates for better nu- trition is too common to call for any special remark. As for the definition of a sporangium, it is probably not best to define it too rigidly, but to use the term as one of convenience. From THE MICR0SP0RANG1UM 41 this standpoint, there is no objection to speaking of the four groups of mother-cells in Lenuia as four sporangia, which have had quite an exceptional origin. The phenomenon is too unique as yet among Angiosperms to justify any generali- zation. The growth of mother-cells and the enlargement of the spo- rangial cavities usually result in reducing to a thin plate the sterile tissue separating the two sporangial cavities on each side of the anther. As dehiscence approaches, this plate usually disappears, and the two sporangial cavities become fused into a single loculus of the anther (Fig. 7). In the mature condi- tion, therefore, such an anther contains two loculi or " pollen- sacs." While this represents the ordinary condition of the mature anther, among the Araceae it is reported that the single loculus of the anther is formed by the fusion of four sporan- gial cavities, and in Sassafras it is well known that the four remain separate. In case an anther contains only two sporan- gia, as among Asclepiadaceae, there is no fusion, and each loculus is a single sporangial cavity. The mechanism for the dehiscence of anthers is extremely varied (Fig. 15), and needs much more investigation than it has received. By far the most common method is by means of a longitudinal fissure, a definite stomium developing, as in Lilium (Fig. 7), and opening by means of the drying of the anther-walls, the contraction of the epidermal cells being- greater than that of the endothecial cells with their thick bands. There is also dehiscence by a short apical fissure, as in Solarium and certain Ericaceae; by a terminal pore, formed by the dis- organization of a small portion of tissue, as in certain Erica- ceae; by hinged valves, as in Berberis, Sassafras, and Hama- melis; and by irregular breaking and exfoliating of superficial tissues, as in the axial stamens of Naias. The details of these methods, and of others, should be investigated from the stand- point of the development of the mechanism, for snch as we have are too vague and superficial to be of much significance. The mother-cell stage of the microsporanginm is regarded as the end of the history of the sporophyte in this direction, chiefly because the division of the mother-cell, preceded by a more or less prolonged rest, is a reduction division, and in con- sequence the resulting cells have the feature most characteristic 42 MORPHOLOGY OP ANGIOS PERMS of gametophytic tissue — namely, the reduced number of chromo- somes (Fig. 53). With this division, therefore, the history of the male gametophyte begins. This line of demarcation be- tween sporophyte and gametophyte is easy to define, but does Fig. 15. — Forms of stamens. 1, Calandrinia compressa; 2, Solanvm Lycopergicum; 3, Galanthus nivalis; 4i Cyclamen europaeum; 5, Ramnndia pyrenaica; 6. 7, Cassia lenitiva\ 8, Pyrola rotund i folia; 9, Arctostayhylos Uva-ursi; 10, A.al/pina; 11, Yaccinium uliginosum: 12, Pyrola unijlora; 13, Medinilln (after Baillon); 14, Yaccinium Oxycoccus; 15, Calceolaria Pavonii; 16, Tozzia alpina; 17, 18, tjihbaldia procumbens; 19, Galenjt.iis angusti folia ; 20, 21, Erythraea Csntaureum ; 22, 23, Me- lissa officinalis; 24, 25, Calla palustris; 26, Xyctandra (after Baillon); 27, 28, Globularia cordifolia; 29, 30, Theobroma Cacao; 31, Pinguicula vulgaris; 32, Garcinia. — From Kerner's Ptfanzenleben. not result in so simple a conception of the alternating genera- tions as to begin the gametophyte with the germinating spore, for it involves the simultaneous origin of four gametophytic generations from the mother-cell through an intermediate divi- sion. The claim that the reduction division must be regarded THE MICROSPORANGIUM 43 as ushering in the gametophyte was first urged by Stras- burger,12 whose paper closes as follows: " The reduction in number of the chromosomes takes place, among the higher plants, in the mother-cells of the spores, and it is consequently these which must be regarded as the first term of the new generation. They assert this their true sig- nificance in that thev usually isolate themselves from cohesion with other cells and become independent, although this inde- pendence is only of practical utility in the case of the products of their division — that is, of the spores. Hence the center of gravity of the developmental processes which take place in both micro- and macrosporangia of Cryptogams and Phanerogams does not lie in those cells, cell-rows, or cell-aggregates which give rise to the sporogenous tissue and have been designated ' archesporium ' by Goebel. The archesporium still belongs to the sexually developed asexual generation; it is only the spore mother-cells which initiate the new sexual generation ; consequently the presence or absence of a well-defined arche- sporium is not a matter to which importance should be attached." LITERATURE CITED 1. Purkixje. J. E. De cellulis antherarum fibrosis nee non de gra- nopnm pollinarium formis commentatio phytotomica. Yrati- slaviae. 1830. 2. Nageli, C. Zur Entwicklungsgesehichte des Pollens. Zurich. 1842. 3. Hofmeister. W. Xeuere Beobachtungen fiber Enibryobildung der Phanerogamen. Jahrb. Wiss. Bot. 1 : 82-188. pis. 7-10. 1858. 4. Rosaxoff. S. Zur Kermtniss des Baues mid der Entwicklungs- gesehiehte des Pollens der Mimoseae. Jahrb. Wiss. Bot. 4: 441- 450. pis. 31-32. 1865. 5. Rohrbach. P. Die Samenknospe der Typhaceen. Bot. Zeit. 27 : 479-480. 1869. 6. Kauffmaxx. N. Ueber die mannlichen Blfithe von Casnarina quadrivalvis. Bull. Soc. Nat. Moscou 41 : 374-382. 1869. 7. MAGNUS. P. Zur Morphologie der Gattung Xaias L. Bot. Zeit. 27: 7('i9-773. 1869. Also Beitriii$, Zamn- chellia, Lilaea, Piperaceae, Juglandaceae, Polygonaceae, and others. In the case of laterally cauline ovules apical growth of the axis may be checked, so that the lateral ovule appears to arise in a terminal position from the bottom of the sporangial chamber, as among the Compositae; or the apical growth may be continued into the sporangial chamber as a dome-shaped (Anagallis arvensis) or columnar (Spergularia rubra, Amaran- taceae, etc.) structure upon which numerous lateral ovules are 46 THE MEGASPORAXGIUM 47 borne, giving rise to the so-called " free central placentation " of the older botanists. Canline ovules have also been reported Fig. 16. — A, BalanopTiora polyandra, with archegonium-like megasporangia ; x 15. B. B. dioiea. a younger stage showing the mother-cell just after the first division; x 200. C, B. polyandra," style'' with a pollen-tube growing down into the "stylar canal"'; x 105. I\ B. dioiea, longitudinal section of a nearly ripe seed; the sus- pensor is not shown in this section. E, a similar section through the endosperm and embryo, showing the suspensor. — After IIofmeister.8 in Myzodendron punctulatum by Johnson,22 and they doubtless occur in other Santalaceae; and in Sparganium simplex, Lilaea 48 MORPHOLOGY OF ANGIOSPERMS subulata, and certain of the Araceae by Campbell 37- 47> 4t> ; and there is no doubt that numerous other cases await discovery. It should be remembered also that in many cases of epigyny the ovules are probably to be regard- ed as cauline, and if these be added to the cases already indi- cated, it becomes evident that cauline ovules are exceedingly common and occur in all grades of Angiosperms. In this connection the curious condition in Loranthaceae and Balanophoraceae may be consid- ered, a condition that may have some connection with their pecul- iar habits. In 1858 Hofmeis- ter 4j 5 studied various species and outlined the prominent feature i of these groups, describing and Yio.ll -Balanophoraglobosa A l,nu- illustratinD. geveral stages in tbe cell us with mother-cell of embryo- o o sac (shaded) ; the epidermal cells development of the archegomum- above the mother-cell give rise to the l^g megasporailgiuill, and also of outgrowth resembling the neckof an endosperm and embrVO of archegonmm. B, later stage in which l the mother-cell has divided into two Balanophoraceae (Fig. 16), and cells, both of which "very often de- a]g0 foe puzzling " lliamelon " ill velop into embryo-sacs": x 166. — T , 0 , , • AA/T 4S Loranthaceae. Subsequent mves- After Lotsy.48 1 tigators have in the main con- firmed and extended this work, the most important modifi- cation being in the interpretation of the embryo; and even here Hofmeister's figures are nearly identical with those of the most recent papers (cf. Fig. 16 with Fig. 107V In 1882 Treub 18 described the development of the pistil of Loranthus sphaerocarpus (Fig. 19). A structure (" mame- lon") arises from the bottom of each of the three or four sporangial chambers and grows until it completely fills it, and in this structure hypodermal archesporia appear and develop megaspores in the usual way. It is a fair question whether the " mamelon " is a growth of the axis, whose ovules, represented by separate archesporia, are mechanically hindered from any superficial development; or whether it is an ovule THE MEGASPORANGIUM 49 without an integument, in which there are several archesporia. Hofnieister favored the latter view, while Trenb inclined to the former, as his explanation of it as a fusion of rudimentary ovules and placentas would seem to indicate. In 1883 Treub 19 discovered exactly the same structure in Loranthus pentandrus. In 1895 the same investigator29 described Balanophora elon- gata as having no ovule or placenta. In 1896 this was con- firmed by Van Tieahem 34 for B. indica ; and in 1899 bv Lotsy 4S for B. globosa (Fig. 17). Lotsy claims that in B. globosa there are no flowers, carpels, placenta, or ovules ; but that a hypodermal cell in a protuberance of the floral axis gives rise to the embryo-sac, while the epidermal cells over it de- velop a long, style-like organ resembling the neck of an archegonium. Hofmeister de- scribes and figures the pollen- tube of B. polyandra as grow- ing down into this " stylar canal," as he called it (Fig. 16, C). It would appear from the figures that the " protu- berance of the floral axis " is a megasporangium without in- teguments, and that the so- called " stvle " is a remarkable outgrowth of the nucellus. The pollen-grains, as figured by Hofmeister, therefore, come in contact with the nucellus, as in Gymnosperms. In this connection attention may be called to the remarkable beak developed by the nucellus of Euphorbia corollata as de- scribed by Miss Lyon 41 (Fig. 18), a beak which suggests the same general tendency of the nu- cellns which has reached such an extreme expression in Balano- phora. The investigation of Rhopaloenemis phalloides (Balano- phoraceae) by Lotsy,52 however, as well as the case of Balano- Fio. 18.— Euphorbia corollata. Longitudi- nal section Bhowing an excessive pro- longation of the nucellus; x 650. — After Lyon." 50 MORPHOLOGY OF ANGIOSPEKMS phora, suggests the explanation. Lotsy finds that the enlarged tip of the flower axis soon completely fills the cavity of the ovary, and that one or more hypodernial cells of this axis form the mega- spores (Fig. 20). This is exactly the case of Loranthus, and suggests that in the allied Balanophora the B Fig. 19. — Loranthus sphaerocarpus. A, longitudinal section of a young flower showing the " mamelon " (m) ; x 25. J3, longitudinal sec- tion of a "mamelon" at a later stage showing two hypodermal archesporial cells ; x 300. — After Treub.18 same " mamelon " is present, but with no carpellary investment, the naked nucellus (as the "mamelon" would seem to be in this case) de- veloping the remarkable neck-like outgrowth of sterile tissue. In both families it seems certain that the megasporangia are cauline. Foliar ovules are related to the carpels in a variety of ways. By far the most common position is for the ovules to arise in a line along each side of one of the two prominent vas- cular bundles of the carpel, a very common position for the sporangia of ferns. In the older morphology this line was thought to represent the abutting margins of the infolded car- pellary leaf, and hence such ovules were called " marginal." In fact, this double line of ovules, and the dehiscence of many carpels along it, seemed to the supporters of meta- morphosis to prove the foliar nature of the carpel. As might be expected from the behavior of sporangia in ferns, there are cases in which ovules arise without such close connection with a prominent vascular bundle. For example, in Butomus, Nymphaea, Nuphar, Obolaria, Bartonia, and many species of Gentiana, the ovules arise from the whole inner surface of the carpel. In the older terminology these were called " super- ficial " ovules, and associated with them, curiously enough, were THE MEGASPORANGIUM 51 the occasional cases in which the ovules arise from the other vas- cular bundle (the " midrib " of the infolded leaf theory), as in Brasenia, Cabomba, and Astro car pus (Eichler,8 2: 17). Ac- cording to Warming 7 a third category is necessary to include such cases as Zannichellia, Ranunculus, and Sedum, in which he says the ovules are " basal or axillary." The general conclusion seems evident that ovules may arise from any free surface within the cavity of the ovary, whether it be morphologically carpel or axis ; and further, that if the cavity of the ovary becomes obliterated by the enlarged tip of A Fig. 20. — Rhopalocnemis pJialloides. A, longitudinal section through the "mamelon" before the appearance of archesporial cells. £, later stage showing the two mega- spore mother-cells which develop directly into embryo-sacs, x 116. — After Lotst.m the axis, as probable in Loranthaceae and Balanophoraceae, megasporangia arise from the hypodermal cells of the axis without the definite organization of ovules. The morphological nature of the ovule was much discussed by the older morphologists. According to the theory of meta- morphosis it was necessary to interpret it as a transformation of some one or more of the vegetative members. The most prevalent view was that it is a transformed leaf-bud arising from the margin of the carpellary leaf, as in the well-known case of Bryophyllum ; and ITofmeister claimed that the ovule of Orchis is a trichome because it arises from a single epidermal 52 MORPHOLOGY OP ANGIOSPERMS cell. When cauline ovules came to notice, Schleiden, End- lieher, and others took the extreme position that all ovules are cauline, even those evidently parietal upon carpels. This view was opposed by Van Tieghem,6 Celakovsky,9 and especially by Warming.10 The last-mentioned paper is noteworthy for its presentation of the origin and development of the ovule, as well as for its discussion of its morphology. These writers maintained that the ovule is always foliar in origin, and their explanations of cauline ovules are interesting on account of their ingenuity. This view was also maintained by Eichler in his Bluthendiagramme, where an historical resume of the whole subject may be found. The most interesting feature of the whole discussion, however, is the persistent idea that ovules could not be both foliar and cauline. These last observers, hav- ing established the foliar origin, disproved the bud character of ovules, since the members of leaf-buds arise in acropetal succession, while the nucellus and integuments are basipetal. It was urged that the ovule is a transformed leaf-lobe or leaf- outgrowth, and that this view homologized them with the spo- rangia of ferns. This was a decided step in advance, and it only remained to abandon the doctrine of metamorphosis, and to see that the ovules (sporangia) hold no necessary relation to either leaf or stem, but are themselves long-established and independent members of the plant body, with a history that antedates that of either stem or leaf. The length of time from the beginning of megasporangia to their maturity is very indefinitely known, as most investigators do not seem to have kept such a record. It must be extremely variable, as in the case of the microsporangia, and related to the seasonal habit of the plant. In Sallx and Popuhts Cham- berlain39 found that the megaspore mother-cells are not distin- guished until the renewal of growth in the spring, although the microsporangia pass the winter in the mother-cell stage; and this lateness of development may be usual in the megaspore series. Enough cases have been observed, however, to show that a much earlier development may often occur. For example, in Acer rubrum Mottier 27 discovered the mother-cell stage in March or earlier, the indication being that this is the winter condition; Chamberlain39 found the four megaspores of Tril- lium recurvatum fully formed early in April, when the plants THE MEGASPORANGIUM 53 were not more than 5 centimeters high, and the embryo-sac of Hepatica ready for fertilization while the ground was still frozen ; we have seen embryo-sacs of Epigaea ready for fertil- ization in the autumn, pollination probably occurring the fol- lowing spring; and Schaffner 56 has found that in Erythronium the archesporial cell begins to enlarge about the first of October and nuclear changes occur, and that by the first of December the nucleus is very large and the mother-cell stage reached, which persists until early spring. The subject should be investigated especially in connection with vernal herbs and early blooming shrubs and trees. The details of the development of the ovule have been ad- mirably given by Warming 10 and Strasburger,13 supplement- ing and correcting the earlier observations of Hofmeister,4, 5 and the literature since has grown so extensively that full cita- tion is impossible (Fig. 21). At first the epidermis of the mem- ber upon which the ovule is to appear is even, and in the hypo- dermal layer the archesporium may or may not be evident. A slight protuberance is developed by cell-divisions, which are usually only radial in the epidermal layer, resulting in a more extended plate of cells ; but in the hypodermal layer they are variable, resulting in a mass of tissue, or in many cases in but a single axial row of cells. In any event, the protuberance becomes more and more prominent and constitutes the nucellus of the nascent ovule. After the nucellus has become prominent, an annular out- growth begins at its base, and with greater or less rapidity develops into the inner and often only integument, in most cases overtopping the nucellus (Fig. 3, I). In case there is an outer integument, its annular primordium becomes visible soon after the inner integument is well under way (Fig. 21). If the aril be placed in this category, it has been observed that this third integument arises much later than the other two, usually after fertilization, as in Asphodehis, Aloe, Nympliaca, Podo- phyllum, Euonymus, Celaslrux, Myristica, etc., although in all these cases its point of origin does not seem to be well estab- lished. Disregarding the aril, two integuments seem to be the rule among Monocotyledons, the only recorded exception we have noted being Crinum, although, doubtless, there are others. Two integuments prevail among the Archichlamydeae also, the 1/ /v. Fio. 21. — Lilium philaJelphicum. A, ovule before the appearance of integuments, showing a single hypodermal archesporial cell which is also the megaspore mother- cell. B, beginning of inner integument. C, beginning of outer integument. D, E, later stages. F, G, the ovule anatropous and the megaspore germinating, x 175. 54 THE MEGASPORANGIUM 55 Umbelliferae being the most notable exception. On the other hand, a single integument is characteristic of the ovules of the Sympetalae, as well as of the Umbelliferae, and some other Archichlamydeae, such as species of Ranunculus, Legmninosae, etc., the integument being very massive and in comparison with the very small nucellus constituting the bulk of the ovule. There seems to be every indication that two integuments are characteristic of the ovules of the more primitive Angiosperms ; that they persist among Monocotyledons even among the most highly specialized families ; but that among Dicotvledons thev are replaced in the higher groups by the single massive integu- ment. The fact that the single integument is more massive even than both integuments when there are two suggests that it represents two integuments in the sense that their primordia are no longer differentiated. This is very far from meaning that two integuments have fused to form the single one, but that a single integument is developed by the same region that in other cases produces two. Certain exceptional cases in the development of integu- ments may be noted. Among the Loranthaceae and Balano- phoraceae no integuments are formed ; and the same claim is made by Chauveaud 2i' 65 for Cynanchum (Asclepiadaceae), perhaps to be explained by Asclepias (Frye66), in which the integument might be mistaken for part of a naked nucellus. The same claim is made for Santalaeeae, and it may be true of most of them; but in Myzodendron punctulatum Johnson 22 has described a single-layered integument that does not cover the free end of the embryo-sac. This suggests an abortion of the integument, which in other members of the family may not have been recognized or may even have been sup- pressed. The ovule of Houstonia is said by Lloyd 61 to have no integument. The ovule of Hippuris long had the reputation of having no integument, as reported by Schleiden,1 Unger,2 and Schaeht.3 In 1880, however, Fischer 15 in reinvestigating it discovered that a single integument is formed, but closes over the nucellus so tightly as to give the appearance of a naked nucellus. Oliver 21 discovered exactly the same behavior in his new genus Trapella, except that the integument is very massive. The same thing has also been observed by !Murbeck " in the parthenogenetic Alchemilla alpina, in which the single 56 MORPHOLOGY OF AXGIOSPERMS integument so completely coalesces with the nucellus and closes the micropvle that the ovule resembles a naked nucellus. Zin- ger 43 observed that the massive inner integument in Canna- bineae is completely coalescent with the thick outer one over the apex of the nucellus, and the micropylar canal becomes entirely closed by the development of tissue. In cases of chalazogamy and persistent parthenogenesis such behavior of the integu- ments may be expected, as well as in other cases whose habits do not suggest it. In most cases, the ovule does not merely become distinct from the surface of the member that produces it, but is borne upon a stalk-like base or funiculus. It is generally stated that n ~\-\\- Fig. 22. — Forms of ovules (diagrammatic). A, orthotropous ; B, anatropous ; C, cam- py lotropous ; m, micropvle ; e, embryo-sac ; «, nucellus; c, chalaza;/, funiculus. — After Prantl in Engler and Prantl's Nat. Pflanzenfam. the ovules of Gramineae have no funiculus, but it would be im- possible to draw an exact line between its presence and absence. The direction of growth of the ovule seems to be related to the orientation of the micropvle in reference to the pollen- tube. Mirbel gave to the resulting forms the names ortho- tropous, campylotropous, and anatropous (Fig. 22). In the first case the growth continues without the development of any curvature, a fact generally true of terminal cauline ovules. Orthotropous ovules are quite common, being found among Monocotyledons in the Restiaceae, Eriocaulaceae, Xyridaceae, certain Araceae, Commelinaceae, etc. ; and among Dicotyledons in the Piperaeeae, Urticaceae, Polvgonaceae, Oistaceae. etc. These are relatively primitive families of Monocotyledons and Archichlamydeae, and confirm the natural impression that the THE MEGASPORANGIUM 57 original angiospermous ovules were straight. The campylotro- pous ovule, in which the whole body of the ovule curves, is the rarest type, among Monocotyledons characterizing the Grami- neae, Scitamineae, etc., and among Dicotyledons the Cheno- podiaceae, Caryophyllaceae, Cruciferae, Capparidaceae, Reseda- ceae, etc. These families are more or less specialized members of their alliances, and none of them belong to the Sympetalae. By far the most common form of ovule is the anatropous, and although it is extensively displayed among Monocotyledons and Archichlamydeae, it is present almost without exception among the Sympetalae, and may be regarded as the most highly spe- cialized type of ovule. In its development an anatropous ovule i- at first straight or nearlv so, but very earlv develops a curva- ture at a level with the origin of the first or only integument. As the integuments grow the curvature increases, and usually before the outer integument is complete the nucellus is inverted against the funiculus (Fig. 21). For this reason, in anatropous ovules with two integuments the outer one is not developed on the side toward the funiculus. In abnormal material of Salix petiolaris both anatropous and orthotropous ovules have been observed in the same ovary (Fig. 6). The archesporium, as in the microsporangia, is recognized bv the increasing size and the different reaction to stains of one or more hypodermal cells. Doubtless all of the hypodermal cells are potentially arehesporial, and there is reason for be- lieving that the deeper cells of the nucellus, most of which are probably derivatives from the original hypodermal layer, may be also. In the vast majority of cases, however, only cells of the hypodermal layer show those changes that are character- istic of arehesporial cells. It is not always easy to determine just how many hypodermal cells are to be included in the ar- chesporium, for there is often complete gradation from cells with the size and staining reaction of undoubted arehesporial cells to those showing neither increase in size nor the character- istic staining reaction. This is to be expected in case all the hypodermal cells are potentially arehesporial, and there is no definite point in its history when such a cell ceases to be merely hypodermal and becomes clearly arehesporial. For this reason, the number of cells recorded as constituting; the archesporium in many plants can not be regarded as precise, but as approxi- 58 MORPHOLOGY OP ANGIOSPERMS mate. The prevailing habit, however, is to limit the arche- sporium to the single hypodermal cell that terminates the axial row of the micellus. This seems to have resulted in the more Fig. 23.— Longitudinal sections of ovules showing: multicellular archesporia. A. B, Astilbejaponica, x 550; after Webb." C, Salix glaucophylla, x 600; after Cham- berlain.36 D, Rosa livida, x 224; after Strasburger.13 £, Alchemilla alpina, x 275 ; after Murbeck.67 F, Callipeltis cuctdlaria ; after Lloyd.81 G, Quercus veluUna, x 720 ; after Conrad.63 highly specialized groups in reducing the nucellus within the epidermis to this axial row, as Lilium, many Orchida- ceae (Dumee44), Lobeliaceae (Marshall-Ward14), Rubiaceae (Lloyd61), Compositae, and many other sympetalous groups. In such cases the nucellus in longitudinal section shows only three rows of cells. It is of interest to note the recorded cases in which the archesporium is said to consist of more than a single cell (Fig. 23). In 1870 Strasburger 13 described the several-celled ar- chesporium of Rosa livida, and in 1880 Fischer15 reported a similar archesporium in Geum, Sanguisorba, Agrimonia, Ru- THE MEGASPORANGIUM 59 bus, and Cydonia, indicating that this is the prevailing tend- ency among the Rosaceae. In 1882 Guignard 17 added Erio- botrya to the list, and in 1901 Murbeck 57 found an archesporial group in Alchemilla alpina. Recently, however, Pechoutre 63 has made a general survey of the Rosaceae, and in all of the fourteen genera studied, well distributed among the tribes, there was found a many-celled archesporium, showing a remarkable persistence of this character throughout a large family. Among the closely allied Saxifragaceae also, Webb 60 has found in Astilbe this same type of archesporium. In 1891 Treub 23 published his account of Casuarina, re- porting that the archesporium is a group of hypodermal cells, and that the derived sporogenous tissue forms a large central mass within the nucellus (Fig. 24). The account and the fig- Fio. 24. — Casuarina. Longitudinal sections of nucellus. A, section showing two pri- mary sporogenous cells (shaded ) ; x 190. £, later stage showing extensive sporog- enous tissue: x 190. C, pollen-tuhe (with heavier walls) among the elongated sterile megaspores ; x 67. D, portion of nucellus at a much earlier stage than C, showing numerous megaspore mother-cells ; x 157. — After Tkeub.*3 ures suggest that all of the sporogenous tissue may not be derived from the hypodermal layer. In 1894 Miss Benson 28 discovered that a several-celled archesporium is present in Fa- 62 MORPHOLOGY OF ANGIOSPERMS The archesporial cells behave as do those of the mierospo- rangrum, and in case the archesporiimi is a plate of cells, the resemblance is striking. In the large majority of cases, how- ever, the archesporiimi is a single cell, and often by transverse division it gives rise to a primary parietal cell and a primary sporogenons cell (Fig. 26). That the former cell, or plate of cells, as it is in the case of a several-celled archesporiimi, repre- sents the primary parietal layer of the microsporangiimi seems clear. In recognition of this fact Strasburger called it the " tapetal cell," but for reasons given under the microsporangiimi we shall call it the parietal cell— that is, a cell that develops in part the wall of the em- bedded sporangium. Mottier 25 has reported a very peculiar case in Ari- saema, in which the single archespo- rial cell divides by anticlinal walls into three or four cells, each of which then cuts off a parietal cell. Just how7 far this is exceptional behavior B Fig. 26. — Salix glaucophylla. Lon- remains to be seen, but it intro- duces an interesting problem as to the application of the term archespo- riimi. The behavior of the primary gitudinai sections of nuceiius, parietal cell is exceedingly varied. x 631. A, single hypodermal . _ . - archesporial cell (a). B, arche- A] sporial cell has given rise to pri- mary parietal cell (t) and pri- mary sporogenous cell (m). — After Chamberlain.36 periclinal divisions to occur, result- ing in a long row of parietal cells, corresponding to the parietal layers of the microsporangiimi. In case there is a plate of archesporial cells the radial rows of parietal cells are very conspicuous, as in the Rosaceae and many of the Amentiferae (Figs. 23, B, D, E). In other cases the parietal rows become lost by the formation of anticlinal walls. If the mother-cell broadens rapidly, the first divisions of the pri- mary parietal cell may be anticlinal, followed by periclinal divisions, as in Ruta graveolens (Guignard 17) and Potamoge- ion foliosus (Wiegand 54). The deep-placing of the sporoge- THE MEGASPORANGIUM 63 nous cells beneath parietal tissue occurs in Potamegeton (Wie- gand,54 Holferty55), Triticum (Koernicke33), Agraphis (Vesque12), Triglochin (Vesque 12), Lysichiton (Campbell47), Rosaeeae, Saxifragaeeae, many Leguminosae (as Lupinus, Cer- cis, Acacia), Euphorbiaceae, Cuphea (Guignard 1T), Fuchsia (Vesque12), Mesembvianthcmum (Guignard17), and doubtless manv other Monocotyledons and Archichlamydeae. From a conspicuous deyelopment of parietal tissue there is a complete gradation to its entire suppression. A few peri- clinal diyisions of the parietal cells may occur or none at all. Sometimes in case the periclinal diyisions haye been abandoned, one or more anticlinal diyisions may be induced by the broad- ening of the mother-cell, as the single periclinal diyision in Typha ( Schaffner 3tage, showing four nuclei at each end of the sac-like pollen-tube. — After Xkmec.6* pressure of the surrounding tissue, there being no opportunity for early isolation and rounding off as in miero>porangia. It may lie of interest to note that sometimes after the first divi- sion of the mother-cell the outer dauc'hter-cell divides bv an anticlinal rather than a periclinal wall, as observed in Butomus 76 MORPHOLOGY OF ANGIOSPERMS (Marshall-Ward8), Jeffersonia (Andrews37), and Potamoge- ton (Holferty 97). In the case of Cynomorium (Juel121), the two cells resulting from the first division of the mother-cell are very unequal, the micropylar one being the smaller. This smaller cell divides longitudinally and the larger one trans- versely. Transitions to this condition in the formation of oblique walls sometimes occur, as in Delphinium (Mottier36). The case of Fatsia japonica has been referred to above. As already indicated, the completion of a tetrad is by no means always attained, for there is every gradation between a row of four megaspores and an undividing mother-cell that func- tions directly as a megaspore. The explanation of this tend- ency to shorten the megaspore series is probably connected with the fact that only one megaspore of the tetrad functions. The number of megaspores formed by the mother-cell has been reported for very many plants, but the records are by no means of equal value. The reasons for this are obvious. One is that the sterile axial cells of the nucellus often simulate megaspores, so that too large a number might easily be reported, and great care is necessary to distinguish them ; and another is that the technique of the earlier observers did not always permit cer- tainty. By far the most important source of inaccuracy, how- ever, is the hasty examination of a great number of forms by a single investigator. Two megaspores might be reported, when the same ovule collected a few hours later might have shown four megaspores. As a consequence, much of the avail- able data can be used only in a very general way as indicating tendencies of groups. Among the Monocotyledons, about one-third of those investi- gated are reported as forming complete tetrads, in another third the mother-cell does not divide, while the remaining forms show every intergradation. Although one might expect the complete tetrads to be characteristic of the more primitive Monocotyledons, and the undividing mother-cell characteristic of the higher families, there is as yet no such evidence, both conditions occurring in all grades of Monocotyledons. The greatest variability is found among the Liliaceae, pos-i- bly because more of the species have been investigated (compare Fig. 28 with Figs. 35 and 36). For example, without attempt- in- to include all the recorded cases, in Hemerocalli* (Stras- THE FEMALE GAMETOPHYTE i ^ burger5), Trillium (Chamberlain52), and Galtonia (Schnie- wind-Thies 95) four megaspores are reported, although in the last genus only two may appear; in Anthericum (Strasbur- ger5), and Tricyrtis and Yucca (Guignard12) three; in Alli- um (Strasburger 5), and Agraphis and Ornithogalum (Guig- nard12) two; while in Lilium, Fritillaria, Funkia, Tulipa, Convallaria (Wiegand 80), and Erythronium (Schaffner 98) the mother-cell does not divide. It may be of interest to note the records of other investigators in reference to some of these genera. For example, Ikeda 106 reports four megaspores in Tricyrtis hirta, and Vesque4 three in Hemerocallis, Allium, and Convallaria. Among the more primitive aquatic families, Zostera (Ro- senberg92) and Potamogeton (Wiegand,66 Holferty97) have three or four megaspores; in Typha (Schaffner49) there is no division of the mother-cell; and among the Alismaceae, Alismd (Schaffner44) and Limnocharis (Hall109) have an undividing mother-cell, while Bulomus (Ward8) has three and sometimes four megaspores. Among the Gramineae the complete tetrad is common (Fischer6), but Guignard12 reports only two megaspores in Cornucopiae. Among the Araceae Mottier 27 reports two megaspores in Arisaema, and Campbell75 the same number in Dicffenbacliia, while in the allied Lemna (Caldwell 62) the mother-cell does not divide. Among the Pontederiaceae (Smith53) there are four mega- spores, while Guignard 12 reports only two in Commclina. Among the higher families, Narcissus (Guignard 12) has an undividing mother-cell; Iris (Guignard12) has three mega- spores (Vesque4 reports four), and Tritonia and Sisyrinchium (Strasburger5) four: the Scifauiineae have three or four, excepting Costus (Humphrey40), in which the mother-cell does not divide; while the orchids Gymnadenia (Strasburger5) and Orchis (Vesque4) have a row of three or four megaspores. That detailed study would show that many of these numbers are not constant is indicated in several instances. In Arisaema, in which two megaspores are customary, Mottier 27 found one case in which the transverse wall did not form, the elongated mother-cell appearing with a nucleus at each end ; while in Dief- 78 .MORPHOLOGY OF ANGIOSPERMS fenbachia, of the same family, Campbell 75 states that the inner one of the two cells may divide, forming a row of three mega- spores. Among the Pontederiaceae, Smith °3 found great varia- tion in the development of the megaspores. While sometimes the row of four is formed by equal successive divisions, it is more common for the mother-cell to elongate greatly, with its nucleus near the micropylar end. In this position two succes- sive and rapid divisions of the nucleus occur in any order or direction, and four usually naked cells are the result, the inner- most being much the largest and speedily obliterating the others, becoming the functioning megaspore (Fig. 29, B). In Arena fatua, Cannon 86 found that four cells are formed ; or the mother-cell mav contain four nuclei without anv cell walls, the three outermost disappearing, the innermost forming the nu- cleus of the functioning megaspore. • In Potamogeton foliosus, Wiegand 66 found that the second divisions in forming the row of four are not accompanied by walls, and Ilolferty 97 found in Potamogeton nutans that the outermost wall may not appear even when there is nuclear division. Such cases emphasize the fact that there mav often be the greatest variation in the devel- opment of megaspores, and that a number reported for a species bv a hasty observer should not be regarded as a fixed one, or even possibly the customary one. The only generalization that seems to be safe in reference to the Monocotyledons, aside from the fact of their great irregu- laris, is that more of them than of the Dicotvledons have reached the condition of an undividing mother-cell. Among the Archichlamydeae, nearly all the species investi- gated have three or four megaspores, and both of these numbers are represented in almost every family in which more than one species has been studied. Upon the whole, however, a row of three megaspores seems to be more common than one of four. For example, among the Ranunculaceae, of eleven genera stud- ied only four have been reported as having four megaspores, and in all of these cases three megaspores have also been observed. The four genera referred to are Aquilegia, in which five mega- spores were also observed, Delphinium, Ranunculus (Fig. 27), and Thalictrum, and in each of these cases different observers have given different numbers. In Culthu, which ordinarily has three megaspores, Mottier 36 occasionally found nuclear divi- THE FEMALE GAMETOPHYTE 79 sion, unaccompanied by a wall, in the outermost cell of the row of three. There is every evidence that in this family the inner cell of the first division always divides, and the other one may or may not divide, resulting in three or four megaspores. Almost the only exception to three or four megaspores, in case the mother-cell divides, rioted among Archichlamydeae is Vis- cum articulatum (Treub13), in which the four or five mother- cells divide only once, the inner cell becoming the functional megaspore. This same variation is found in at least twenty other fam- ilies of the Archichlamydeae. Probably the most variable case recorded is that of Salix glaucophylla (Chamberlain40), in which there may be three megaspores, or two, or the mother-cell may not divide. The few cases anions; Archichlamydeae in which the mother- cells are not known to divide are three genera of Piperaceae (Peperomia, Piper, Heckeria) investigated by Johnson,79' 114 but the allied Saururus ( Johnson 8~) has a row of three mega- spores; Alchemilla alpina (Murbeck 94), but this is associated with the occurrence of a large mass of archesporial tissue ; the Cactaeeae ( D' Hubert 3?') ; and at least Slum cicutae folium among the Umbelliferae. In Juglans cordiformis Karsten 115 finds great variabilitv, the mother-cell functioning directly as the megaspore or giving rise to a row of three or four megaspores, the two outer ones never functioning, the two inner ones appar- ently having an equal chance, and in many cases developing two sacs. Among the Araliaceae also, Ducamp 112 reports that the mother-cell becomes the megaspore directly or produces a row of three or four megaspores. The same is true of the Balanophora- ceae, as shown by Lotsy 82 in Ehopalocnemis, and by Chodat and Bernard 83 in Helosis; but the conditions in this family are so peculiar that the phenomenon does not seem significant. In Casuarina (Treub25) (Fig. 24) and Quercus (Conrad78), in which there is a large mass of sporogenons cells, there is no division of mother-cells to form spores. The behavior of the numerous mother-cells of Casuarina is remarkable, a certain number developing as embryo-sacs, a larger number remaining sterile and becoming very much elongated, and still others be- coming tracheid-like cells. It is apparent, therefore, that among the Archichlamydeae 80 MORPHOLOGY OP AXGIOSPERMS the mother-cell very rarely fails to divide, but that there is a strong tendency to suppress one of the divisions and form a row of three megaspores. Among the Sjmpetalae the complete tetrad appears with remarkable uniformity. This is associated with a very small nucellus, most frequently only the epidermal layer investing the tetrad row, and the suggestion is evident that there may be some causal relation between these two facts. Occasionally, however, one of the divisions is suppressed, and a row of three megaspores is the result, the only cases we have found being Vaccinium and Lycium (Vesque4), Lobelia (Marshall- Ward8), Lonicera and Nicotiana (Guignard 12), and Trapella and Sarcodes (Oliver21- 24). Among the Rubiaceae Lloyd 10- reports that while each mother-cell forms a tetrad there are usually no walls (Fig. 33), as in Arena (Cannon86) and Eich- liornia (Smith53). Among the Verbenaceae Treub 14 reports that in Avicennia officinalis the mother-cell does not divide; in Aphyllon unifiovum Miss Smith102 reports that the mother- cell does not divide, although Koch19 figures a row of four megaspores in Orobanche; in the parthenogenetie Antennaria alpina Juel 74 finds that the mother-cell does not divide, how- ever, in A. dioica, in which fertilization regularly occurs, a row of four megaspores is formed. Undoubtedly more numer- ous exceptions will be found, but the evidence seems clear that the complete row of four megaspores is almost universally pres- ent among the Sympetalae. As has been stated, the reduction in the number of chromo- somes occurs during the first mitosis in the megaspore-mother- cell, whether a row of four, or three, or two megaspores is to be formed, or the mother-cell is to function directly as a mega- spore. In Lilium, the first described form in which the mother- cell does not divide to form megaspores, the beginning of a cell-plate is clearly visible in the spindle during the first mito- sis, and at the second mitosis there is also a rudimentary cell- plate. Since the other cytological characters of these two mito- ses are identical with the first two mitoses in forms that have the row of four megaspores, it might be suggested that the rudimentary plate is a survival, indicating that the ancestors of Lilium once produced the row of four, and making Lilium in this respect a specialized rather than a primitive form. This THE FEMALE GAMETOPHYTE 81 seems reasonable, but it must be noted that the rudimentary plate occurs also at the third mitosis, and so may be a reminis- cence of a much more remote ancestry with cellular prothallia. In connection with the reduction division it is of interest to note the number of chromosomes found among Angiosperms. The following table, arranged in the Engler sequence of fami- lies, although more extensive than any hitherto published, is far from complete. The numbers in parentheses were inferred rather than actually counted: The Number of Chromosomes recorded for Certain Angiosperms Plant. CHROMOSOME NUMBERS. Zostera marina Naia8 major Tritieum vulgar? Tradeseantia Eichhornia crassipes Pontederia cordata Chlorophytum Sternbergianum . F'nikia Sieboldiana Allium fistulosum ursinum ' ' Cepa I/ilium " Martagon ii ii " camdidvm " croceum " philaddphicum " tigri 'mi in FrUilliirin imperialis " Meleagris Tulipa Gesneriana En/fli ron in in americamum Galtonia candicans Scilla non-scripta sibirica Muscari neglectum {'onvallaria majalia Trillium grandiflorutn " recwrvativm Leueojum vemum Alstroemeria " psiflacea Iris squalens Cim mi indica Cypripedium barbatum Ganieto- phyte. 6 6 8 8 12 16 8 1-2* 24 8 8 ,si 12 12 12 12 12 12 12 12 12 8 12 8 8 8 24 16? 18 6 6 12 8 8 12 3 16 Sporophyte. 12 12 (16) 16 mostly 16 (32) 16 (48) 16 (16) . 16 mostly 16 24 24 24 24 24 24 •24 24 16 24 16 or less (16) variably 8-16 4^ more than 16 (36) 12 12 (24, mostly 16 24 6 32 16 Observer. Rosenberg92 Guignard 71 Overton 28 Koernicke43 Strasburger20 Smith53 Strasburger 20 Stra^burger72 Strasburger 20 Guignard26 Schaffner55 Strasburger 20 Guignard 16 Sargant 41 Strasburger20 Schaffner* Strasburger20 Guignard 26 Schn.-Thies95 Schaffner98 Schn.-Thies95 Overton 28 Schn.-Thies95 Strasburger20 i t Wiegand80 Atkinson 64 The author-. < >verton28 SI rasburger10 Guignard 26 Strasburger72 Wiegand *° Strasburger20 Year. 1901 1899 1893 1*96 isss 1898 1898 1888 1900 1888 1891 1898 1888 1884 1896 1888 l.v-s 1897 1897 1888 1891 1901 1901 1901 1893 1901 1888 1888 1900 1899 1902 1893 isss 1891 1900 1900 18S» * One anther constantly 14. 82 MORPHOLOGY OP ANGIOSPERMS The Number of Chromosomes — Continued Plant. Orchis mascula Himantoglossum Mrcinum Oymnadenia conopsea. Listera ovata Neottia nidus-avis Nymphaea alba CeratophylUim submersum Aconitum Napellus , Htlleborus foeiidus Paeonia spectabilis Podophyllum peltatum. . . Alchemilla alpina * , Asclepias Cormiti , tuberosa Asperula Crueianella Antennaria alpina* ' ' dioica Siiphium integrifolium . . . " laciniatum CHROMOSOME NUMBERS. Gameto- phyte. 16 16 16 16 16 32 48 12 12 12 12 12 8 32 10 10 12 10 40-50 12-14 8 8 Sporophyte. (32) (32) (32) (32) (32) (64) (96) 24 (24) 24 mostly 16 (24) 16 32 20 20 24 20 40-50 counted 20 (16) 16 f Observer. Strasburger 20 Guignard 26 Guignard 59 Strasburger 72 Strasburger 108 Overton 28 Strasburger 20 Strasburger 30 Overton 28 Mottier s* Murbeck94 Strasburger 96 Prye" Lloyd105 Juel88 Merrell 77 Land 81 Year. 1888 1888 1888 1891 1891 1898 1900 1902 1893 1888 1894 1893 1895 1901 1901 1901 1902 1902 1900 1900 1900 1900 It is evident from the table that Strasburger and Guignard were pioneers in this work and that they still remain the most active contributors. It is of interest to note that when atten- tion was first directed to this subject, the number of chromo- somes reported for the sporophyte, while exceeding that of the gametophyte, was not precisely twice that number. The sub- ject is one of great difficulty, and doubtless the countings of competent investigators have often been influenced by their theories, while their followers have been content too often with confirming a reported number. Variations from the character- istic number are numerous. In the gametophyte the number of chromosomes in the antipodals is frequently irregular, with a tendency to higher numbers ; but an explanation may be found in the irregular nuclear divisions which present some of the characters of amitosis (Miss Sargant41). Variations are even more frequent in the sporophyte, but it is well known that mitoses are frequently irregular, and it is easy to imagine that a chromosome may fail to split or that an unequal distribu- * Parthenogenetic. f More than 16, probably 24, in endosperm. THE FEMALE GAMETOPHYTE 83 tion to the daughter nuclei may occur. The high numbers reported for the endosperm are doubtless to be explained by the triple fusion. In the great majority of cases the gametophyte number has been counted only in the mother-cells, and the sporophyte num- ber in the tissues of the ovule or young embrvo. Still, occa- sional counts throughout the entire life-history show that the reduced number that occurs in the division of the mother-cell is maintained up to the time of fertilization, whether the inter- val be short, as in Angiosperms, where only from three to five nuclear divisions intervene between reduction and fertiliza- tion, or long, as in the liverworts, where the gametophyte is the more permanent generation and the sporophyte is a compara- tively temporary structure. Whv the number of chromosomes should be so constant, and V 7 why a reduction in number should take place, are the most important questions in this connection. The constancy of the numbers has led many to believe that the chromosome is a permanent organ of the nucleus, just as the latter is a perma- nent organ of the cell ; but no one would assign such a reason for the constant recurrence of six stamens in a lily. There is other evidence in favor of the individuality of the chromosomes, but it does not seem to be sufficient. The physiological advan- tages are evident, for the constancy in number enables each parent to transmit an equal number of chromosomes to the off- spring, and the reduction prevents the constant geometrical increase in the number of chromosomes which would otherwise occur. Strasburger 108 says: "The morphological cause of the reduction in the number of the chromosomes and of their equal- ity in number in the sexual cells is, in my opinion, phylogenetic. I look upon these facts as indicating a return to the original generation from which, after it had attained sexual differentia- tion, offspring were developed having the double number of chromosomes. Thus the reduction by one-half of the number of chromosomes in the sexual cells is not the outcome of a gradually evolved process of reduction, but rather it is the reap- pearance of the primitive number of chromosomes as it existed in the nuclei of the generation in which sexual differentiation first, took place. . . . The reduction in the number of chromo- somes takes place, in the higher plants, in the mother-cells of 84 MORPHOLOGY OF ANGIOSPERMS the spores, and it is consequently these which must be regarded as the first term of the new generation." In case the mother-cell divides, only the innermost mega- spore functions, its growth involving the digestion and absorp- tion of the other megaspores, as well as more or less of the sur- rounding sterile tissue. Ordinarily the elongating megaspore encroaches upon the others until they become merely a cap upon it; but among the Eanunculaceae Guignard 12 found in Cle- matis and Helleborus, and Mottier36 in Delphinium, that the nucellus elongates so rapidly that the sterile megaspores are not crowded into a cap, but their disorganization leaves a nar- row cavity. The same thing occurs in Jeffersonia, as shown by Andrews,37 and doubtless among many other Archichla- mydeae. The known exceptions to the functioning of the inner- most megaspore are so few that they deserve special mention, as possibly indicating some peculiar condition. Among the Monocotyledons, Agra phis (Scilla) and Dieffcn- hachia are the only exceptions we have noted. In the former, Treub and Mellink 10 observed that the outer one of the two megaspores becomes the embryo-sac, but the inner one also de- velops a sac to the four-nucleate stage, an observation later con- firmed by Guignard 12 for other species of the genus. In Agra- phia nutans Vesque 4 observed the uppermost of a row of three megaspores functioning, but the ordinary divisions within the embryo-sac, up to four nuclei, were also observed in two or even all of the megaspores. The same observer also reports that in Yucca gloriosa all four megaspores show sac tendencies, while in Uvularia each spore in a row of two developed an embryo-sac to the four-nucleate condition. In Dieffenbachia, Campbell 7r> says that the mother-cell divides very unequally, the outer one being the larger and developing the embryo-sac. In Galtonia candicans (Liliaceae) Schniewind-Thies 95 lias ob- served an interesting transition to the condition of Lilium and similar forms. The mother-cell usually gives rise to a row of four megaspores, but occasionally only two spores appear, one of which may pass over directly into the embryo-sac. Among the Archichlamydeae. in Juglans cordiformis (Kar- sten i), the two chalazal megaspores may both develop em- bryo-sacs; the outermost megaspore of the row often functions in Stellaria Holostea (Vesque4) and in Rosa, and sometimes THE FEMALE GAMETOPHYTE 85 in the latter the two outer begin the formation of embryo-sacs; and in Eriobotrya Guignard 12 found that while ordinarily the innermost megaspore of three functions, the middle or the outer one may form the embryo-sac, and even all three may begin its formation. The same author x l also reports great irregularity in Acacia, in some species the innermost of four megaspores functioning, in others the next outer one, and in still others the middle one of a row of three. In Loranthus also, Treub 13 finds that the outermost megaspore of three persist- ently functions. Among the Aralia- ceae (Ducamp112) usually the inner- most of four megaspores functions, but occasionally one of the middle cells may become the embryo-sac. Such cases serve to emphasize the megaspore character of all the cells of the row\ Among the Sympetalae, the only well-established exception is that of Trapella, in which Oliver 21 finds that the outermost of four megaspores func- tions, and in one case the next cell, while the innermost megaspore devel- ops the remarkable haustorium (Fig. 32). In Asclepias tuberosa, although the innermost of the row of four mega- spores ordinarily functions, Frye118 has observed cases in which the outer- most megaspore functions, and others in which the two innermost develop together; while Yesque 4 reports that in Salvia pratensis the outermost of the four megaspores functions. In Crucianella (Lloyd 105) all four megaspores, which in this case are not separated by cell-walls, may begin to germinate (Fig. 33). Guignard12 also includes Pyr t C D Fig. 35. — Lift "in philadelphicutn. .1. archesporial cell which is also the mc£raspore mother-cell; B, synapsis; C, stage just before splitting ofspirem; D, longitudinal >plittintr ofspirem (best seen in threads at the left); x 466.— Negatives l>y W. J. 6. Land. Fig. 36. -LiUum phiiadelphicum. E, mitotic figure of the reduction division showing the short, thick chromosomes characteristic of this stage; F. binucleate embryo- sac : '/. four-nucleate embryo-sac : ^//'double fertilization "; in the egg the darker nucleus is the male and the lighter one the female; just beyond the egg three nuclei are t'u>iiiLr; the antipodal polar nucleus forms about one-half of the complex, while the micropylar polar nucleus and the male nucleus form the other half, the male nucleus being on the rijrht and touching both polar nuclei. E-6 x 466; 11 y 520.— Negatives by W. .!. G. Land. THE FEMALE GAMETOPIIYTE 89 great obscurity, and very little lias been added to the original suggestions concerning it. The most important departure from the ordinary history- is that shown by Peperomia pellucida, as described by Camp- bell70 and Johnson79 (Fig. 38), Gunnefa (Schnegg103), Tril- lium (Ernst116), and Tulipa as described by Guignard.89a In Peperomia the nuclei of the embryo-sac do not show any of the polarity that is so marked a feature in other forms. The first four nuclei are large, and arranged peripherally like the spores of a tetrad. Divisions continue until sixteen parietal nuclei, rather evenly distributed, are found in the sac. One of the nuclei at the micropylar end of the sac becomes somewhat larger and is surrounded by a fairly defined mass of cytoplasm with a limiting membrane, this cell functioning as the egg. Another micropylar cell is similarly organized, and from its position A CD Fig. 38. — Peperomia pellucida. A, longitudinal section of an ovule with a four-nucleate embryo-sac showing no polarity ; x 295. £, embryo-sac at time of fertilization ; tt pollen-tube ; o, oospore ; e, group of nuclei fusing to form endosperm nucleus ; p, peripheral nucleus of embryo-sac ; «, synergid ; v, vacuole : x 520. C, D, groups of nuclei fusing to form endosperm nucleus; x 520. — After Johnson.78 mav be called a svneraid. Eight of the remaining nuclei mass together, are surrounded by a common cytoplasmic investment, and after fertilization unite to form a great fusion-nucleus that functions as the primary endosperm nucleus. The remaining six nuclei remain in their parietal position and are finally cut 90 MORPHOLOGY OF ANGIOSPERMS £ ... A off by walls, showing no tendency to migrate toward the posi- tion of antipodal cells. This remarkable history is regarded by Campbell as repre- senting a primitive phase of the embryo- sac of Angiosperms ; a view from which Johnson dissents, and in a more recent pa- per 114 he shows that in the allied Piper and HecJceria the eight-nucleate stage of the embryo-sac is reached in the usual way. It is tempting Fig. 39. — Gunnera. A, embryo-sac with nine nuclei, i showing no polarity. B, later stage lowing sixteen to cnnn£ct Slicn a ^ac nuclei; s, synergid nuclei; 0, oosphere nucleus: near as that of Pepet'Onua center, a group of six nuclei fusing to form endosperm with SUCh as that of nucleus ;' near base, seven antipodal nuclei. — After , . Schnegg.103 Gnetum, and theo- retically it repre- sents what one might expect to be an earlier condition of the female gametophyte among Angiosperms : but Johnson in- fers from the testimony of Piper and HecJceria, jnst referred to, that this particular sac of Peperomia is specialized rather than primitive. In Gunnera, according to Sclmegg,103 there is no polarity in the early stages of the embryo-sac, and the nuclear divisions are not simultaneous but irregular, so that there is no definite eight-nucleate stage of the sac. Before fertilization there are " at least " eight nuclei, and very commonly one or more of the nuclei divide so that nine or ten and sometimes even sixteen nuclei are found (Fig. 39) : in which case, as in Peperomia, the primary endosperm nucleus is formed by the fusion of a considerable number of nuclei. A similar lack of polarity has been observed in Tulipa sylrestris by Guignard,890 and in Tril- lium grandiflorum by Ernst110: in the latter case at least two of the nuclei of the eight-nucleate sac have been known to di- vide, giving rise to a sac with ten nuclei. In the embryo-sac of JugJans regia Xawaschin 3S has indi- THE FEMALE GAMETOPHYTE 91 cated a lack of the usual definite organization, the male cells being described as " wandering " in the cytoplasm of the sac and fusing with one of several free nuclei which function as eggs but have not organized into an egg-apparatus. This loose- ness of organization in the cells of the embryo-sac has also been observed by Karsten115 in several species of Juglans, and he emphasizes the resemblance to Gymnosperms, believing that Angiosperms are derived from them, with such forms as Gnelum as the point of origin. What may be called minor irregularities in the structure of the female gametophyte have been described in a number of forms. The reported occurrence of only one synergid in Orni- Fig. 40.— Helosis avyanensis. A, binucleate embryo-sac with antipodal nucleus already disintegrating. B, later stage; micropylar nucleus has divided twice, giving rise to two synergids, an egg (not shown), and the micropylar polar nucleus which gives rise to the endosperm; no antipodals. (7, remains of synergids and egg; the "pseudo-endosperm" nucleus dividing; no trace of antipodals. — After Chodat and Bernard. es thogalum nutans, Santalum, Gompfirena, and Loranthus, has long been known. In Loranthus Treub 13 savs that this is due to the fact that the primary micropylar nucleus divides only once, but it is also possible that the mother-nucleus of the 92 MORPHOLOGY OF ANGIOSPERMS lOl 61 0} fc \0\ \—a u m< A '-;-.- p — e synergids may not always divide. In the same category Casu- arina, as reported by Treub,25 has long been included ; but a recent study of the genus by Frye 119 has shown that the usual three micropylar nuclei occur. Fischer 6 reports the occurrence of two eggs in Gomphrcna, which Strasburger suggests may have come from division of the nor- mal egg. In Loranthus and Casuarina Treub also states that there are no antipodals ; but Frye's 119 recent investigation of the latter form has resulted in the discovery of three antipodals, which occur either at the chalazal extremity of the expand- ed portion of the sac, or in the tubular haustorial elongation. In Helosis guayanensis (Balanopho- raceae) Chodat and Bernard 83 state that the primary antipodal nucleus (binu- cleate stage) rarely divides, but soon de- generates, which means also the absence of an antipodal polar nucleus (Fig. 40). The same phenomenon has been ob- served by Hall 109 in Limnocharis, the primary antipodal nucleus remaining undivided. Several cases have also been reported in which regularly formed po- lar nuclei approach one another but do not fuse before endosperm formation, as in Balanophora elongata (Treub32), confirmed also in B. indica by Van Tie- ghem,45 and in B. globosa by Lotsy 69 ; but in the allied Bhopalocnemis (Lot- sy s2) the polar nuclei fuse. In the or- chid Gymnadenia also, Marshall-\Yaront, Salvia, Xico- ti'ina, and Sarcodes, as well as certain members of families characterized by a striking development of the antipodal cells. 98 MORPHOLOGY OP ANGIOSPERMS In other instances the activity of the antipodal cells is shown by their great increase in size and usually multinucleate condition, and also by their more or less extensive division. Among the Monocotyledons, the Sparganiaceae, Gramineae, and Araceae are conspicuous for their strongly developed antip- odal cells. In Sparganium simplex Campbell63 describes the Fig. 45. — Sparganium simplex. Lower end of embryo-sac showing a large mass of antipodal cells. — After Campbell.63 antipodal cells as at first very small, but immediately after fertilization they enlarge to several times their original size, their nuclei dividing. Finally, a conspicuous hemispherical mass of 100 to 150 uninucleate cells is formed, at this stage the endosperm having hardly at all developed (Fig. 15). The strong development of antipodal cells among the Gramineae has long been known, Fischer 6 having reported in 1880 that each antipodal cell of Ehrarta panicca divides once, and of Alopecurus pratensis three or more times. More recently Cannon86 found in Arena fatua that the antipodal cells be- come thirty-six or more in number before fertilization, and begin to disorganize with the beginning of endosperm devel- opment. Westermaier 23 has described a growth of antipodal tissue in Zea and other grasses before fertilization, and Guignard n0 has found as many as twelve multinucleate cells in the much narrowed antipodal end of the embryo-sac of Zea. It is of interest to note in this connection that in 1882 the same investigator12 found in Cornucopiae undivided but prominent and often binucleate antipodal cells. Among the Araceae Campbell 78 states that there is a general tendency for the antipodals to develop strongly, often dividing and forming a tissue, and in Lysichiton himtschatccnse the same observer63 finds that at the time of fertilization the antipodal nuclei have THE FEMALE GAMETOPHYTE 99 increased remarkably in size, and after fertilization the cells increase rapidly and divide, forming a group of eight or more cells with remarkably large nuclei. In addition to these three monocotyledonous families, a prominent antipodal region has been found in Triglochin maritima (Hill76), in which there are three to fourteen cells ; very large but undivided antipodals have been found in Lilaea (Campbell56), Commelina ( Guig- nard12), Ornithogahun, Gladiolus, and Crocus (Mottier36), Xurcissus and Iris (Guignard 12), and Romulea (Ferraris 120) ; and Ikeda 106 reports that in Tricyrtis the antipodals fill up the " chalazal protuberance," become elongated with it, and reach their maximum length just before fertilization. Among the Archichlamydeae, the Ranuneulaceae are espe- cially characterized by the activity of the antipodal cells, shown both by their great size and multinucleate condition, and also by their divisions. We have records of twelve genera, and in all of them the antipo- dals are conspicuous. In 1879 Strasburger "' reported the an- tipodals of Myosurus as very prominent, and in 1882 Guig- nard 12 described the antipodals of Erianthis as large, those of Clematis as very large and bi- nucleate, and those of Ilepatica as forming a great group and becoming multinucleate after fertilization. In 1890 Wester- maier 23 reported large antipo- dals in Ranunculaceae, among them Nigella; and in 1895 Mottier36 investigated a num- ber of genera and described the antipodals of Delphinium tri- corne as very large, growing with the embryo-sac, and persisting till after fertilization; those of Calilia palustris as lame, pyriform, and multinu- cleate ; those of Aquilegia canadensis as growing enormously Fig. 46. — Aconitum Napellus. Longitudi- nal section of embryo-sac after fertili- zation, Rhowing the three very large antipodals : nuclei of endosperm in mi- tosis; x 45. — After Osterwalder.80 100 MORPHOLOGY OF ANGIOSPERMS before and after fertilization and becoming multinucleate; those of various species of Ranunculus, Anemonella, and Thal- ictrum dioicum as very large: and those of Hepatica as growing very much until after fertilization. Since then Overton n0 has found that the antipodals of Thalictrum purpuras- cens become remarkably large, reach- ing the center of the sac ; Misa Dunn 8y has reported that in Del- phinium exaltatum three very large antipodals persist even in the oldest seeds with no indication of degen- eration; Miss Lyon has noted as many as twenty-live antipodal cells in Hepatica ; and Osterwalder 60 has figured exceedingly large antipodals in Aeon it inn Xa pell us (Fig. 46). The whole family is characterized, therefore, bv the activity of its an- tipodal cells, exhibited more by their great increase in size than by divi- sion. Among the Amentiferae Miss Benson 31 reports a row of six or more superposed antipodals in the very narrow antipodal end of the sac in Castanea vulgaris, the lowest one being figured as the largest and multinucleate, the whole structure resembling the antipodal region of many Compositae. Around the base of this elongated antipodal region there are developed such tracheid- like cells as occur in the micellus Fig. 47. — Aster novae-angliae. Longitudinal sec- tion of embryo-sac just before fertilization: m, micropyle: .«. synersrid ; o, oosphere; e, en- dosperm nucleus : t, jacket: A, lower antipodal cell: tour other antipodal cells shown, the upper with four nuclei and the others with two; x 407. — After Chamberlain.35 THE FEMALE GAMETOPHYTE 101 of Casuarina, but in this latter instance thev are derived from mother-cells. Other Arehichlam- ydeae with active antipodals are Heckeria (Johnson 114 ), in which they are sometimes six to eight in number; Asarum (Hofmeister 2), in which thev are very lono-, ex- tending at fertilization from one- ihird to one-half the length of the embryo-sac, and sometimes divi- ding; Jefferson ia diphylla (An- drews37), in which they become about one-half as long as the embryo-sac ; Eriobotrya ( Guig- nard 12), in which they are large; and Anoda (Guignard 12), in which they are prominent and often binucleate. Among the Sympetalae the Compositae are especially note- worthy for the extensive develop- ment of the antipodal region (Fig. 47). In this familv the chalazal end of the elongated sac is very narrow and the antipodals are superposed. In a number of eases, as in Doronicum, Petasites, and Taraxacum, there are usually only three antipodals, but they remain active ; while in Tussila- c]o (Guignard 12) there are usual- ly four: in Senecio (Alottier29) two to six; in Silphium (Aler- rell77) three to eight; in Conyza 'Guignard12) eight to ten; in Aster novae-angliae (Chamber- lain35) three to thirteen; and in Antennaria duel57) they con- tinue to divide until quite a tis- Fio. 48. — A, Sherardia arvensie. Em- * bryo-sac before fertilization : low- er antipodal acting as an hausto- rium. B, CaUipdtit cucullaria, showing lower antipodal still act- ive after embryo and endosperm are considerably advanced. — After Lloyd.106 102 MORPHOLOGY OF ANGIOSPERMS sue is formed (Fig. 41). This record indicates that the divisions are variable in number even in the same species, and it may be noted in this connection that while Schwere 42 states that there are only three antipodals in Taraxacum, Hegelmaier 9 had previously reported four or five, and more than three have been observed frequently in this laboratory. In many of these cases all the cells usually contain two or more nuclei, and the end cell toward the chalaza often becomes vesicular and multi- nucleate, breaking through the sac and encroaching upon the chalazal tissue. It seems to be clear that in the Compositae this development of antipodals is practically an aggressive haus- torium for the embrvo-sac ; while in the Ranunculaceae the antipodals doubtless serve as an haustorium, but do not invade the neighboring tissue. Certain Rubiaceae also contain active antipodals, since Lloyd G" has found that in Vaillantia hispida while two of the antipodals are insignificant, the third is very prominent and remains active for a long time. The same au- thor 105 has more recently found the same to be true of the Galieae (Fig. 48), and he also has found four to ten antip- odals in Diodia virginiana. Balicka-Iwanowska 68 has also noted enlarging and persistent antipodals in Plantaginaceae and Campanulaceae, and their division in Dipsaceae as in the Com- positae. In Asclepias, although three active antipodals are usual, Frye 11S has observed compact antipodal tissue consisting of seven or eight cells; and in A. Cornuti he has noted the occurrence of tracheid-like cells at the base of the embrvo-sac, such as occur in Casuarina and Castanea. There seems to be no reason to question the ordinary view that the antipodal cells are vegetative cells of the gametophyte. Their polarity as contrasted with that of the egg-apparatus, and their behavior when they function confirm it. The occa- sion for their activity seems to be to supply the embryo-sac with nutritive material absorbed from without at a time when the endosperm has not been organized or other means of obtain- ing nutrition are not available. In Monotropa uniflora Shi- bata l— has found that the three small antipodals disintegrate after fertilization, but that when fertilization is prevented they may enlarge enormously and fill a considerable portion of the sac. The character of the active antipodals among the more primitive Monocotyledons and in the Eanunculaceae may be THE FEMALE GAMETOPHYTE 103 regarded as indicating a primitive condition of the nutritive tissue in the female gametophytes of Angiosperms ; but the antipodals of many of the Compositae are organized into an aggressive haustorium which can only be regarded as a very specialized organ. The enlargement of the embrvo-sac and the nature of its development, both before and after fertilization, are extremely varied. The enlargement is directly related to the digestion of the contiguous tissue. In a few cases this destruction is not extensive, and more or less of the nucellar tissue is permanent (perisperm) and is used for the storage of reserve food, as in the Scitamineae, Piperaceae, Chenopodiaceae, Phytolaccaceae, Caryophyllaceae, Xymphaeaceae, etc. In most cases, however, the destruction of the nucellar tissue is complete to the integu- ment, and even that is sometimes involved, as in Allium odo- rum, certain orchids, and Astilbe (Webb111). Frequently the tissue at the apex of the nucellus remains as a cap on the em- bryo-sac, as in Arisaema (Mottier27) and other Araceae, Lemna (Caldwell02), Liliaceae, Silphium (Merrell77), and many other forms, and this is frequently accompanied by more or less elongation and even division of the capping cells. Frequently a definite nutritive jacket invests the embryo- sac, consisting of one or more layers of cells with deeply stain- ing contents CFigs. 47, 50). For the most part this is a single layer derived from the integument, but in Armeria it is derived from the nucellus, and in Erodium one layer is derived from the nucellus and the other from the integument. This jacket has been called a tapetum, and such it is in function. In using the term, however, there is danger of confusing it with the tapetnm of ordinary sporogenous tissue. This jacket has been definitely observed as conspicuous in Helosis (Chodat and Bernard83), Sium, many Scrophulariaceae (Balicka-Iwanow- ska6S), Campanula (Barnes18), Stylidaceae (Burns85), and certain Compositae, and by Billings 10° in numerous sympeta- lous forms, among the most conspicuous being Lobelia, Primu- laceae (except Leptosiphon) , Linum, Forsythia, ximsonia, Menyanthes, Polemoniaceae, Myoporum, Globularia, Scaevola, Calendula, etc. In many cases the micropylar end of the sac destroys all of the nucellar tissue capping it, and protrudes more or less 104 MORPHOLOGY OF ANGIOSPERMS into the micropyle, as in Hemerocallis, Crocus, Gladiolus, Romulea (Ferraris120), Alchemilla (Murbeck ,J4), in which the sac pushes through to the tegumentary tissue closing the micro- pyle, Medicago, Torenia asiatica (Strasburger 5), Labiatae, Vaillantia (Lloyd67), Diodia and the Galieae (Lloyd105), and many other forms. In Vaillantia the mother-cell migrates into the micropyle and develops there. While ordinarily the embrvo-sac is relatively broad and rounded at its micropylar extremity, this is by no means so commonly true of the antipodal end. If the antipodals are ephem- eral, the growth of the antipodal region is frequently checked after the first division of the megaspore nucleus, and through the growth of the rest of the sac it becomes a very small pocket, as in Typha, Potamogeton, Sagittaria, certain Gramineae, Pontederia, Lilium, Oenothera, etc. (Fig. 79). It is generally true that the antipodal region of the sac is narrower than the micropylar, but its growth is not often checked so completely and so early as in the eases cited. In other cases, the antipodal region of the sac grows very active- ly, elongating toward the chalazal region and penetrating it more or less deeply, resulting in a very nar- row and elongated sac. Such an antipodal region must be regarded as an haustorium that digests and absorbs its way into the chalazal tis- sue. Illustrations of this are very numerous, as in Gramineae, Tricyrtis ( Ikeda 10c), Scitamineae, Saururaceae, Loranthaceae, Polygalaceae, Lythraceae, Aceraceae, and most Sympetalae. In penetrating the chalaza the antipodal tip usually remains narrow, but in Saururus (Johnson87), Scitamineae (Hum- phrey40), Cuphea (Guignard 12), Campanula (Barnes18), etc., Fig. 49. — Saururus cernuus. Longi- tudinal section of embryo-sac; after the first division of the en- dosperm nucleus the micropylar cell has given rise to endosperm tissue, while the other cell has become a large vesicular hausto- rium.— After Johnson.* THE FEMALE GAMETOPHYTE 105 it has been observed to enlarge more or less abruptly, forming a bulbous chalazal haustorium. In Carina indica this becomes much larger than the rest of the embryo-sac; and in Saururus cernuus Johnson"7 describes the embryo-sac as elongating rap- idly, broadening below, the upper part remaining narrow, the completed sac resembling a long-necked flask (Fig. 49). In addition to the various forms of haustorial apparatus described above as developed in connection with the embryo-sac, certain extreme cases deserve special mention. It has long been known that among the Santalaceae (Santalum, Thesium, Usyris, etc.) the embryo-sac develops a micropylar tube that passes through the micropyle and enters the cavity of the ovary, and that in some of them {Thesium, etc.) there is also an antip- odal tube (see Guignard 17). These remarkable tubular or vermiform haustoria obtain nutritive material beyond the ovule. Later, Johnson 22 described in detail the haustorial apparatus of Myzodendron, another genus of Santalaceae. The young sac is broad above and narrowed toward the antipodal end. After fertilization the antipodal region develops rapidly, penetrates the chalaza, enters the placental axis, and curving passes down it to the base of the flower, where its tip dilates and becomes embedded in the " vascular cup " formed by the three diverging carpellary bundles. Rigidity is given to this remarkably elon- gated tube by numerous cross-walls, but these are lacking in the placental region. Among the Amentiferae (Miss Benson31) vermiform caeca are often sent out from the embryo-sac. In Fagus sylvatica this tubular outgrowth penetrates to the base of the nucellus, the primary endosperm nucleus passing into it, but not the antipodals, which are anchored by thick walls. In Castanea vulgaris the caecum develops from the side of the sac just above the narrow antipodal prolongation, is entered by the endosperm nucleus, and passes down between the nucellus and the integu- ment. In Cur [>in us Betulus the chalazal region is sometimes riddled by the long caeca from the several embryo-sacs; and in Corylus Avellana a short caecum appears after fertilization. In Casuarina, as shown by Frye,139 a conspicuous vermi- form caecum is developed much as among the Amentiferae. From the antipodal extremity of the sac a long tube penetrates the chalazal region, into which the endosperm nucleus passes 106 MORPHOLOGY OF ANGIOSPERMS and sometimes the antipodals. This haustorial tube was ob- served to begin its development at different stages in the history of the sac, sometimes being evident in the two-nucleate stage of the sac, sometimes not having begun in the seven or eight- nucleate stage. One of the strangest cases is that of Trapella, as described by Oliver.21 In this the innermost megaspore of a row of four becomes extremely elongated, penetrates the chalaza, and divides longitudinally, the two cells being very active, as indi- cated by their contents and numerous starch grains. In this form the synergids enlarge and persist on the apex of the sac (Fig. 32)/ Among the Scrophulariaceae, such as Pedicularis, Rliinan- thus and its allies, etc., Tulasne, Hofmeister, Tschirch, Schlot- terbeck, and others have described the numerous vermiform tubes that develop from the embryo-sac and " ruminate " the integument and destroy its tissue, although they did not recog- nize their origin ; and similar tubes have been found in certain Labiatae. Recently Balicka-Iwanowska 68 has investigated the embryo-sacs of many Scrophulariaceae, as well as other allied Sympetalae, and has discovered a remarkably constant occur- rence of haustorial outgrowths from the sac at both micropylar and chalazal ends, filled in later by endosperm cells. The common case is for the broad micropylar end of the sac to de- velop four prongs, and for the narrower chalazal end to fork, as seen not merely among Scrophulariaceae, but also among Utriculariaceae, Pedaliaeeae, and Plantaginaceae. The devel- opment of these haustoria is related to the thickness of the integument, which in these groups seems to be a source of nutri- tive supply. There are all stages in the development of the haustoria, but the general tendency in this region of the Sympet- alae is very marked. A striking case is that of the well-known Torenia asiatica, mentioned above, in which the sac does not develop outgrowths, but protrudes bodily beyond the micropyle, touching the funiculus, and even reaching the ovary wall. All of these haustorial outgrowths are supplied with active endo- sperm cells or nuclei. It is stated that all species of Campanulaceae (Balicka- Iwanowska 68), Lobeliaceae (Billings100), and Stylidaceae (Burns 85) develop both micropylar and chalazal haustoria, and B Fig. 50. — A, Gldbularia cordifolia, the mieropylar end of the embryo-sac has grown out into an extensive haustorium furnished with nuclei from the endosperm ; ,/', funiculus; after Billixos.'oo B, Plantago lanceolata, longitudinal section of ovule after embryo is somewhat advanced, showing extensive haustorial system: after Balicka-Iwanowska.68 C\ StyliJium squameUosum, embryo-sac after second division of endosperm nucleus: e, egg', /'. pollen-tube; after Burns.86 /', BybUt gigantea. longitudinal section of seed with branching haustoria in both mieropylar and antipodal regions; h, haustorium; g, embryo; e, endosperm; after Lano.m 107 108 MORPHOLOGY OF ANGIOSPERMS that often finger-like processes are put out at the side or base of the sac, extending toward the vascular bundles; and in Sty- lidaceae, immediately after the entrance of the pollen-tube, the micropylar part of the embryo-sac grows out into an enormous haustorium much larger than the rest of the sac (Fig. 50). As a result of his investigations of Polypompholyx and Byblis, Lang 91 not only discovered conspicuous haustoria, but used this character, along with others, such as the nucellus with a single row of axial cells, the tapetum de- rived from the single integument, and the united petals, to remove these genera from the archiehlamy- deous Droseraceae to the sympetalous Lentibulariaceae. The whole subject of the mecha- nism for the nutrition of the embryo- sac deserves more detailed attention than it has received. In his study of the fleshy plants, D'Hubert,33 on the basis of the appearance and disap- pearance of starch, concludes that the antipodals nourish the sac before fer- tilization, the synergids nourish the nuclei of the pollen-tube and then the nucleus of the egg at the time of fertilization, and the polar nuclei nourish the fertilized ec:2; and aive rise to the endosperm (Fig. 51). Such details may prove true for the ( 'actaceae and other fleshy plant-/' but the larger field is to be traversed first, which embraces all of the mor- phological structures used in obtaining nutritive supplies for the structures within the embryo-sac, both before and after fer- tilization. Just what mechanism supplies what structure is a subordinate detail and very difficult to prove, besides being an exceedingly improbable division of labor among structures so * D'Hubert concludes that starch is characteristic -?4. 1898. 55. Schaffner. J. H. Karyokinesis in the Root Tips of Allium Cepa. Bot. Gazette 26: 225-238. pis. 21-22. 1898. THE FEMALE GAMETOPHYTE 117 56. Campbell, D. H. The Development of the Flower and Embryo in Lilaea subulata HBK. Annals of Botany 12: 1-28. pis. 1-3. 1898. 57. Juel, H. O. Parthenogenesis bei Antennaria alpina (L.) R. Br. Bot. Centralbl. 74: 369-372. 1898. 58. Nemec, B. Ueber den Pollen der petaloiden Antheren von Hya- cinthus orientalis. Bull. Int. Acad. Sci. Boheme. 1898. 59. Guignard, L. Les centres cinetiques chez les vegetaux. Ann. Sci. Nat. Bot. VIII. 6: 177-220. pis. 8-10. 1898. 60. Osterwalder, A. Beitrage zur Embryologie von Aconitum Napellus L. Flora 85: 254-292. pis. 11-15. 1898. 61. Goldflus, Mlle. M. Sur la structure et les fonctions de l'assise epitheliale et des antipodes chez les Composees. Jour. Botani- que 12: 374-384. pis. 1-6. 1898; 13: 87-96. 1899. 62. Caldwell, O. W. On the Life-History of Lemna minor. Bot. Gazette 27: 37-66. figs. 59. 1899. 63. Campbell, D. H. Notes on the Structure of the Embryo-sac in Sparganium and Lysichiton. Bot. Gazette 27: 153-166. pi. 1. 1899. 64. Atkinson, G. F. Studies on Reduction in Plants. Bot. Gazette 28 : 1-26. pis. 1-6. 1899. 65. Fullmer, E. L. The Development of the Microsporangia and Microspores of Hemerocallis fulva. Bot. Gazette 28: 81-88. pis. 7-8. 1899. 66. Wiegand, K. M. The Development of the Microsporangium and Microspores in Convallaria and Potamogeton. Bot. Gazette 28: 328-359. pis. 24-25. 1899. 67. Lloyd, F. E. The Comparative Embryology of the Rubiaceae. Bull. Torr. Bot. Club 28: 1-25. pis. 1-3. 1899. 68. Balicka-Iwanowska, G. P. Contribution a Tetude du sac em- bryonnaire chez certaines Gamopetales. Flora 86: 47-71. pis. 3-10. 1899. 69. Lotsy, J. P. Balanophora globosa Jungh. Eine wenigstens ortlich-verwittwete Pflanze. Ann. Jard. Bot. Buitenzorg II. 1 : 174-186. pis. 26-20. 1899. 70. Campbell, D. H. Die Entwicklung des Embryosackes von Pepe- romia pellncida Kunth. Ber. Deutsch. Bot. Gesell. 17: 452-456. pi. 31. 1899: also, A Peculiar Embryo-sac in Peperomia pellu- cida. Annals of Botany 13 : 626. 1899. 71. Guignard. L. Le developpement du pollen et la reduction dans le Naias major. Arch. Anat. Micr. 2 : 455-509. 1899. 72. Strasburger, E. Ueber Reduktionstheilung, Spindelbildung, und Cilienbihlner irn Pflanzenreich. Hist. Beitr. VI. Jena, 1900. 73. Goebel, C. Organography of Plants. Translated by I. B. Balfour. Oxford. 1900. IIS MORPHOLOGY OF ANGIOSPERMS 74. Juel, H. O. Vergleichende Untersuchungen iiber typische und parthenogenetische Fortpflanzung bei der Gattung Antennaria. Handl. Svensk. Vetensk. Akad. 33: no. 5. pp. 59. 2)ls. 6. Jigs. 5. 1900 ; review in Bot. Zeit. 59 : 131. 1901. 75. Campbell, D. H. Studies on Araceae. Annals of Botany 14: 1-25. pis. 1-3. 1900. 76. Hill, T. G. The Structure and Development of Triglochin mari- timum L. Annals of Botany 14: 83-107. pis. 6-7. 1900. 77. Merrell, W. D. A Contribution to the Life-History of Silphium. Bot. Gazette 29: 99-133. pis. 3-10. 1900. 78. Conrad, A. H. A Contribution to the Life-History of Quercus. Bot. Gazette 29 : 408-418. pis. 28-29. 1900. 79. Johnson, D. S. On the Endosperm and Embryo of Peperomia pellucida. Bot. Gazette 30: 1-11. pi. 1. 1900. 80. Wiegand, K. M. The Development of the Embryo-sac in some Monocotyledonous Plants. Bot. Gazette 30: 25-47. pis. 6-7. 1900. 81. Land, W. J. G. Double Fertilization in Compositae. Bot. Gazette 30: 252-260. pis. 15-16. 1900. 82. Lotsy, J. P. Rhopalocnemis phalloides Jungh., a Morphological- systematical Study. Ann. Jard. Bot. Buitenzorg II. 2: 73-101. pis. 3-U. 1900. 83. Chodat, R., and Bernard, C. Sur le sac embryonnaire de VHe- losis guayanensis. Jour. Botanique 14: 72-79. pis. 1-2. 1900. 84. Bernard, C. H. Recherches sur les spheres attractives chez Lilium candidum, etc. Jour. Botanique 14= 118-124, 177-188, 206-212. pis. 4-5. 1900. 85. Burns, G. P. Beitrage zur Kenntniss der Stylidiaceen. Flora 87: 313-354. pis. 13-lk. 1900. 86. Cannon, W. A. A Morphological Study of the Flower and Em- bryo of the Wild Oat, Avena fatua. Proc. Calif. Acad. Sci. III. 1 : 329-364. pis. 49-53. 1900. 87. Johnson, D. S. On the Development of Saururus cernuus L. Bull. Torr. Bot. Club. 27: 365-372. pi. 23. 1900. 88. Juel, H. O. Beitrage zur Kenntniss der Tetradenbildung. Jahrb. Wiss. Bot. 35 : 626-659. pis. 15-16. 1900. 89. Dunn, Louise B. Morphology of the Development of the Ovule in Delphinium exaltatum. Proc. Amer. Assn. Adv. Sci. 49 : 284. 1900. 89a. Guignard, L. L'appareil sexuel et la double fecondation dans les Tulipes. Ann. Sci. Nat. Bot. VII. 11 : 365-387. pis. 9-11. 1900. 90. . La double fecondation dans le mais. Jour. Botanique 15: 37-50. 1901. 91. Lang, F. X. Untersuchungen iiber Morphologie, Anatomie, und Samenentwicklung von Polypompholyx und Byblis gigantea. Flora 88: 149-206. pi. 12. figs. 80. 1901. THE FEMALE GAMETOPHYTE 119 92. Rosenberg, O. Ueber die Embryologie von Zostera marina. Bih. Handl. Svensk. Vetensk. Akad. 273: no. 6. pp. 26. pis. 2. 1901. 93. . Ueber die Pollenbildung von Zostera. Meddel. Stock- holms Hogsk. Bot. Inst. pp. 21. 1901. 94. Murbeck, S. Parthenogenetische Embryobildung in der Gattung Alchemilla. Lunds Univ. Arsskrift 362 : no. 7, pp. 46. pZs. 6. 1901. 95. Schniewind-Trtes. J. Die Reduktion der Chromosomenzahl und die ihr folgenden Kerntheilungen in den Embryosackmutter- zellen der Angiosperinen. Jena, 1901. 96. Strasburger, E. Einige Bemerkungen zu der Pollenbildung bei Asclepias. Ber. Deutsch. Bot. Gesell. 19 : 450-461. pi. 24. 1901. 97. Holferty, G. M. Ovule and Embryo of Potamogeton natans. Bot. Gazette 31: 339-31(3. pis. 2-3. 1901. 98. Schaffxer, J. H. A Contribution to the Life-History and Cytology of Erythronium. Bot. Gazette 31 : 369-387. pZs. 4-9. 1901. 99. Frye. T. C. Development of the Pollen in some Asclepiadaceae. Bot. Gazette 32 : 325-331. pi. 13. 1901. 100. Billings, F. H. Beitrage zur Kenntniss der Samenentwicklung. Flora 88: 253-318. 1901. 101. Atkinson, G. F. On the Homologies and Probable Origin of the Embryo-sac. Science 13: 530-538. 1901. 102. Smith. Amelia C. The Structure and Parasitism of Aphyllon uniflorum Gray. Contrib. Bot. Lab. Univ. Penn. 2: 111-121. pis. 13-15. 1901. 103. Schnegg. H. Beitrage zur Kenntniss der Gattung Gunnera. Flora 90: 161-208. figs. 28. 1902. 104. Oliver, F. W. On a Vascular Sporangium from the Stephanian of Grand 'Croix. New Phytologist 1 : 60-67. pi. 1. 1902. 105. Lloyd, F. E. The Comparative Embryology of the Rubiaceae. Mem. Torr. Bot. Club 8 : 27-112. p/s. 8-15. 1902. 106. Ikeda, T. Studies in the Physiological Functions of Antipodals and Related Phenomena of Fertilization in Liliaceae. 1. Tri- cyrtis hirta. Bull. Coll. Agric. Imp. Univ. Tokyo 5: 41-72. pis. 3-6. 1902. 107. Guignard, L. La double fecondation chez les Solanees. Jour. Botanique 16: 145-167. figs. 45. 1902. 108. Strasburger, E. Ein Beitrag zur Kenntniss von Ceratojyhyllum submersnm und phylogenetische Erorterungen. Jahrb. Wiss. Bot. 37: 477-526. ph. 9-11. 1902. 109. Hall. J. G. An Embryological Study of Limnocharis emargi- nata. Bot. Gazette 33: 214-219. pi 9. 1902. 110. Overton. J. B. Parthenogenesis in Thalictrum purpurascens. Bot. Gazette 33 : 363-375. pis. 12-13. 1902. 9 120 MORPHOLOGY OF AXGIOSPERMS 111. Webb. J. E. A Morphological Study of the Flower and Embryo of Spiraea. Bot. Gazette 33: 451-460. figs. 28. 1902. For cor- rection of names see Rehder in Bot. Gazette 34: 246. 1902. 112. Due amp, L. Recherches sur l'embryogenie des Araliacees. Ann. Sci. Xat. Bot. VIII. 15 : 311-402. pis. 6-13. 1902. 113. MrRBECK, S. Ueber Anomalien im Baue des Nucellus und des Embryosackes bei parthenogenetischen Arten der Gattung Al- chemilla. Lunds Univ. Arsskrift 38- : no. 2. pp. 10. pi. 1. 1902. 114. Johnson, D. S. On the Development of Certain Piperaceae. Bot. Gazette 34: 321-340. pis. 9-10. 1902. 115. Karsten, G. Ueber die Entwicklung der weiblichen Bluthen bei einigen Juglandaceen. Flora 90: 316-333. pi. 12. 1902. 116. Ernst, A. Chromosomenreduction, Entwicklung des Embryo- sackes und Befruchtung bei Paris quadrifolia L. und Trillium grandiflorum Salisb. Flora 91 : 1-46. pis. 1-6. 1902. 117. Endriss, W. Monographic von Pilostyles ingae (Karst.) (Pilo- styles Uiei Solms-Laub.). Flora 91 : 209-236. pi 20. figs. 29. 1902. 118. Frye, T. C. A Morphological Study of Certain Asclepiadaceae. Bot. Gazette 34: 389-413. pis. 13-15. 1902. 119. . The Embryo-sac of Casuarina stricta. To be published in Bot. Gazette 35 : 1903. 120. Ferraris, T. Ricerche embriologiche sulle Iridaceae. I. Em- briologia del G. Romulea Maratti. Ann. R. Istit. Bot. Roma 9: 221-241. pis. 6-7. 1902. 121. Juel, H. O. Zur Entwicklungsgeschichte des Samens von Cyno- morium. Beih. Bot. Centralbl. 13 : 194-202. figs. 5. 1902. 122. Shibata, K. Experimentelle Studien iiber die Entwickelung des Endosperms bei Monotropa. (Vorlaufige Mitteilung.) Biol. Centralbl. 22: 705-714. 1902. CHAPTER VI THE MALE GAMETOPHYTE The reduced number of chromosomes appears at the first mitosis in the pollen mother-cell, which is therefore the first gametophytic cell (Fig. 53). In every case, so far as known, two divisions occur in rapid succession, giving rise to four microspores. Strasburger 8 has called attention to the two modes of division. In one case, most frequent among ]\Iono- cotyledons, a wall follows the first nuclear division, dividing the mother-cell into two hemispherical cells ; the second nuclear division is also followed immediately by the formation of a wall, making two equal cells from each of the hemispheres (Fig. 54). In the other case, more characteristic of the Dicotyledons, the two nuclear divisions occur before any walls are formed, all the walls beine; then formed simultaneouslv and in such a way that each of the four cells has the form of a triangular pyramid with a spherical base — that is, each cell is the quadrant of a sphere (Figs. 55, 56). The former method has been called successive, the latter simultaneous division. The two modes are not sharply characteristic of the two great groups of Angio- sperms, but the successive method is dominant among Mono- cotyledons and the simultaneous among Dicotyledons. In any event the result is a tetrad, a group of four cells each of which i- a microspore. In successive division there is a bilateral ar- rangement of the microspores, and in simultaneous division the arrangement is tetrdhedral ; but both arrangements sometimes occur in the same sporangium. The arrangement of the tetrad is not always restricted to these two methods (Fig. 57). Wille15 has described varying arrangements of microspores in the tetrads of species of Juncus and Orchis mascula; and in Typha Schaffner 34 not only found 121 122 MORPHOLOGY OF ANGIOSPERMS the tetrads indiscriminately tetraliedral or bilateral, but fre- quently the four spores are in a row. A tetrad consisting of four spores in a row has also been found by Strasburger 5i and 5 C75?1.0n 5 (,7 8 Fig. 53. — Lilium Martagon. A. transverse section of young mierosporangium, showing two nuclear mitotic figures in sporogenous cells and one in a hypodermal cell ; such figures show 24 chromosomes, the sporophyte number; x 200. B. chromosomes of a mitotic figure in the wall of a mierosporangium, showing 24 chromosomes; x 6, mitotic figure of the first division showing the short, thick chromosomes characteristic of the reduction division; E, later stage of first division, showing vertical view of the 12 chromo- somes; F. side view of same stage showing 12 chromosomes passing to the upper pole, only 10 for the lower pole being in sight; G, formation of wall between daughter nuclei ; II, second division ; /, formation of walls. — After Stkasburger.™ like that of Lilium are not rare, where the mother-cell gives rise directly to a single megaspore. As stated, in 1886 Wille 15 found no tetrad in Asclepias syriaca; and in 1892 Chau- veaud 20 observed the reduction division of the pollen mother- 124 MORPHOLOGY OF ANGIOSPERMS cells of Cynanchum, but seems not to have noted the formation of a tetrad; but the tetrad, consisting of a row of four micro- spores, and referred to above as discovered by Strasburger and by Frye in 1901 in a number of species of Asclepias and in Cynanchum, was so unusual as to disguise its tetrad nature, and Fig. 55.— Podophyllum peltatum. Mitosis in pollen mother-cell. A, telophase of first division ; 2?, late anaphase of second division ; G, telophase of second division ; the nuclei of the four microspores are formed, but the cell walls, as is characteristic of simultaneous division, have not yet appeared. — After Mother.26 besides, the enlargement and consequent readjustment of the spores soon break up the row (Fig. 58). The first record of the occurrence of a tetrad in Asclepias seems to have been made bv Stevens 41 in 1898 ; and the fourth independent discovery of it was by Gager 5S in 1902. Elving,7 Wille,15 and Stras- burger12 showed that in various species of the Cyperaceae a tetrad is formed although only one microspore becomes func- tional, the other soon disorganizing. Juel 50 has recently made a thorough study of Carex acuta (Fig. 59). He finds that the two characteristic nuclear divisions take place, and that a cell-plate is formed at each division. The cell-plates are soon resorbed, however, so that the four nuclei lie free within the wall of the mother-cell. Three of the nuclei then disintegrate, while the fourth becomes the nucleus of the single functional microspore; and the wall of the mother-cell, inclosing the four nuclei, becomes the wall of the microspore. In Zostera marina Rosenberg :'~ has described the tetrad division of the remarkably elongated mother-cell (Fig. 11). The divisions are longitudinal and in parallel planes, resulting in four remarkable filiform THE MALE GAMETOPHYTE 125 microspores lying side by side, and measuring 3 by 2,000 p> when mature. That this is a tetrad is evident from the rapid succession of the divisions, the reduction of chromosomes, and the formation of four spores from a mother-cell. In some cases a mother-cell may give rise to less than four microspores, or may produce more than the normal number (Fig. 60). In 1886 Wille 15 summarized the work of previous investigators, notably of Hofmeister, Tangl, Wimmel, and TschistiakofT, and added the results of his own investigations. The following lists are made up largely of forms investi- gated by Wille himself: Two microspores from a mother-cell are occasion- ally found in Conrallaria m ultiflo ra, A spa rag us offi- cinalis, Aconitum Napellus, Euphorbia Lathyrus, Be- gonia sp., Saxifraga caespi- tosa, Azalea indica, and Syringa vulgaris. Three microspores were found in Begonia sp., Saxi- fraga caespitosa, Azalea in- dica, and Lonicera coerulea. Five microspores were found in Funhia ovata, Fi- caria ranunculoides, Stel- laria glauca, Scleranthus annuus, Prunus Cerasus, Rumex Patientia, Azalea indica, Lonicera coerulea, Si/ringa persica, and Sym- Fig. 56. — Scrophula/ria nodosa. Section of mi- crosporangium showing appearance of spores formed by the simultaneous method ; the inner tapetum of mierosporantfium consists of greatly elongated cells which are very glandular in appearance, x 275. phytum officinale. Six microspores were found in Hemerocallis fulva, Ficaria ranunculoides, Elatine hexandra, Cornus sanguinea, Lonicera coerulea, and Fuchsia sp. Seven microspores were counted with certainty in Fuchsia sp. and fourteen are reported rather doubtfully: eight i< given for Azalea indica, and eight to twelve for Lonicera coerulea, 126 MORPHOLOGY OF AXGIOSPERMS but it was not absolutely certain that in case of the higher num- bers all the microspores came from the same mother-cell. In Hemerocallis fulva Strasburger 10 has counted nine microspores from a single mother-cell ; and later Juel 33 and Fullmer 44 re- ported six to eight in the same species. More recently Miss Lyon 40 has found five or six microspores of equal size produced by a single mother-cell of Euphorbia corollata. According to Wille, two microspores result from a failure of the mother-cell to undergo the second division. When three are formed, the first division is unequal, and only the larger cell divides. Five or more microspores are formed by subsequent division of one or more members of the tetrad. Strasburger,10 Juel,33 and Fullmer,44 in their studv of Hemerocallis fulva found an explana- tion of the irregular numbers. Strasburger found that chromosomes which fail to pass to either pole at the first B C D mitosis give rise to small microspores. Juel in his more recent study con- firms Strasburger, and finds that even single Fig. 57. — Variation in the arrangement of the spores of a tetrad. A-C, Orchis mascula, x 380; after Wille.™ D-E, Typha latifolia, x 400; after ScHAFFNER.34 chromosomes which be- come separated may divide and give rise to nuclei and organize cells. Fullmer attributes the supernumerary microspores to the division of one or more members of the tetrad. Perhaps no phase of plant cytology has received so much attention as the nuclear divisions in the pollen mother-cell. It is an interesting fact that the cytological characters of these two mitoses agree minutely with those in the megaspore mother- cell. The pollen mother-cell can be positively identified by the appearance of the synapsis stage (Fig. 54, 5), even before any rounding off or separation takes place. While yet in the spirem Fig. 58. —Development of male gametophyte in Asclepias. A-J>, A. Cornuti; C-E, A. tuberosa. A, section of young mierosporangium showing archesporial cells: £, portion of the single layer of elongated mother-cells; C, later 6tage showing two mother-cells, the lower one dividing and showing 10 chromosomes, the gametophyte number; Z>, second division of mother-cell, by which the row of four microspores is formed: E, microspore showing tube nucleus (t) and generative nucleu6 (g). A, x 200 ; J3-E, x 800.— After Fbye." 127 12S MORPHOLOGY OP AXGIOSPERMS stage the chromatin thread splits longitudinally throughout its entire length (Fig. 61, A, B). The double thread then seg- ments transversely into the number of chromosomes characteris- D A Fig. 59. — Development of microspores in Heleocharis palvstris and Cars.r acuta. A-B, Heleocharis. showing the single functional microspore and three disorganizing micro- spores. x.380; after Strasblrgeb.14 G-I, Carex: ' '. mother-cell; Z>, second division; F, four nuclei, only three of winch are shown within the mother-cell * E and F should be reversed i ; £, later stage than F; the nucleus of the functional micro- spores is preparing for division; % Fig. 62. — Oypripedium apectabile. Section of microsporangium, showing microspores in various stages of division into tube and generative nuclei; although the divisions are nearly simultaneous throughout the microsporangium, it will he seen that in some cases the nuclei are in the spirein stage, while in others the tube ;ind genera- tive nuclei are easily distinguished ; x 300. not the morphological equivalent of the vegetative or prothallial cells of the Gymnosperms and heterosporous Pteridophytes. A generative cell is formed by the more or less distinct or- ganization of the cytoplasm about the generative nucleus. This cell usually lies free in the body of the spore, but is often cut off by a distinct wall, as in Typha (Schaffner 31), Sparganium ( Campbell 43), Naias (Campbell29), Con call aria (Wiegand 45), Neottia (Guignard '•'), Populus (Chamberlain30), Asclepias 131 MORPHOLOGY OF ANGIOSPERMS (Trye56), and Barcodes (Oliver18). Both methods are often found in the same species and even in the same anther, as in Lilium (Fig. 63). The free generative cell finallv assumes a varietv of forms, the most common being lenticular, the cytoplasm massing chiefly Pig. 63. — Male gametophyte at time of shedding. J3, 0, Lilium auratum; the others L.tiijiinum; x 500. A, generative cell against side of microspore ; .5, generative cell in body of microspore ; the two male nuclei already formed ; <~\ three male nuclei within generative cell, an unusual case; /'.two male nuclei, differing in size, within generative cell; E, tube-nucleus divided, giving rise to six nuclei: /'. nn unusual case, showing tube-nucleus, two generative cells (g), and a " prothallial " cell (pn. — After Chamberlain.33 at two opposite poles of the nucleus. In some cases a spherical form is maintained, as in Acer (Mottier22) ; in others the len- ticular form passes into the vermiform, becoming elongated and THE MALE GAMETOPHYTE 135 even coiled or twisted, as in Tradescantia (Coulter and liose 14) ; or the cytoplasm of the spindle-shaped generative cell may taper into elongated whip-like filaments that more or less encircle the tube-nucleus, as in Eichhornia (Smith39). In Erythronium Schaffner55 found that the generative nucleus is larger than the tube nucleus and is surrounded by a densely staining amoeboid-form mass of cytoplasm. It is altogether probable that the size and form of free generative cells varies with age and external conditions, so that they may be relatively large or small ; or spherical, lenticular, spindle-shaped, or ver- miform in the same species. It is very common to find them at first spherical and later lenticular, as has been frequently ob- served in Lilium. In Lilium tigrinum Chamberlain 32 often found a small cell cut off by the microspore before the appearance of the tube and generative nuclei, and the same cell was noted after the division of the generative nucleus (Fig. 63). A similar cell was found by Smith 39 in Eichhornia crassipes and by Campbell 43 in Spar- ganium simplex. It is suggestive of a true vegetative or pro- thallial cell, two of which so commonly occur among the Gym- nosperms ; but the phenomenon is too unique as yet among Angiosperms to deserve more than a mention. The tube-nucleus usually increases much in size, and under certain conditions has been found to fragment, as in Lilium, in which Chamberlain 32 found four and in one case eight tube- nuclei ; in Eichhornia, in which Smith39 found two tube-nuclei in half the pollen-grains examined; in Hemerocallis, in which Fullmer 44 reports the frequent occurrence of two to six tube- nuclei ; and in Asclepias, in which Frye 56 observed a fragment- ing nucleus. This phenomenon is doubtless not uncommon in certain conditions of nutrition. The generative nucleus or cell may divide in the pollen- grain, even long before dehiscence, as in Sagittaria (Schaff- ner 31 ) ; or the generative cell may pass into the tube before division, sometimes not dividing until immediately before fer- tilization. The time of this division seems to hold no relation to the great plant groups, and may be variable in the same genus or even species. For example, in Lilium tigrinum it often takes place in the grain, but in L. phi/adelphicum rarely so; and in this last species it may occur either in the grain or at 10 136 MORPHOLOGY OF ANGIOSPERMS any time in the tube Up to its completed growth. The variable relation of the time of this division to the great groups may be illustrated by the following record: Anions; Monocotyledons the generative nucleus divides in the pollen-grain in Potamogeton (Wiegand45), Alisma (Schaff- ner2S), Sagittaria (Schaffner 31), Avena (Cannon46), Triti- cum and other grasses (Golinski 21), Lemna (Caldwell42), and Lilium (Chamberlain32); and in the pollen-tube in Symplo- carpus (Duggar47), Tradescantia (Coulter and Rose14), Eich- hornia (Smith39), Lilium (Chamberlain32), Convallaria (Weigand45), Erythronium (Schaffner55), and the Orchids (Guignard9). In examining this record it might be concluded that the early division of the generative cell within the pollen- grain is a more primitive character in general than the later division in the pollen-tube. Even if this should prove to be true for the Monocotyledons, it can hardly be claimed for the Dicotyledons, as the following record shows: Among Dicotyledons the generative nucleus or cell divides in the pollen-grain in Bhopalocnemis (Lotsy51), Papaver, Hesperis, Archangelica, and Mertensia (all by Strasburger 12), Nicotiana Tabacum (Guignard59), Sambucus (Halsted 1T), and Silphium (Merrell48) ; and in the pollen-tube in Pepero- mia (Johnson49), 8 alias (Chamberlain30), Ranunculus (Coul- ter37), Lathyrus (Strasburger12), Euphorbia (Miss Lyon40), Staphylea (Strasburger12), Acer (Mottier22), Vinca, Nemo- phila, Digitalis, and Torenia (all by Strasburger12), Campa- nula (Barnes13), and Datura laevis (Guignard59). It is evident that the two conditions are found among Dicotyledons in both primitive and high groups, and even in the same family (as Solanaceae), and that neither one has any claim to be regarded as an essentially primitive character. The male nuclei, formed by the division of a generative nu- cleus, are possibly always associated with cytoplasm in such a way that definite male cells are organized. The nucleus is often the only conspicuous feature, and in every case it finally constitutes the bulk of the male cell. In fact, in most of the plants studied only the male nucleus has been demonstrated in the pollen-tube and embryo-sac. In the following citations " male nucleus " and " male cell " are used to indicate whether cytoplasm was demonstrated or not. Various forms of male THE MALE GAMETOPHYTE 137 cells and nuclei have been described, but it is evident that the form as well as the size may change decidedly in the course of its history. For example, SchafTner 31 notes that the male nuclei in Sagittaria are at first spherical, but after pollination become bean-shape or spindle-shape. In Silphium Merrell 48 observed the originally spherical male nuclei become much elongated, more or less curved, and even spirally twisted while still within the pollen-grain (Fig. 64) ; and in Triticum and other grasses Go- linski 21 implies the same changes in form in describing the occurrence of male nuclei within the pollen-grain as " not unlike the antherozoids of a fern or of Cham." It has been re- peatedly observed that the spherical nuclei of the oblong or lenticular male cells of Lilium increase in size and become vermiform and variously curved and coiled after discharge from the pollen-tube, and the same phenomenon was observed by Miss Thomas 53 in Caltha. It seems to be generally true that the male cells when formed free in the body of the grain are at first spherical, but soon become oblong or lenticular. In a forthcoming paper by Koernieke it will be shown that in Lilium only male nuclei are found in the pollen-tube; at least there are no male cells as ordinarily figured. This claim is of special interest, since in Lilium male cells are clearly organized in the pollen-grain. The increase in size and change of form so often described as taking place in the tube or sac are probably phenomena of the male nucleus rather than of the male cell. There are well- known cases, however, in which the spherical or oblong form persists throughout the history of the nucleus. For example, in Peperomia (Johnson49) the male nucleus is spherical even in contact with the egg, and the same is true of several other forms recently investigated in connection with double fertilization. Fig. 64. — A, microspore of SilpM- um integrifolium, showing tube- nucleus and two male nuclei. B, later stage in S. terebinthina- ceum, showing the two male cells. C, single male cell of S. integrifolium, showing spiral form. — After Merkell.48 13S MORPHOLOGY OF ANGIOSPERMS There is also indication that the two male nuclei mav be- *j come differentiated in form, as in the case of Alisma, in which Schaffner 2S found the upper male nucleus in the pollen-tube elongated or spindle-shaped, and the lower one spherical. It is also probable that in cases of double fertilization the two male nuclei often assume different forms in the embrvo-sac. Four male nuclei have been reported by Strasburger 12 as some- times occurring in Camassia Fraseri, and Chamberlain 32 has observed three nuclei within a single male cell in Lilium dura- tion (Fig. 63, C). This recalls the spermatogenesis of Gymno- sperms, in which the generative cell gives rise to a stalk cell and two male cells, but it may have no further significance than that any active cell may be induced to divide by favorable conditions. The morphology of the structures included in the male gametophyte of Angiosperms is obscure. In 1884 Stras- burger12 suggested that only an antheridium is developed within the pollen-grain, the vegetative or prothallial tissue, rep- resented in many Gymnosperms, having been entirely sup- pressed. The same view has been developed in several papers from this laboratory, and in 1898 Belajeff 36 reiterated it in a discussion including both Gymnosperms and Angiosperms. According to this view, the larger tube-cell is the antheridium wall that develops a tubular outgrowth, used at least in Angio- sperms as the carrier of the male nuclei, while the generative cell and its product is the spermatogenous part of the antherid- ium. It is not exact to say that according to this view the whole pollen-grain is an antheridium, but that in its germina- tion the pollen-grain develops only an antheridium. Another view, which seems to be the only alternative, is that while only an antheridium is present its sole representative is the generative cell, the tube-cell not being any more a part of the gametophyte than is the embryo-sac. The divergence between the two views, therefore, has to do only with the nature of the tube-cell. In any event, it is important to note, as contra- dicting a very common statement, that the pollen-tube is not the male gametophyte. The development of the pollen-tube and the passage of the male nuclei to the embryo-sac are so directly connected with fertilization that they will be considered in the next chapter. THE MALE GAMETOPHYTE 139 LITERATURE CITED 1. Reichenbach, H. G. De pollinis Orchidearum genesi ac structura et de Orchideis in artem ac systema regigendis. Leipzig. 1852. 2. Schacht. H. Ueber den Bau einiger Pollenkorner. Jahrb. Wiss. Bot. 2 : 1U7-168. pis. 1^-18. 18(50. 3. Hofmeister, W. Neue Beitrage zur Kenntniss der Embryo- bildung der Phanerogamen. Abhandl. Konigl. Sachs. Gesell. Wiss. 6: 533-672. pis. 1-21. 1859. 4. Rosanoff, S. Zur Kenntniss des Baues und der Entwicklungsge- schichte des Pollens der Mimoseae. Jahrb. Wiss. Bot. 4: 441- 450. pis. 31-32. 1865. 5. Luerssen*. C. Zur Controverse iiber die Einzelligkeit oder Mehr- zelligkeit des Pollens der Onagrarieen, Cucurbitaceen und Cory- laceen. Jahrb. Wiss. Bot. 7 : 34-60. pis. 3-5. 1869. 6. Strasburger, E. Befruchtung und Zelltheilung. Jena. 1877. 7. Elfving, F. Studien iiber die Pollenkorner der Angiospermen. Jenaisch. Zeitsch. Xaturwiss. 13 : 1-28. 1879 ; Quart. Jour. Micr. Sci. 20: 19-35. 1880. 8. Strasburger, E. Zellbildung und Zelltheilung. Ed. 3. Jena. 1880. 9. Gognard, L. Recherches sur le developpement de l'anthere et du pollen des Orchidees. Ann. Sci. Nat. Bot. VI. 14: 26-45. pi. 2. 1882. 10. Strasburger, E. Ueber den Theilungsvorgang der Zellkerne und das Verhaltniss der Kerntheilung zur Zelltheilung. Archiv. Mikr. Anat. 21 : 476-590. pis. 25-27. 1882. 11. Corry, T. H. Structure and Development of the Gynostegium, etc.. in Asclepias Cornuti. Trans. Linn. Soc. Bot. London 2 : 173-207. pis. 24-26. 1884. 12. Strasburger. E. Neue Untersuchungen iiber den Befruchtungs- vorgang bei den Phanerogamen. Jena. 1884. 13. Barnes, C. R. The Process of Fertilization in Campanula amer- icana. Bot. Gazette 10: 349-354. pi. 10. 1885. 14. Coulter, J. M., and Rose, J. X. The Pollen Spore of Tradescantia rirginica. Bot. Gazette 11 : 10-14. pi 1. 1886. 15 Wille, N. L'eber die Entwickelungsgeschichte der Pollenkorner der Angiospermen und das Wachsthum der Membranen durch Intussusception. Christiania. 1886. 16. Goebel, C. Outlines of Classification and Special Morphology. English translation. 1887. 17. Halsted, B. D. Three Nuclei in Pollen Grains. Bot. Gazette 12: 285-288. pi. 16. 1887. 18. Oliver. F. W. On Sarcodes sanguinea. Annals of Botany 4: 303-326. pis. 17-21. 1890. 14:0 MORPHOLOGY OF ANGIOSPERMS 19. Guignard, L. Nouvelles etudes sur la fecondation. Ann. Sci. Nat. Bot. VII. 14 : 163-296. pis. 9-18. 1891. 20. Chauveaud, G. L. Sur la fecondation dans les cas de polyembry- onie. Compt. Rend. 114: 501. 1892. 21. Golinski, St. J. Ein Beitrag zur Entwicklungsgeschichte des An- droeceums und des Gynaeceums des Griiser. Bot. Centralbl. 55 : 1-17, 65-72, 129-135. pis. 1-3. 1893. 22. Mottier, D. M. Development of the Embryo-sac in Acer rubrum. Bot. Gazette 18 : 375-377. pi. 34. 1893. 23. Humphrey, J. E. Nucleolen und Centrosomen. Ber. Deutsch. Bot. Gesell. 12: 108-117. pi. 6. 1894. 24. Belajeff, W. Zur Kenntniss der Karyokinese bei den Pflanzen. Flora. Erganzungsband, 1894. 25. Strasburger. E. Karyokinetische Probleme. Jahrb. Wiss. Bot. 28 : 151-204. pis. 2-3. 1895. 26. Mottier, D. M. Beitrage zur Kenntniss der Kerntheilung in den Pollenmutterzellen einiger Monokotylen und Dikotylen. Jahrb. Wiss. Bot. 30 : 169-204. pis. 3-5. 1897. 27. Campbell, D. H. The Structure and Development of the Mosses and Ferns. London and New York. 1895. 28. Schaffner, J. H. The Embryo-sac of Alisma Plantago. Bot. Gazette 21 : 123-132. pis. 9-10. 1896. 29. Campbell, D. H. A Morphological Study of Naias and Zannichel- lia. Proc. Calif. Acad. Sci. III. 1 : 1-62. p/s. 1-5. 1897. 30. Chamberlain, C. J. Contribution to the Life History of Salix. Bot. Gazette 23 : 147-179. pis. 12-18. 1897. 31. Schaffner, J. H. Contribution to the Life History of Sagittaria variabilis. Bot. Gazette 23 : 252-273. pis. 20-26. 1897. 32. Chamberlain, C. J. Contribution to the Life History of Lihum Philadelphicum ; the Pollen Grain. Bot. Gazette 23: 423-430. pis. 35-36. 1897. 33. Juel, H. O. Die Kerntheilungen in den Pollenmutterzellen von Hemerocallis fulva und die bei denselben auftretenden LTn- regelmassigkeiten Jahrb. Wiss. Bot. 30: 205-226. j)ls. 6-8. 1897. 34. Schaffner, J. H. The Development of the Stamens and Carpels of Typha latifolia. Bot. Gazette 24: 93-102. pis. 4-6. 1897. 35. Lawson, A. A. Some Observations on the Development of the Karyokinetic Spindle in the Pollen Mother -cells of Cobaea scandens. Proc. Calif. Acad. Sci. III. 1: 169-184. pis. 33-36. 1898. 36. Belajeff, W. Die verwandtschaftlichen Beziehungen zwischen den Phanerogamen und den Cryptogamen in Lichte der neues- ten Forschungen. Biol. Centralbl. 18: 209-218. 1898. 37. Coulter. J. M. Contribution to the Life History of Ranunculus. Bot. Gazette 25 : 73-88. pis. 4-7. 1898. THE MALE GAMETOPHYTE 141 38. Chamberlain, C. J. Winter Characters of Certain Sporangia. Bot. Gazette 25: 124-128. pi. 11. 1898. 39. Smith. R. W. A Contribution to the Life History of the Ponte- deriaceae. Bot. Gazette 25: 324-337. pis. 19-20. 1898. 40. Lyon, Florence M. A Contribution to the Life History of Euphorbia corollata. Bot. Gazette 25: 418-426. £>Zs. 22-24. 1898. 41. Stevens, W. C. The Behavior of the Kinoplasm and Nucleolus in the Division of the Pollen Mother-cells of Asclepias Cornuti. Kansas Univ. Quarterly 7 : 77-85. pi. 15. 1898. 42. Caldwell, O. W. On the Life History of Lemna minor. Bot. Gazette 27 : 37-66. Jigs. 59. 1899. 43. Campbell, D. H. Notes on the Structure of the Embryo-sac in Sparganium and Lysichiton. Bot. Gazette 27: 153-166. pi. 1. 1899. 44. Fullmer, E. L. The Development of the Microsporangia and Mi- crospores of Hemerocallis fulva. Bot. Gazette 28 : 81-88. pis. 7-8. 1899. 45. Wiegand, K. M. The Development of the Microsporangium and Microspores in Convallaria and Potamogeton. Bot. Gazette 28: 328-359. ills. 24-25. 1899. 46. Cannon, W. A. A Morphological Study of the Flower and Em- bryo of the Wild Oat, Avena fatua. Proc. Calif. Acad. Sci. III. 1 : 329-364. pis. 49-53. 1900. 47. Dlggar, B. M. Studies in the Development of the Pollen Grain in Symplocarpus foetidus and Peltdndra undidata. Bot. Gazette 29 : 81-98. pis. 1-2. 1900. 48. Merrell, W. D. A Contribution to the Life History of Silphium. Bot. Gazette 29 : 99-133. pis. 3-10. 1900. 49. Johnson, D. S. On the Endosperm and Embryo of Peperomia pellucida. Bot. Gazette 30 : 1-11. pi. 1. 1900. 50. Juel, H. O. Beitrage zur Kenntniss der Tetradenbildung. Jahrb. Wiss. Bot. 35: 626-659. pis. 15-16. 1900. 51. Lotsy, J. P. Rhopalocnemis phalloides Jungh., a Morphological- systematical Study. Ann. Jard. Bot. Buitenzorg II. 2 : 73-101. pis. 3-14, 1900. 52. Byxbee, Edith. The Development of the Karyokinetic Spindle in the Pollen Mother-cell of Lavatera. Proc. Calif. Acad. Sci. III. 2: 63-82. pis. 10-13. 1900. 53. Thomas, Ethel M. On the Presence of Vermiform Nuclei in a Dicotyledon. Annals of Botany 14: 318-319. 1900. 54. Strasburger, E. Einige Bemerkungen zu der Pollenbildung bei Asclepias. Ber. Deutsch. Bot. Gesell. 19: 450-461. pi. 24. 1901. 55. Schaffner. J. H. A Contribution to the Life History and Cy- tology of Erythronium. Bot. Gazette 31 : 369-387. pis. 4~9. 1901. 142 MORPHOLOGY OF ANGIOSPERMS 56. Frye, T. C. Development of the Pollen in some Asclepiadaceae. Bot. Gazette 32: 325-331. pi. 13. 1901. 57. Rosenberg, O. Ueber die Pollenbildung von Zostera. Meddel. Stockholms Hogsk. Bot. Inst. pp. 21. 1901. 58. Gager, C. S. The Development of the Pollinium and Sperm Cells in Asclepias Cormiti. Annals of Botany 16: 123-148. pi. 7. 1902. 59. Guignard, L. La double fecondation chez les Solanees. Jour. Botanique 16: 145-167. figs. 45. 1902. 60. Cheauveaud, G. L. De la reproduction chez le dompte-venin. Diss. Paris. 1902. CHAPTER VII FERTILIZATION In various ways the male gametophyte reaches the stigma. The literature dealing with pollination has become very exten- sive, and can not even be recapitulated here, especially as it is an ecological subject. The development of tubes from pollen- grains lodged upon stigmas has long been known, but the rela- tion of the tubes to fertilization was long misunderstood. An historical account of the early views of fertilization among An- giosperms, together with the citation of literature, was given bv Schacht1 in 1850, and by Hofmeister 2 in 1851. A few notes from Schacht's account may not be without interest, and the reproduction of some of his figures will serve to show the technique of the time and to illustrate how theories may in- fluence interpretation (Fig. 65). In 1681 Malpighi discovered the ovule and the embryo- sac, and also examined the pollen, but regarded it as a useless secretion. Xo important advance was made until 1823, when Amiei discovered the pollen-tube on the stigma of a Portulaca and succeeded in tracing the tube to the ovule. In 1826 Bron- gniart traced the pollen-tube in many plants, and in Pepo macrocarpus saw hanging from the micropyle the end of the tube that had passed into the embryo-sac ; " but," says Schacht, " he misinterpreted the phenomenon, for he regarded the pol- len-tube as a fertilizing tube through which the fertilizing con- tents were brought to the embryo-sac, there to be taken up by the ' embryonal vesicle,' a coll arising in the sac." In 1826 Robert Brown described the development of the integuments, and later traced the pollen-tubes of orchids and asclepiads from the stigma to the micropyle. Tn 1833 the knowledge of the subject may be summarized as follows : there had been observed 143 144 MORPHOLOGY OP ANGIOSPERMS the pollen-grain with its pollen-tube and some contents, as well as the ovule with its integuments and embryo-sac; and the pollen-tube had been traced from the stigma to the embryo-sac. Fig. 65.— A-C, Orchis Mono; D, 0. latifolia; F, O. maculata; F Canna limbata. A-B, young ovules, x 150; C, end of pollen-tube enlarging, x J 00; D, later sta^re with two nuclei visible in embryo, x Hii>: F, more advanced embryo, x 208; F, considerably later stage, x 125.— After Schacht.' In 1835 Schleiden, the founder of the cell-theory, traced the pollen-tube in a large number of widely separated families. He claimed to have seen the tube enter the micropyle, press into the embryo-sac, and then become itself the embryonal vesicle, the beginning of the embryo. He thought that the contents of the pollen-tube not only give rise to the embryonal vesicle, but that the end of the tube, nourished by the embryo-sac, becomes the future plant. FERTILIZATION" 145 In 1842 Hartig described an " egg " in the embryo-sac, and claimed that the pollen-tube carries a substance that fertilizes the egg, a view which Schleiden promptly opposed. In the same year Amiei reiterated his previous views and claimed for Orchis and other plants the preexistence in the embryo-sac of a cell which, through the influence of the pollen-tube, becomes the embryo. Schacht opposed this claim, and suggested that such antiquated ideas be abandoned. At the same time, Hugo von Mohl described the egg-apparatus in Orchis Morio, and warmly supported Amici's views. In his conclusion Schacht says : " The tendency to error is so bound up in human nature that the work of one's mind, like that of his hand, is never perfect, and consequently I do not consider my work free from error and misconception, but I have tried to minimize these as much as possible. In the chief A C Fig. 66. — A, Staph if lea; tip of pollen-tube showing division of generative nucleus. B, Orchis latifolia ; end of pollen-tube showing tube nucleus (in advance) and the two male nuclei. C, Monotropa Hypopitys; fusion of sex nuclei, male nucleus more deeply shaded. />, the same stage just after fertilization, showing tirst division of endosperm nucleus, x 450. — After Strasburger.8 matter, the origin of the embryo from the pollen-tube, no one can convince me that there has been any error or misconcep- 146 MORPHOLOGY OF ANGIOSPERMS tion." Nevertheless, in his text-book, published a few years later, he says that " fertilization ,;I is accomplished in plants, as in animals, by the union of male and female elements. It is only since 1875 that detailed information has gradu- ally accumulated; and not until 1884 (Strasburger 8) were the cells concerned in fertilization clearly pointed out (Fig. 66). The tube-cell of the pollen-grain in various ways pushes tli rough the exine a papillate protrusion of the intine that develops into the pollen-tube with greater or less rapidity. Crowding among the loose papillate cells of the stigma, the elongating tubes enter the conducting tissue of the style. Ordi- narily the style is solid, and the tubes grow along the conducting strand, which they disorganize more or less, obtaining from it their nutritive supply. In case there is a stylar canal the tubes either pass down it, as in Pontederia (Smith28) and Erythro- nium (Sehaffner 51), nourished by the lining glandular cells, or they may penetrate the stylar tissue about the tube, as in Campanula (Barnes9) and Jugla ns ( Xawaschin 20). In many cases the tube enters the ovary cavity close to the micropyle; in others it must traverse more or less of the cavity, being guided " to the micropyle by various mechanical and nutri- tive contrivances. Although ordinarily pollen-tubes are developed only in con- tact with the stigma, in cleistogamous flowers tubes have been observed issuing from pollen-grains still in the anther, the tips being directed toward the stigma. In Asclepias also multi- tudes of tubes sometimes start from the unremoved pollinia. The time elapsing between pollination and fertilization, as inferred from the presence of pollen-tubes in the embryo-sac, is extremely variable, and seems to hold no relation to the dis- tance traversed, as shown by Hofmeister,3 in comparing Crocus, in which a style 6 to 10 cm. long was traversed in one to three davs, with Arum, in which a stvle onlv 2 to 3 mm. long was traversed in five days. The range in time is probably repre- sented by the following illustrations: In Limnocharis emargi- nata Hall 57 found a two-celled embryo in material killed eighteen hours after pollination, and thinks that in this case fertilization probably occurs the first night after pollination. Probably the most accurate estimate of the time is that by Mottier 26 for Lilium, in which the time between artificial pol- FERTILIZATION 147 lination and fertilization (as shown bv fusion) was sixty-five to seventy-two hours. Guignard u6 has recorded an interval of two days between pollination and fertilization in Nicotinana Taba- cum. Juel 63 found by artificial pollination that fertilization occurs in Cynomorium four days after pollination, sixteen days after pollination embryos of various sizes being found. Hofmeister 2 noted the interval as one to three days in Crocus, five davs in Arum, from ten davs to several month- among the Orchidaceae, and in Colchicum autumnale not less than six months (Xovember to May). In the last case, as is well known, pollination sometimes occurs before there is any appearance of ovules. Miss Benson 15 found three weeks elaps- ing in Fagus sylvatica between pollination and the entrance of the tube into the embryo-sac, and the same interval is reported by D'Hubert 17 for certain Cactaceae. In Hamamelis virgini- ana Shoemaker 62 has found that pollination occurs from Octo- ber to December ; that the tubes develop at once and grow rapidly until cold weather; that during January and February the tube may be found safely embedded in the hairy part of the carpel ; and that growth is resumed in the spring, fertilization occurring about the middle of May, five to seven months after pollination. The pollen-grains of Hamamelis show great resist- ance to low temperature, Shoemaker citing eases in which they produced tubes after exposure to a week of cold, the tempera- ture sometimes being as low as —15° C. Among the Amentif- erae, however, the interval becomes even more extended. Miss Benson 15 reports that it is one month in Betula alba, two months in Carpinus Betulus, three months in Alnus glutinosa, four months in Corylus Arellano, and Quercus Robur, and as much as eleven months in certain other oaks ; while in Q. velu- tina Conrad36 found the interval between pollination and fer- tilization to be thirteen months. Baillon had long before noted that no indication of ovules is present in Quercus at the time of pollination. Goebel 10 has associated these long intervals with the woody habit, citing Ulmus, Quercus, Fagus, Juglans, Citrus, Aesculus, Acer, Cornus, and Bobinia as illustrations, and stating that the interval is almost a year in American oaks that take two years to ripen their seed. Such cases bear a striking resemblance in this regard to many Gymnosperms. A recent study of Monotropa uniftora by Shibata C5 indi- 148 MORPHOLOGY OF ANGIOSPERMS cates that the interval between pollination and fertilization in any given species may be dependent upon temperature. In the ease of Monotropa, under normal conditions fertilization takes place about five days after pollination ; but by lowering the tem- perature the interval is lengthened, and at 8-10° C. fertilization is prevented. In Shibata's experiments it was shown that light, atmospheric pressure, and mechanical injury seem to exert no influence upon fertilization and subsequent phenomena, but that the structures of the embryo-sac are very sensitive to temperature. In a long pollen-tube, or in one that persists for a long time, it is common to observe the formation of successive cellulose plugs (Propfen) that shut off the growing tip, with its cells and nuclei, from the cavity behind, as fully described by Stras- burger 4 and Elfving.6 Sometimes the plugs are so large and persist in such a series that they become conspicuous objects, as in Gymnadenia conopsea (Marshall-Ward7), Campanula americana (Barnes9), Sarcodes sanguined (Oliver11), etc. In such forms as the Amentiferae and others, in which the tube and its contents remain imbedded in the stylar tissue for a period varying from one month to over a year, the tip of the tube is cut off by a plug, its wall thickens, and it passes into what might fairly be called an encysted condition, as suggested 1 >v Miss Benson 15 in connection with Carpinus. The branching of pollen-tubes, so conspicuous a phenome- non among Gymnosperms, is also found among certain Angio- sperms. Hofmeister 3 observed branching tubes among Mono- cotyledons in Potlios longifolia and Hippeastrum aulicum. Among the Amentiferae it seems to be very common, Miss Benson15 observing forking tubes in several of the genera (Corylus, Carpinus, etc.) she studied, and in Quercus a cluster df short branches at the end of the tube; while Xawaschin 20, 30 states that the tubes of Juglans and Ulmus branch profusely, and recently a similar branching has been noted by Billings 6fi in Carya (Hicoria). Zinger 31 also described the pollen-tubes of the Cannabirieae as ending in numerous swollen sac-like 1 tranches. The breaking up of the tip of the tube into short branches is doubtless a common phenomenon, probably associ- ated with the rhizoidal habit, hut free branching seems to be characteristic chiefly of chalazogamic forms. FERTILIZATION 149 In 1891 Treub12 announced the phenomenon of chalazog- amy in l asuarina. He found the pollen-tube penetrating the ehalazal region of the ovule, instead of entering through the micropyle. In this case the pollen-tube becomes associated with the numerous elongated sterile megaspores, and doubtless they are of service in rendering the passage easy; and later it enters the antipodal region of the embryo-sac and approaches the egg- apparatus from that direction (Figs. 67, 24 C). In 1893 Nawaschiri14 reported chalazogamy in Betula', and in 1894 Miss Benson 15 not only observed the phenomenon in Betula, but also added Alnus, Coryhts, and Carpinus to the list of chalazogamic plants. In all of these cases Miss Benson ob- served the tubes following a course parallel with the vascular strands <>f the raphe, thus reaching and penetrating the chalaza. In Cory- hts and Carpinus the tube enters a more or less conspicuous caecum developed in the antipodal region of the sac, traverses it, and comes in contact with the egg; but in Alnus the tube traverses the nucel- his to the micropylar region above the embryo-sac, and then turns and enters it as though it had come by way of the micropyle. In 1895 Xawasehin 20 added Juglans cine- rea and -/. regia to the list. In the latter species the tube does not pass down the stylar canal or traverse the cavity of the ovary, but advances through the tissue of the style and of the ovary wall until opposite the insertion of the single ovule1 that fills the ovary cavity. It then leaves the ovary wall and pierces the chalaza, branching freely in the nu- cellus, which is described as " veined " by tubes surrounding the sac on all sides. The male nuclei discharged into the sac were seen " wandering " in its cytoplasm and fusing with one of sev- eral free cells that function as eggs but have not organized an egg- apparatus. Recently Billings r,e has discovered chalazogamy in Carya olivaeformis, the common pecan, the details conforming almost exactly to those given by Xawaschin for Juglans regia. Fig. 67. — Casuarina suberom. Ay pollen-tube entering ehalazal end of embryo-sac, x 270; B, stage showing (Treub's interpretation) formation of endosperm before fer- tilization, x 180. After Tkeib.12 150 MORPHOLOGY OB' ANGIOSPERMS In 1898 Nawaschin 30 described some remarkable variations in the course of the pollen-tube in Ulmus pedunculata and U. montana. In addition to tubes following the ordinary chala- zogamic route, some instead of penetrating the chalaza pass from the funiculus across the short outer integument, and thence into and upward through the inner integument to the top of the nucellus, when they turn across to the bottom of the micropyle and so enter the nucellus from the usual direction; others follow the same route except that they pass directly from the funiculus into the inner integument ; while still other tubes branch profusely and apparently with no definiteness within both the funiculus and integument. In the same species, there- fore, pollen-tubes may enter the sac either at the antipodal or micropylar ends, and may either pass with great directness or branch profusely. The behavior of the pollen-tubes in Ulmus suggested that there might be other routes than through the micropyle or through the chalaza, and this has been observed in other forms. In his study of the Cannabineae in 1898, Zinger 31 discovered that the two thick integuments completely coalesce over the apex of the nucellus, and the micropyle is entirely closed by tissue. The pollen-tube either bores its way through the tissue filling the micropyle or pierces the two integuments, reaching the nucellus and branching about its apex, and finally sending one very slender branch into the embryo-sac. With these facts before them, Pirotta and Longo 41 pro- posed the term " acrogamy " for the entrance of the pollen-tube directly through the micropyle ; " basigamy " for its entrance through the chalaza (Casuarina, Betula, Alnus, Corylus, Carpi- nus, Juglans, and sometimes Ulmus) ; and " mesogamy '' for its entrance by intermediate routes (sometimes Ulmus, and Cannabineae). In the following year Longo 49 described a case of mesogamy in Cucurbita, in which the pollen-tube traverses the tissues of the funiculus and outer integument before enter- ing the micropyle. Practically the same phenomenon has been observed by Murbeck 50 in Alchemilla arvensis, in which the micropyle is entirely closed by the growth of the integument, and the pollen-tube enters the ovule at the chalazal end, trav- erses the entire length of the integument within its tissues, and thus enters the micropylar extremity of the embryo-sac. FERTILIZATION 151 True chalazogamy, therefore, has as yet been found only among the Anient iferae, but such an intermediate condition as shown by Ulmus, Cucurbita, and Alchemilla, in which the pollen-tube enters the ovule at the chalazal end, but traverses the integument instead of the nucellus, suggests that chala- zogamy is an exceptional condition derived from the ordinary route of the pollen-tube through the micropyle. In certain cases the tube reaches the micropyle by passing along more or less of the surface of the integument ; in other cases it enters the tissues of the integument, and finally it penetrates deeper, entering the chalazal tissue. This seems to be a natural sequence of events that resulted in chalazogamy, which there- fore would hold no relation to a primitive condition of Angio- sperms or to their classification. In passing through the micropyle the pollen-tube is more or less compressed, and upon reaching the wall of the embryo- sac may broaden out upon it. In some cases (p. 94) the synergids have already pierced the wall of the embryo-sac, but in most cases it must be pierced by the tube. Upon entering the sac the tube either passes between the synergids, as in Ponte- deria (Smith28), Euphorbia (Lyon29), sometimes Salix (Chamberlain23), etc. (Fig. 44) ; or between the sac-wall and one synergid, as in Alisma ( Schaffner 22), Lilium (Coulter23), Ranunculus (Coulter27), Fagus (Benson15), Silphium (Mer- rell35), etc. Recently, however, Guignard 56 has reported that in Nicotiana TabaCum and Datura laevis the tube passes into a synergid and discharges its contents into the broken-up body. So far as our own observation goes, the usual route of the tube is between the sac-wall and one of the synergids, but this may well vary even in the same species. Within the sac the tip of the tube usually becomes much swollen, often appearing pouch-like, as in Alisma, Erythronium, Ranunculus, Silphium, etc., due probably to the rapid absorption of material from the synergid. As a rule, one synergid is disorganized by its contact with the tube; but in Salix (Chamberlain23) (Fig. 44), Sil- phium (Merrell35), Nigella (Guignard53), etc., cases of fer- tilization have boon observed in which both synergids remained intact; while in Erir/eron (Land38) both synergids are fre- quently disorganized. D'Hubert 1T has made the interesting observation in connection with his studv of the Cactaceae that 11 152 MORPHOLOGY OF ANGIOSPERMS the nucleus of one synergid moves toward the tube upon its entrance into the sac, and that the nucleus of the other synergid moves toward the nucleus of the egg. In case the tube passes between the synergids it advances directly toward the egg-nucleus ; but in case it passes along the wall of the sac the tip of the tube curves toward the egg-nucleus. In any event, the tip of the tube, in which a thin area ( pit ) is developed, is directed toward the egg-nucleus when the dis- charge takes place. Under the pressure developed by the turgor of the end of the tube, and re- sisted by the small caliber of the tube in its passage through the micropyle and sac-wall, the membrane of the pit is ruptured, and a discharge of the contents results. The perforated tip of the pollen-tube, after the discharge, has been demonstrated fre- quently, as seen by Schaffner 4+ in Sagittaria (Fig. 68). The discharge seems to be forcible enough to empty the end of the tube of most of its contents, the most important ones being the two male nuclei. Cases have been reported in which only one male nucleus is said to be discharged, as in Alisma (Schaffner22) and Sagittaria (Schaffner24), the other being recognized as degenerating in the tube. However, the frequent presence of disorganizing bodies within the tube after fertilization (Fig. 71), and numerous observations of the discharge of both male nuclei, and especially the rapidly multi- plying illustrations of " double fertilization," incline to the be- lief that the discharge of both male nuclei into the sac is usual. The passage of the male nucleus through the cytoplasm of the egg toward the female nucleus may be attended by an increase in size and change in form, but the changes are not so conspicuous as those that occur in the male nucleus that passes deeper into the sac to fuse with the polar nuclei. For example, in Calilia palusfris ^liss Thomas44 found the male nuclei very Fig. 68. — Sagittaria variabilis. Pollen- tube in the act of discharging; four centrosomes represented; x 900. — After Schaffner.34 FERTILIZATION 153 small and oblong or lenticular on extrusion, the one passing to the polar nuclei increasing very much in size, the other very little. In Tricyrtis hirta Ikeda 5S found the male nucleus that passes to the polar nuclei showing •'enormous change in size and shape " as it passes through the sac. There is usually more or less elongation of male nuclei at the time of discharge or afterward, but in MonotrO[ja uitifiora Shibata 54 has seen them elongated when entering the sac, but becoming more nearly spherical as fusion progresses. In the pollen-grain at the time of shedding the generative nucleus stains blue and the tube nucleus red with a combination like cyanin and erythrosin. This reaction is maintained, the male nucleus staining blue even after coming into contact with the nucleus of the egg which stains red; but as fusion proceeds the male nucleus takes less and less of the cyanin and finally stains with erythrosin like the nucleus of the egg. The fusion of the male and female nuclei may be very rapid, as observed by Guignard 4S> 53 in Zea and Ranuncula- ceae ; or the two may be long in contact without fusion, as noted by Johnson37 in Peperomia. The behavior of the chromatin during fusion has received but little attention. Mottier 26 fig- ures the chromatin when the nuclei are partly fused, and the statement is generally current that the nuclei fuse in the resting condition (Fig. 69). In view of the independence of the pater- nal and maternal chromatin dur- ing fertilization in Gvnmo- -perms, as recently noted by several investigators, it would be well to reexamine the subject in Angiosperms, especially since most observers have paid little ot no attention to this phase of the problem. Since it has been in connec- tion with fertilization and at- tendant phenomena that the cen- trosome problem has come into greatest prominence, it may not be inappropriate to refer to the subject at this point. Guignard, Schaffner, and others have Fig. 69. — Lilium candidum. Fusion of sex nuclei ; the synergids appear as dense homogeneous masses. — After ' MuTTlEIi.2* 154 MORPHOLOGY OF AXGIOSPERMS regarded the centrosome as a permanent organ performing an important function in mitosis and in fertilization. Even the " quadrille of the centers," described by the zoologist Fol, was identified by these observers. Centrosomes in the vascular plants have been figured by many other prominent botanists, including Humphrey,10 Strasburger,18 Campbell,19 and Mot- !:i: !;:.'.v\ Fig. 70.— Figures of centrosomes in vascular plants. A, Lilium Martagon, the reduction division at germination of megaspore ; 12 chromosomes may be counted: x 600; after Guignabd.13 B, Larix europaed, first division of pollen mother-cell: x 600; after Stbasbubgeb." C, Delpkinwim tricorne, first division of megaspore mother- cell; "at upper pole are centrospheres " ; x 588; after Mother.9' Z>, Sagittaria variabilis, first division of pollen mother-cell ; x 640; after Schaffner24 E, Lilium candidum, reduction division at germination of megaspore; after Bernard.47 F, Piilotum triquetrum, first division of spore mother-cell; x 800: after Humphrey.18 G, Equisetum telemateia, tetrad of four spores ; x 960; after Campbell.18 tier26 (Fig. 70). Most botanists, following Strasburger. have publicly renounced any belief in the centrosome as an organ of FERTILIZATION 155 vascular plants, and many others have made a tacit renuncia- tion. To say that all the figures that have been drawn have Fig. 71. — Double fertilization. A. Helianthus annuus, showing the two coiled male nuclei, one fusing with the egg-nucleus and the other with the endosperm nucleus; after Nawaschin.*1 B, Ira, the two polar nuclei not yet fused; after Guigxard.3* C. Silphium laciniatum: spt, up?, male nuclei: o, oosphere; e, endosperm nucleus; sy, synergid ; pt, pollen-tube ; x, two conjectural bodies often seen in the pollen- tube after the male nuclei have been discharged : x 525 ; alter Land.38 been mere products of the imagination would be a radical state- ment, and one doubtless verv far from the truth. In our opinion the observations, figures, and descriptions, like the pollen-tube embryos of Schleiden and Schacht, furnish an exam- ple of the extent to which even a careful and conscientious scientist may be influenced by preconceived opinion. Our knowledge of the phenomenon called " double fertili- zation " (Fig. 71) dates from 1898, when Nawaschin 33, 34 an- 156 MORPHOLOGY OF ANGIOSPERMS nounced at a meeting of the Russian Society of Naturalists in August that it occurs in Lilium Martagon and Fritillaria ten- ella. In 1S99 Guignard32 observed the same phenomenon in Lilium pyrenaicum, Fritillaria meleagris, and Endymion nutans. During 1000 the literature of the subject increased rapidlv. Nawaschin40 added Juglans, Delphinium elatum, Rudbechia speciosa, and Helianthus annuus to the list, and in certain orchids (Arundina and Phajus) he found the second male nucleus consorting with the polar nuclei, but there was no fusion. Guignard 39 described the phenomenon in species of Tulipa (Fig. 72), also42 in Scilla, Narcissus, Reseda, and Hibiscus; and Strasburger 43 not only added Himantoglossum, Fig. 72.-^4, embryo-sac of Tulipa sylvestris, showing nuclei scattered irregularly, each nucleus surrounded by a rather definitely limited portion of the cytoplasm; x B, T. Celsiana, showing double fertilization in sac like that shown in A ; the male nuclei recognized by vermiform appearance; x 333. — After Giignard.39 certain species of Orchis, and Monotropa Hypopitys, but dis- cussed the whole subject. ^liss Thomas 14, 45 reported double fertilization in Caltha palustris; Guignard42 announced it in Ranunculus Flammula, Helleborus foetidus, Anemone nemo- rosa, Clematis, VUicella, and Nigella sativa, and independently confirmed its occurrence in Caltha palustris. Land 3S found it in species of Erigeron and Silphium : it was observed repeatedly FERTILIZATION 157 in this laboratory in Lilium philadelphicum (Fig. 36, H), L. iritjrinum, and Anemone patens Nuitalliana; and at the close of 1900 .Miss Sargant 46 published a resume and general discus- sion of the subject. More recently, Guignard 4S has described double fertilization in Zea and Xaias major; Land has discov- ered it in Cnicus and possibly in Taraxacum ; while Guignard 53 has added Nigella damascena and Ranunculus Cymbalaria', and Frye60 has described its occurrence in Asclepias Cornuti. Karsten 55 has also confirmed the occurrence of double fertili- zation in Juglans, investigating several species; Shibata 54 has added Monotropa uniflora, Ikeda 58 Tricyrtis hirta, Strasbur- ger 59 Ceratophyllum demersum, Guignard56 species of Nico- tiana and Datura, as well as of Capsella and Lepidiumf4 Wvlie 6T Elodea, and Frve °8 Casuarina. It will be seen that the phenomenon is not restricted to a few groups, but is widely displayed among both Monocotyledons and Dicotyledons ; among the former having been observed in Xaiadaceae, Hydroeharitaceae, Gramineae, Liliaceae, Amaryl- lidaceae, and Orchidaceae ; and among the latter in Juglanda- ceae, Ceratophyllaceae, Ranunculaceae, Cruciferae, Resedaceae, Malvaceae, Ericaceae, Asclepiadaceae, Solanaceae, and Com- positae. Probably it is not safe to. infer the general occurrence of double fertilization, although the observations already include sixteen families, about forty genera, and over sixty specie-, besides inferential testimony in other species from the form and activity of both male nuclei and from the phenomenon of xenia. In any event, it is common enough to demand a general explana- tion of its significance, its place in the history of Angiosperms. and especially whether it is really fertilization or merely triple fusion. It has certainly introduced among' structures alreadv difficult of interpretation a phenomenon that immensely in- creases the difficulty. The subject will be discussed briefly under endosperm (Chapter VIII), and only such general details presented here as have been observed in connection with the process. It is claimed by Guignard for Lilium, and confirmed by Miss Thomas in Caltha, that the first male nucleus extruded from the tube passes to the polar nuclei. The frequently vermi- form and spiral character of this nucleus has suggested the possi- bility of independent motion; but this form is by no means con- 158 MORPHOLOGY OF ANGIOSPERMS stant, and Strasburger,43 in examining the process in living material of Monotropa, demonstrated the passage of the male nucleus in the streaming protoplasm of one of the cytoplasmic strands connecting the primary endosperm nucleus or the polar nuclei with the egg-apparatus. This is confirmed by Guig- nard,53 who has described and figured the very small male nucleus passing down the broad cytoplasmic strand that con- nects the egg-apparatus with the antipodals and envelops the primary endosperm nucleus in Xigella, Damascena, Ranunculus Cymbalaria, and Anemone nemorosa, and which is doubtless true of the other Kanunculaceae. It seems probable that the male nucleus is generally carried along one of these strands : but it is not improbable that the vermiform nuclei occasionally acquire some power of independent motion. It is during this passage that the male nucleus may increase much in size (Thomas,44 Ikeda 58) and may even assume the vermiform character; although all such changes may have occurred before discharge from the pollen-tube, even in the pollen-grain, as observed by Merrell 3f' in Silphium. The male nucleus, how- ever, may retain its small size and oval form even in contact with the polar nuclei, as observed by Guignard 32 in Endymion, and by other observers since. In Juglans Karsten 55 believes that in all cases the polars are fertilized before the egg; but in Xicotiana Tabacum Guignard 56 reports that sometimes the egg- is fertilized first and sometimes the polars, so that probably there is no definite order in the two fusions. Every possible order in the fusion of the three nuclei has been observed, so that the 'triple fusion is brought about in a variety of ways. As might be expected, it is often the case that the polar nuclei have already fused when the pollen-tube enters the embrvo-sac, and the male nucleus unites with the fusion nucleus, as in Tricijrtis, Eanunculaceae, Datura, Erigeron, Sil- phium, etc. ; although even in this case the polar nuclei may not always lose their individuality. The two polar nuclei and the male nucleus have also been observed to fuse all together, as in Zea (Guignard48) and other plants, in which the vermiform male nucleus seems to bind the polar nuclei together. In Xicoti- ana (Guignard 56) the male nucleus comes in contact with either polar nucleus or both. In Lilium Martagon the male nucleus usually fuses first with the upper polar nucleus, and later the FERTILIZATION 159 lower polar nucleus enters the combination, as was also observed by Shiu~ta :'4 in Monotropa uniflora; but in Lilium it has been observed that if the lower polar nucleus happens to be the more favorably placed the male nucleus fuses with it first. In Ascle- pias Cor nut i (Frye 60) both male nuclei are vermiform and more or less curved, and one of them was observed in contact with a polar nucleus near the antipodal cells, the mieropylar polar nucleus being- some distance away and nearer the egg- apparatus. That the male nucleus may thus traverse much of the embryo-sac is also shown in Nigella darnascena and Anem- one nemorosa, in both of which Guignard 53 observed the male nucleus uniting with the fusion nucleus near the prominent antipodal cells. At present there is a decided tendency among botanists and zoologists to distinguish two distinct phenomena in fertiliza- tion— namely, the stimulus to growth and the mingling of ances- tral qualities. Strasburger 43 regards the latter process as the essential one, and the stimulus to growth as only providing the conditions which make it possible to obtain the advantages resulting from a mingling of ancestral plasma masses. In a later paper 59 he makes the statement that fluctuating variations do not furnish a starting-point for the formation of new species, but that it is the principal function of fertilization, through the mingling of ancestral plasma masses, to keep the species characters constant. The essence of fertilization lies in the union of organized elements. It was to insure this essentially generative fertilization that, in the course of phylogenetic devel- opment, the inability of the sexual cells to develop independ- ently became more and more marked. The term generative fertilization is used in contrast with vegetative fertilization, which is merely a stimulus to growth. Hence Strasburger re- gards the fusion of the male nucleus with the polar nuclei as merely vegetative fertilization, and lacking the essential feature of a sexual fusion. It is worthy of note that Ernst 61 finds in Paris quadrifolia and Trillium grandiflorum a striking differ- ence between generative and vegetative fertilization, the fusion of the male nucleus with the egg-nucleus being complete, so that a typical resting nucleus is formed ; while the polar nuclei begin to form spirems even before the male nucleus arrives, and in the group of three nuclei — the two polar nuclei and the male 160 MORPHOLOGY OF AXGIOSPERMS nucleus — three spirems are distinguishable, a case observed also in this laboratory by Miss Laetitia Snow in Lilium philadel- phicum. In such cases it is very probable that there is no union of the chromatin (Fig. 73), and it is known that in Pin us there is no fusion of the chromatin of the two sex nuclei before the 'gss,, 7H , w& feiSp / vv"* A B Fig. 73. — Paris quadri/olia. A, two polar nuclei in spirem stage; male nucleus (m) shown just above; £, the two nuclei and male nucleus in spirem stage; x 1^50. — After Ernst.61 :;-. binucleate stage of the proembryo is reached, and the majority of published figures show this condition. However, Land describes a complete fusion of the polar nuclei of Silphium before the union with the second male nucleus. On the whole, it is to be regretted that the phrase " double fertilization " has been applied to this phenomenon, since it is far from established that it is to be regarded as real fertiliza- tion. During this uncertainty it would seem convenient and sufficient to speak of it as " triple fusion." It is also mislead- ing to speak of the vermiform male nuclei as " antherozoids " or " spermatozoids " in the sense that they are something mor- phologically distinct from the other male nuclei of Angiosperms. Whatever the ordinary male nuclei of Angiosperms may be these vermiform nuclei are. Probably male cells are always organ- ized, and we consider thern as morphologically sperm mother- cells ; but it is also probable that only the male nuclei become FERTILIZATION 161 vermiform and take part in fusion. In preparations of Lilium we have seen a vermiform nucleus still enclosed bv the cvto- plasm of the male cell. It would be strange morphology to base the definition of a sperm-cell upon its form or power of inde- pendent motion. LITEEATURE CITED 1. Schacht, H. Entwickelungsgeschichte des Pflanzenembryon. Amsterdam, pp. 234. pis. 26. 1850. 2. Hofmeister, W. Vergleichende Untersuchungen der Keiniung, Entfaltung und Fruchtbildung hoherer Kryptogamen. Leipzig. 1851. 3. . Neue Beitrage zur Kenntniss der Embryobildung der Phanerogamen. Abhandl. Konigl. Sachs. Gesell. Wiss. 6 : 533- 672. pis. 1-27. 1859. 4. Strasburger, E. Befruchtung und Zelltheilung. Jena. 1877. 5. . Die Gymnospermen und die Angiospermen. Jena. 1879. 6. Elfving, F. Studien iiber die Pollenkorner der Angiospermen. Jenaisch. Zeitsch. Naturwiss. 13: 1-28. 1879; Quart. Jour. Micr. Sci. 20: 19-35. 1880. 7. Ward, H. Marshall. On the Embryo-sac and Development of Gymnadenia conopsea. Quart. Jour. Micr. Sci. 20: 1-18. pZs. 1-3. 1880. 8. Strasburger, E. Neue Untersuchungen iiber den Befruchtungs- vorgang bei den Phanerogamen. Jena. 1884. 9. Barnes, C. R. The Process of Fertilization in Campanula amer- icana. Bot. Gazette 10: 349-354. pi. 10. 1885. 10. Goebel, K. Outlines of Classification and Special Morphology. English translation. 1887. 11. Oliver, F. W. On Sarcodes sanguinea. Annals of Botany 4: 303-326. pis. 17-21. 1890. 12. Treub, M. Sur les Casuarinees et leur place dans le systeme naturel. Ann. Jard. Bot. Buitenzorg 10: 145-231. pis. 12-32. 1891. 13. Guignard, L. Nouvelles etudes sur la fecondation. Ann. Sci. Xat. Bot. VII. 14; 163-296. pis. ri-18. 1891. 14. Nawaschin, S. Zur Embryobildung der Birke. Bull. Acad. Imp. Sci. St. Petersbourg 13: 345-348. 1892; reviewed in Bot. Cen- tralbl. 54: 237. 1893. 15. Benson. Margaret. Contribution to the Embryology of the Amentiferae. I. Trans. Linn. Soc. Bot. London 3 : 409-424. j>/s. 67-72. 1894. 16. Humphrey, J. E. Nucleolen und Centrosomen. Ber. Deutsch. Bot. Gesell. 12: 108-117. pi. 6. 1894. lt)2 MORPHOLOGY OF AXGIOSPERMS 17. D'Hlbert, E. Recherches sur le sac embryonnaire des plantes grasses. Ann. Sci. Nat. Bot. VIII. 2: 37-128. pis. 1-3 > fiy$- 6<8 1896. 18. Strasburger, E. Ka^okinetische Probleme. Jalirb. Wiss. Bot. 28: 151-204. pis. 2-3. 1895. 19. Campbell, D. H. The Structure and Development of the Mosses and Ferns. London and New York. 1S95. 20. Nawaschin, S. Ein neues Beispiel der Chalazogamie. Bot. Cen- tralbl. 63: 353-357. 1895. 21. Mottier, D. M. Contributions to the Embryology of the Ranun- culaceae. Bot. Gazette 20: 241-248. 296-304. pis. 17-20. 1895. 22. Schaffner, J. H. The Embryo-sac of Alisma Plantago. Bot. Gazette 21 : 123-132. pis. 9-10. 1896. 23. Chamberlain, C. J. Contribution to the Life History of Salix. Bot. Gazette 23 : 147-179. jils. 12-18. 1897. 24. Schaffner, J. H. Contribution to the Life History of Sagittaria variabilis. Bot. Gazette 23 : 252-273. pis. 20-26. 1897. 25. Coulter, J. M. Contribution to the Life History of Lilium phil- adelphicum. Bot. Gazette 23: 412-422. j)ls. 32-3^. 1897. 26. Mottier, D. M. Ueber das Verhalten der Kerne bei der Entwick- elung des Embryosacks und die Vorgange bei der Befruchtung. Jahrb. Wiss. Bot. 31 : 125-158. pis. 2-3. 1898. 27. Coulter, J. M. Contribution to the Life History of Ranunculus. Bot. Gazette 25 : 73-88. pis. 1,-1. 1898. 28. Smith, R. W. A Contribution to the Life History of the Ponte- deriaceae. Bot. Gazette 25 : 324-337. pis. 19-20. 1898. 29. Lyon, Florence M. A Contribution to the Life History of Eu- phorbia corollata. Bot. Gazette 25: 418-426. pis. 22-24. 1898. 30. Nawaschin, S. Ueber das Verhalten des Pollenschlauches bei der Ulme. Bull. Acad. Imp. Sci. St. Petersbourg 8: 345-357. pi. 1. 1898 ; reviewed in Bot. Centralbl. 77 ; 26-30. 1899. 31. Zinger, N. Beitrage zur Kenntniss der weiblichen Bliithen und Inftorescenzen bei Cannabineen. Flora 85: 189-253. pis. 6-10. 1898. 32. Guignard, L. Sur les antherozoides et la double copulation sexu- elle chez les vegetaux angiospermes. Compt. Rend. 128: 864- 871. figs. 19. 1899; Rev. Gen. Bot. 11: 129-135. pi. 1. 1899; also Les decouvertes recentes sur la feeondation chez les vegetaux angiospermes. Volume Jubilaire de la Societe de Biologique. Paris. 1899. 33. Nawaschin, S. Resultate einer Revision der Befruchtungsvor- gange bei Lilium Martagon und Fritillaria tenella. Bull. Acad. Imp. Sci. St. Petersbourg 9: 377-382. 1898; reviewed in Bot. Cen- tralbl. 78: 241-245. 1899. 34. . Neue Beobachtungen iiber Befruchtung bei Fritillaria tenella und Lilium Martagon. Bot. Centralbl. 77 : 62. 1899. FERTILIZATION 163 35. Merrell. W. D. A Contribution to the Life History of Silphium. Bot. Gazette 29 '. 99-133. pis. 3-10. 1900. 36. Conrad, A. H. A Contribution to the Life History of Quercus. Bot. Gazette 29 : 408-418. })ls. 28-29. 1900. 37. JOHNSON, D. S. On the Endosperm and Embryo of Peperomia pellucida. Bot. Gazette 30: 1-11. pi. 1. 1900. 38. Land, W. J. G. Double Fertilization in Compositae. Bot. Gazette 30: 252-260. pis. 15-10. 1900. 39. GrUlGNARD, L. L'appareil sexuel et la double feeondation dans les Tulipes. Ann. Sci. Nat. Bot. VIII. 11 : 365-387. pis. 9-11. 1900. 40. Nawaschix, S. Ueber die Befruchtungsvorgange bei einigen Dieo- tyledoneen. Ber. Deutsch. Bot. Gesell. 18: 221-230. pi. 9. 1900. 41. Pirotta, R. and Loxgo, B. Basigamia, mesogamia, acrogamia. Atti R Accad. Lineei V. 9: 296-298. 1900; Bot. Centralbl. 86: 93. 1901. 42. Guignard, L. Nouvelles l'echerches sur la double feeondation ehez les Phanerogames angiospermes. Compt. Rend. 131 : 153- 160. 1900. 43. Strasburger. E. Einige Bernerkungen zur Frage nach der " dop- pelten Befruehtung " bei den Angiospermen. Bot. Zeit. 58 : 293- 316. 1900. 44. Thomas, Ethel M. On the Presence of Vermiform Nuclei in a Dicotyledon. Annals of Botany 14: 31S-319. 1900. 45. . Double Fertilization in a Dicotyledon — Caltha palustris. Annals of Botany 14: 527-535. pi SO. 1900. 46. Sargant, Ethel. Recent Work on the Results of Fertilization in Angiosperms. Annals of Botany 14: 689-712. 1900. 47. Bernard, C. H. Recherches sur les spheres attractives chez Lilium candidum, etc. Jour. Botanique 14: 118-124. 177-18S, 206-212. pis. kr-5. 1900. 48. Guignard, L. La double feeondation dans le mais. Jour. Botan- ique 15: 37-50. 1901. 49. LONGO, B. La mesogamia nella commune zucca (Cucurbita Pepo Linn.). Rend. R. Accad. Lineei 10: 168-172. 1901. 50. Mlrbeck. 8. Ueber das Verhalten des Pollenschlauches bei Al- chemilla arvensis und das \Yesen der Chalazogamie. Lunds Univ. Arsskrift 36: pp. 19. pis. 2. 1901. 51. Schaffner. J. H. A Contribution to the Life History and Cy- tology of Erythronium. Bot. Gazette 31: 369-387. pis. Jt-9. 1901. 52. GUIGNARD, L. Sur la double feeondation chez les Solanees et les Gentianees. Compt. Rend. 133: 126S-1272. 1901. 53. . Double feeondation chez les Ranonculacees. Jour. Botan- ique 15: 394-408. figs. 16. 1901. 54. Shibata. K. Die Doppelbefruchtung bei Monotropa uniflora L. Flora 90: 61-66. 1902. 164 MORPHOLOGY OF AXGIOSPERMS 55. Karsten, G. Ueber die Entwickelung der weiblichen Bliithen bei einigen Juglandaceen. Flora 90: 316-333. pi. 12. 1902. 56. Guignard, L. La double fecondation chez les Solanees. Jour. Botanique 16 : 145-167. Jigs. 45. 1902. 57. Hall. J. G. An Embryological Study of Limnocharis emargi- nata. Bot. Gazette 33: 214-219. pi. 9. 1902. 58. Ikeda, T. Studies in the Physiological Functions of Antipodals and Related Phenomena of Fertilization in Liliaceae. 1. Tricyr- tis hirta. Bull. Coll. Agric. Imp. Univ. Tokyo 5: 41-72. pis. 3-6. 1902. 59. Strasburger, E. Em Beitrag zur Kenntniss von CeratophyUum submersum und phylogenetische Erorterungen. Jahrb. "Wiss. Bot. 37: 477-526. pis. 9-11. 1902. 60. Frye, T. C. A Morphological Study of Certain Asclepiadaceae. Bot. Gazette 34 : 389-413. pis. 13-15. 1902. 61. Ernst, A. Chromosomenreduction, Entwickelung des Embryo- sackes und Befruchtung bei Paris quadrifolia L. und Trillium- grandiflorum Salisb. Flora 91: 1-46. pis. 1-6. 1902. 62. Shoemaker, D. N. Notes on the Development of Hamamelis vir- giniana L. Johns Hopkins Univ. Circ. 21 : 86-87. 1902. 63. Juel, H. O. Zur Entwicklungsgeschichte des Samens von Cyno- morium. Beih. Bot. Centralbl. 13 : 194-202. figs. 5. 1902. 64. Guignard, L. La double Fecondation chez les Cruciferes. Jour. Botanique 16 : 361-368. figs. 20. 1902. 65. Shibata, K. Experimentelle Studien iiber die Entwickelung des Endosperms bei Monotropa. (Vorlaufige Mitteilung) Biol. Cen- tralbl. 22: 705-714. 1902. 66. Billings, F. H. Chalazogamy in Carya olivaeformis. Bot. Gazette 35: 134-135. 1903. 67. Wylie, R. B. A Morphological Study of Elodea canadensis. To- be published in Bot. Gazette 36: 1903. 68. Frye, T. C. The Embryo-sac of Casuarina stricta. To be pub- lished in Bot. Gazette Z5 : 1903. CHAPTEK VIII THE ENDOSPERM The endosperm of Gymnosperms seems to be clearly the vegetative tissue of the female gametophyte, but the morpho- logical nature of the endosperm of Angiosperms (Fig. 74) is not so clear. The c:er- mination of the megaspore begins, as in Gymnosperms, -with free and simul- taneous nuclear di- vision. In Gymno- sperms this con- tinues for some time and is re- placed by cell-for- mation, giving rise to an extensive tis- sue bearing arche- gonia, while in An- giosperms usually only eight free nu- clei are formed be- fore an egg is organ- ized and fertiliza- tion takes place. In both cases endo- Fio. 74. — Two modes of initiating the formation of endo- sperm. A, J'aias major, illustrating free nuclear divi- sion ; there are four free nuclei belonging to the endo- sperm, the lower free nucleus being that of the upper antipodal; x 175. 2?, Datura laevis, nuclear division followed immediately by formation of wall; x 225.— After Guionard.«. *« sperm is formed after fertilization; but in Gymnosperms it is a continuation oi cell division, while in Angiosperms it usually begins with nuclear fusion followed bv simultaneous and often free nuclear 165 166 MORPHOLOGY OF ANGIOSPERMS division. This nuclear fusion is one of the most striking fea- tures of the Angiosperms as contrasted with Gymnosperms, and especially since the discovery of so-called " double-fertilization " the morphological character of the endosperm of Angiosperms is in question. For this reason, we have preferred to discuss it apart from the gametophytic structures concerning which there is no question. As has been said, the endosperm of Angiosperms is usually derived from a fusion nucleus, the constituent members being the micropylar polar nucleus, sister to the egg, and the antipo- dal polar nucleus. If the current homologies are true, this fusion is that of a female and a vegetative nucleus. In many cases a male nucleus also joins in the structure of the primary endosperm nucleus, which is then the result of a triple fusion (Figs. 36, II, and 71-73). How far this male nucleus is an es- sential factor in the formation of the endosperm of Angiosperms is at present unknown, but the rapidly increasing number of plants in which triple fusion has been observed leads to the belief that it may be of general occurrence. It should also bo remembered that in Peperomia pellucida (Johnson31) (Fig. 38)_tbe primary endosperm nucleus is the result of the fusion of no less than eight of the sixteen free nuclei of the embryo- sac; and that in Gunnera (Schnegg47) (Fig. 39) the same sort of multiple fusion occurs. The fusion-nucleus, therefore, may be made up of a variable number of constituents of various morphological character, and hence the significance of the fusion and the nature of the resulting tissue are peculiarly diffi- cult to interpret. While the fusion of these nuclei seems to result in what lias been called a growth-stimulus, endosperm is sometimes formed without any antecedent fusion. For example, in Balanophora (Treub,16 Lotsy26) the polar nuclei do not fuse, but divide independently, the embryo-sac becoming filled with endosperm tissue; and in Helosis (Chodat and Bernard33) after the first division of the nucleus of the megaspore the chalazal nucleus disintegrates so that antipodal cells, and hence an antipodal polar nucleus, are not formed, the endosperm being derived en- tirely from the micropylar polar nucleus. In Antennaria alpi- na duel — found that the polar nuclei do not fuse, although they behave normally in A. clioica, as the same investigator35 has THE ENDOSPERM 1C7 observed. In Lemna Caldwell 24 states that often the polar nuclei do not fuse, in which case he observed that the micro- pylar polar produced some free endosperm nuclei, and probably the antipodal one also. In Limnocharis, one of the Alismaceae, there is also no fusion (Hall 50), since no antipodal polar nucleus is formed, and all the endosperm, which eventually fills the sac, is derived from the micropylar polar nucleus. In Casuarina, according to Treub,12 there are no antipodals or polar nuclei, and the endosperm is formed before fertilization and independently of any fusion (Fig. 67, B). It should be stated, how- ever, that in a recent studv of Casu- arina by Frye 59 abundant endo- sperm was found before the first division of the egg, but probably not before fertilization. For exam- ple, the same investigator 56 found in Asclepias sixteen and thirty-two endosperm nuclei before the first division of the egg, but not before fertilization (Fig. 75). In Piper and Heche ria the development of £' endosperm before the first division of the fertilized egg is even more extensive. Johnson 55 represents twenty-two endosperm cells in a single section of Piper (Fig. 76) and the e^c; has not vet divided. Not a little confusion has arisen bv assuming that fertilization and the first division of the egg are practically simultaneous. In any event, the formation of endo- sperm without antecedent fusion is clear enough in some cases, and indicates that while fusion usually serves to stimulate growth and coll division it is not an absolute prerequisite. In certain orchids Xawaschin 36 states that the polar nuclei do not fuse, but in this case no endosperm is formed. In this connection the experiments of Shibata 57 on Mono- 12 Fio. 75. — Asclepias Corn uti. Large development of endosperm bet'oie di%'ision of fertilized egg : a, an- tipodals ; e, egg : .«, synergids ; x 750.— After Frye." ^em cm Fig. 76. — -4, Piper medium, showing extensive development of endosperm before first division of fertilized egg : x 17"' : Jl-I'. Peperomia pellucida : B. longitudinal section of ripe seed, showing the small embryo, scanty endosperm, and abundant perisperm ; x 55 ; C. terminal portion of a similar section at an early stage of germination ; x 175; i), longitudinal section of a germinating seed, showing the endosperm pro- truding with the embryo; x 55: a, antipodals ; c, cotyledons ; syncrgid; st, stigma; t. tapetar cells. — After Johnson." 169 THE ENDOSPERM 169 tropa it ni flora are of interest. In this case the polar nuclei may fuse in the absence of pollination, but the fusion may be hastened or regulated by pollination. In normal cases fusion of polar nuclei occurs about five days after pollination, but when pollina- tion is prevented the interval may be prolonged to ten days or even longer. Development of the endosperm was also induced experimentally in the absence of fertilization. When pollination is prevented, many of the ovules die within two or three weeks, but in others the sac enlarges and becomes filled with endosperm. This development of en- dosperm was observed in from three to five per cent of the ovules, but at a tem- perature of 28° C, or by using osmotic solutions, en- dosperm was developed by from six to twelve per cent of the seeds. If a fusion nucleus is formed, as is certainly gen- erally the case, it usually begins to divide before the fertilized egg and with much greater rapidity. After fertilization, the egg usually seems to rest for a period while free endo- sperm nuclei are being Fig. 77. — Erigeron philadelphicus. Longitudinal sections of embryo-sac after fertilization. A\ fertilized egg dividing before primary endo- sperm nucleus; J}, primary endosperm nu- cleus dividing before egg; x 550. — After Land32 formed. For example, among the Ranunculaceae (Guignard43) and in As- rlepias (Frye 56) free en- dosperm nuclei are scattered through the sac before the egg divides. But there is every gradation from an approximately simultaneous division of primary endosperm nucleus and fer- tilized egg, as usually in Sagittaria (Schaffner 18), Lilium ( Coulter ,;' I, Nelumbo ' Lyon 4n ), Sorrodes (Oliver "), Senecio (Mottier15), and Erigeron (Land32) (Fig. 77), in which last case sometimes the egg and sometimes the primary endosperm nucleus divides first, to a sac almost or even completely filled with 170 MORPHOLOGY OP ANGIOSPERMS endosperm before the fertilized egg segments, as in Gonyanthes Candida (Treub7), Heclceria (Johnson55), the Stylidaceae (Burns28), and Aphyllon uniflorum (Smith46). Even though the primary endosperm nucleus and the fertilized egg divide simultaneously, the much more rapid divisions of the former result in numerous free endosperm nuclei before the first few segmentations of the egg have been completed. In the cases just cited, in which the segmentation of the primary endosperm nucleus precedes that of the fertilized egg, the division does not begin until after fertilization, and proba- bly this is true in the majority of plants. As a consequence, the impression is current that the act of fertilization is an essential stimulus to the division of the primary endosperm nucleus ; and there seems to be no clear evidence to the contrary when fertilization occurs, unless it be the case of Ranunculus, as reported by Coulter,20 in which free endosperm nuclei were sometimes observed scattered through the embrvo-sac before the entrance of the pollen-tube. To this same category belong those cases of habitual failure of fertilization in which endosperm formation may occur, as in the Balanophoraceae, Antennaria alpina ( Juel 22), Thalictrum purpurase'ens (Overton51), Eich- hornia crassipes (Smith21), etc. It seems to be very rare for the fertilized egg to divide before the primary endosperm nu- cleus, but in Naias major, in which triple fusion occurs, Guig- nard 42 has observed that the fertilized egg divides immediately, and has figured a two-celled embryo by the side of a primary endosperm nucleus in the spirem stage. It is important to note also that in this same species Guignard observed that the male nucleus may fuse with the persistent synergid instead of with the primary endosperm nucleus, in which case there is no endosperm, but a second embryo (Fig. 103). Many cases of two embryos lying side by side with an " unfertilized " primary endosperm nucleus between them were observed. Recently Wylie 60 has observed that in Elodca also the fertilized egg divides before the primary endosperm nucleus. It is evident that the beginning of endosperm formation does not depend absolutely upon any of the causes usually assigned ; and that while it is in general approximately coinci- dent with the segmentation of the fertilized egg, this is merely a coincidence, for it may be independent of fertilization and THE ENDOSPERM 171 even of fusion. Ordinarily it must be dependent upon polar fusion, and in some cases upon triple fusion, as indicated by the behavior in Naias cited above ; but in the failure of these, other conditions may cause nuclear division and the formation of endosperm. While in the majority of plants the endosperm may be re- garded as fully developed, either to remain as a permanent tissue of the seed or to be more or less resorbed by the growing embryo, there are certain plants in which it is abortive or even suppressed. It consists of only a few scattered nuclei, or at most of a parietal layer of free nuclei, in Naiadaceae, most Alis- maceae, Juncagineae, and Hydrocharitaceae, all of which belong to the Helobiales among Monocotyledons. The tendency of the endosperm to become abortive in this particular alliance is evi- dently very strong, although, as Hall50 has shown in Limno- charis, the endosperm may finally develop and become packed about the embryo. With the exception of the Helobiales, disap- pearance of the endosperm seems to be very rare, having been reported in Tropaeolum and Trapa; and among the Orchida- ceae the endosperm seems to be entirely suppressed, the polar nuclei, as a rule, neither fusing nor dividing. Humphrey 17 has called attention to what he calls a pro- gressive series in the development of the endosperm among the Scitamineae, but which seems to be best interpreted as a retro- gressive series. In the Musaceae an abundant starch-bearing endosperm either fills the sac (Ileliconia) or nearly so (Stre- litzia), the peripheral cells often forming an aleurone layer; in Zingiberaceae (Costus) the endosperm is several layers thick in the lower part of the sac and only aleurone-bearing ; in Can- naceae (C. indica) the endosperm is a single aleurone-bearing layer lining the sac; while in Marantaceae (Thalia dealbata) the endosperm is probably not represented at all in the mature seed. Strasburger 4 has called attention to the two general meth- ods of endosperm formation among Angiosperms. In the ma- jority of plants observed it begins with free nuclear division ; but in many cases, chiefly among Dicotyledons, the first division of the primary endosperm nucleus is accompanied by a wall dividing the sac into two chambers (Fig. 74). While these two methods of initiating endosperm formation are quite dif- 172 MORPHOLOGY OF ANGIOSPERMS ferent, the subsequent stages of endosperm development result in all kinds of intergrading conditions, as will be shown later. Even when the endosperm begins with free nuclear division, a rudimentary plate often appears, suggesting derivation from an endosperm in which nuclear division was followed by cell- formation. The history of the development of endosperm initiated by free nuclear division is nearly identical, in most cases, with the history of the female gametophyte in Gymnosperms, modi- fied, of course, by the presence of a developing embryo. It is an interesting fact, also, that the early stages in the develop- ment of the endosperm bear a striking resemblance to early stages in the development of the embryo of Cyeadales and some other Gymnosperms. There is the same simultaneous nuclear division, often the parietal placing, and later the appearance of cell walls. The primary endosperm nucleus, usually in contact with the egg, or nearly so, divides, and subsequent divisions follow with great rapidity, Guignard 41 remarking that in Zea he was unable to follow the course of division, and other observers call- ing attention not only to the great rapidity with which one set of divisions is followed bv another, but also to their simultane- ous character. A common form of statement is that at first the free nuclei remain for a time in the vicinity of the e^e, but sooner or later migrate in every direction toward the wall of the embryo-sac, where they become equally distributed and embedded in a lining cytoplasmic layer. The real fact, how- ever, is that this apparent movement of the nuclei is due to the rapid enlargement of the sac, the cytoplasm becoming more and more vacuolate and finallv occurrino- chieflv as a wall laver. By this increasing vacuolation the nuclei are naturally driven to the wall. In this parietal position free nuclear division con- tinues, until finally walls are formed and a layer of parietal cells is organized. These first walls usually " cut out " only one nucleus in each cell, but in some cases (Con/dalis cava, Staphylea pinnata, Armeria vulgaris, etc.) Strasburger 4 noted that two to four nuclei might be enclosed by a cell wall, but that they afterward fuse to form a single nucleus (Fig. 78). Tischler "'-1 has recently reexamined Corydalis cava and states that when septa THE ENDOSPERM 173 appear many nuclei are always enclosed in each cell and sub- sequently fuse. In this particular case the free nuclear divi- sions are often irregular, and of course the number of chromo- somes is exceedingly variable, a fact very common in all endo- B FIG. 78. — Advanced stages in development of endosperm. A, Reseda odorata, upper part of figure showing free nuclear division, while in lower part nuclear division is accompanied by formation of cell walls; x 860; £, Caltha palustris, showing all nuclear divisions accompanied by formation of walls, x 155; C, Corydalis cava. showing free nuclear division within cells of endosperm; D, the same, showing multinucleate endosperm ; x 860. — After Strasbvkger.4 sperm. The same phenomenon was observed by Humphrey 17 in Carina indica, in which the parietal layer of free nuclei block " containing several becomes blocked out by walls, each « 174 MORPHOLOGY OF ANGIOSPERMS nuclei that apparently fuse into one. The irregular and usu- ally large number of chromosomes found in the nuclei of endo- sperm tissue is doubtless due to " double fertilization v and other nuclear fusions. The parietal plate of cells by division gradually encroaches upon the general cavity of the embryo-sac, either filling it up compactly about the embryo, or leaving more or less of a cavity containing cell sap, which in the coconut becomes of extraordi- nary size. In many cases a fully developed endosperm is more or less displaced by the growing embryo, so that in the mature seed it may be much reduced or even obliterated. Among the Mono- cotyledons the embryo of the Gramineae is at first completely invested by endosperm, but becomes eccentric by displacing it on one side; and the embryo in some Araceae finally replaces all the endosperm ; but for the most part the Monocotyledons are characterized by retaining the endosperm in the mature seed. Among the Dicotyledons, however, it is characteristic of certain families, among the important ones being Cupuliferae, Leguminosae, Cucurbitaceae, and Compositae, for the embryo to have entirely displaced the endosperm at the maturity of the seed, the gain in size being almost entirely in the cotyledons. It must not be supposed that in all cases the formation of endo- sperm continues from the first free nuclear division to a tissue filling the embryo-sac. Illustrations could be introduced show- ing a cessation of endosperm formation at every stage. It may stop with a few free nuclei, or with the parietal placing of free nuclei, or with a parietal plate of tissue. An interesting case is that of Tricyrtis (Liliaceae), recently described by Ikeda,54 in which free endosperm nuclei are distributed through a sac full of cytoplasm, and assume very irregular and bizarre forms, the parietal position never being assumed. The second general method of endosperm formation — namely, that in which the first division of the primary endo- sperm nucleus is accompanied by a wall dividing the sac into two chambers — is found chiefly among Dicotyledons, and among them it is especially characteristic of saprophytic and parasitic forms, Cuscuta being a marked exception in that its endosperm begins with free nuclear division. Usually the wall divides the sac into two approximately equal chambers, but naturally the THE ENDOSPERM 175 relative size of the chambers depends upon the position of the dividing nucleus (Fig. 74). Among Monocotyledons, the endosperm of Sagittaria (Schaffner 1S) develops rapidly in the micropylar chamber into a -walled tissue, the endosperm nucleus of the antipodal chamber enlarging much but not dividing for a long time, when two or three nuclei may be formed, all of them increasing greatlv (Fig. 79). Practically the same thing occurs in Limno- charis (Hall50), but the nucleus of the antipodal chamber en- larges without dividing. In Ruppia rostellata (Murbeck 58) a Fig. 1$.— Sagittaria variabilis. A, two nuclei of endosperm separated by wall : a, an- tipodal, x 200 ; £. compact endosperm tissue developed from upper cell, the lower merely growing large without dividing; x 108.— After Schaffner.18 wall is formed at the first division of the endosperm nucleus, the antipodal chamber remaining small and with undividing nucleus, but a large number of free nuclei being formed in the micro- 176 MORPHOLOGY OF ANGIOSPERMS pylar chamber. In Pqtamogeton (Holferty44) the endosperm is developed only as a parietal layer of free nuclei; but all of these seem to have come from the micropylar endosperm-cell of the first division, the lower one becoming very large but not dividing, a tendency similar to that in Sagittaria and Limno- charis, but without the formation of a transverse wall in the sac. Among the Dicotyledons instances of a chambered embryo- sac are numerous. Hofmeister l has given a long list of them, and these, with others added since, are approximately as follows: Among the Arehiehlamydeae they are the Saururaceae, Loran- thaceae, Balanophoraceae, Santalaceae, Aristolochiaceae, jSTym- phaeaceae, Ceratophyllaceae, Loasaceae, a list composed in the main of primitive or saprophytic and parasitic forms. In fact, the chambered sac is distinctly lacking in the more important and characteristic groups of the Arehiehlamydeae. Among the Sympetalae, chambered sacs occur in the Pyrolaceae, Mono- tropaceae, Vacciniaceae, Hydrophyllaceae (Nemophija), Sola- naceae, Verbenaceae, Selaginaceae, Labiatae, Scrophulariaceae, Orobanchaceae, Bignoniaceae, Pedaliaceae, Acanthaceae, Plan- taginaceae, and Campanulaceae. Although most largely repre- sented among Sympetalae, it will be noted that chambered sacs occur chiefly in saprophytic or parasitic forms, and among the Personales. The phenomenon seems thus to be associated with peculiar conditions of nutrition or a certain configuration of the embryo-sac. In the case of two-chambered sacs anions' Dicotvledons, it does not seem to be common for endosperm to form in both chambers, although this is reported to be the case in Balano- phoraceae, Aristolochiaceae, Pyrolaceae, and Monotropaceae. In the majority of cases the endosperm develops only in the micropylar chamber, in connection with the embryo, as in Sau- ruraceae, Viscum (Loranthaceae), Santalaceae, Nymphaeaceae, Globuhiria (Selaginaceae), Scrophulariaceae, and Orobancha- ceae. In Saururus (Johnson34) the embryo-sac is flask-shaped, the Avail cutting off the neck from the large venter, and the en- dosperm developing only in the former. In Nymphaea and Nuphar (Cook52) the endosperm develops only in the micro- pylar chamber, while the antipodal chamber extends as an haus- torial tube to the chalazal extremity of the ovule. It is of interest to note that until Cook's work the endosperm of these genera was THE ENDOSPERM 177 said to begin with free nuclear division, followed by a wall cut- ting off the micropylar end of the sac; and the same statement in reference to Ceratophyllwn has been disproved recently by Strasburger.49 The endosperm is said to develop only in the antipodal chamber in Loranihus, Yacciniaceae, Verbenaceae, Hebenstreitia (Selaginaceae), Bignoniaceae, and Acanthaceae. In Trapella (Oliver10), a genus of the Pedaliaceae, although the sac is not chambered by a wall, the endosperm develops only in the lower two-thirds, a sort of diaphragm of thick-walled en- dosperm-cells cutting off the broad micropylar end of the sac. B C D Fig. SO. — Ceratophyllum submersum. Development of endosperm and embryo. A, first division of embryo, six cells in endosperm; x 250; B, embryo and endosperm more advanced; x 250; C-D, entire embryo seen from opposite sides, 6' showing the two cotyledons separate and D nearly united ; x 50. — After Strasburger.48 Cases are also known in which more than two chambers are formed in the embryo-sac and followed by ordinary cell-forma- tion. For example, in Ceratophyllum (Strasburger49) at the first division of the primary endosperm nucleus the sac is divided into two approximately equal chambers. The nucleus in the antipodal chamber does not divide again, but at the next division in the micropylar chamber another wall across the sac 178 MORPHOLOGY OF ANGIOSPERMS is formed, so that there are three superposed chambers, and only in the one nearest the micropyle does division proceed. As a result, a dense, small-celled tissue is formed near the embryo (Fig. 80). In Datura laevis (Guignard 48), after the first divi- sion into two chambers {Fig. 74), transverse walls are formed in each, resulting in four superposed chambers in which further division proceeds in various planes. There are also cases in which each division of an endosperm nucleus is accompanied by a transverse wall across the sac, as in Sarcodes (Oliver11), in which the mature sac is several- chambered by a series of delicate transverse walls. The same is doubtless true of Pistia, whose narrow sac contains a row of broad discoid endosperm-cells that lie like transverse chambers. One of the most exceptional cases of wall-formation, however, is that of Peperomia pellucida (Johnson 31), in which the first division of the very large primary endosperm nucleus, formed by the fusion of eight nuclei, is followed by a wall from the fertilized egg to the base of the sac, further divisions following until the sac is packed with forty or more endosperm-cells. In a recent study of Heche ria also, one of the Piperaeeae, the same investigator 55 has found the same general condition as in Peperomia, in that the endosperm is " cellular " from the first, filling the sac before the egg divides. It is worthy of note that the endosperm of Piper (Johnson 55), on the other hand, begins with free nuclear division. It is evident from these differences in closely related genera, also noted by Hofmeister 2 and Ilegel- niaier,8 that methods of endosperm formation can not indicate relationship. The mature and permanent endosperm is a tissue with no intercellular spaces, whose cells are either thin-walled, form- ing an endosperm of delicate texture, or thick-walled, resulting in a horny endosperm, as in palms, umbellifers, etc. In case the thickening of the walls becomes excessive, the endosperm is stony, as in Phytelephas, the palm whose seeds furnish the so-called " vegetable ivory." The endosperm has sometimes been observed to continue its growth after it has filled the sac. Hofmeister describes the en- dosperm of Crinum capense and some other Amarvllidaeeae as bursting the seed-coats, and oven the ovary wall, the cells devel- oping chlorophyll, and the tissue remaining succulent and form- THE ENDOSPERM 179 ing intercellular spaces. A similar extensive growth and escape of the endosperm is reported to occur during the germi- nation of the seeds of Ricinus. In the germination of the seeds of certain Piperaceae (Peperomia and Heckeria) Johnson 53, 55 has described the endosperm as bursting out of the seed-coat, and continuing to jacket the embryo, which at germination is a globular undifferentiated mass of cells, until the root, hypocotyl, and cotyledons are organized. In the same papers Johnson calls attention to the fact that the endosperm of these Piperaceae is not a storage region, but digests, absorbs, and passes on food material to the embryo from the much more abundant peri- sperm, which is the real storage tissue. This restriction of the function of the endosperm Johnson 34 had already pointed out in Saururus, and suggests the probability that this same relation between endosperm and perisperm obtains in all seeds with abundant perisperm as in Polygonaceae, Chenopodiaceae, Phy- tolaccaceae, Caryophyllaceae, etc. The following quotation 55 will serve to make plain the author's point of view: " Observations thus far made lead me to believe that in the peri- sperm-contain ing- seeds mentioned the embryo sporophyte of the second generation is never nourished by the parent sporophyte directly, but always through the intermediate gametophyte. In general, then, we find that the food substance supplied to the embryo by the nucellus may pass through the endosperm and be stored in the embryo during the ripening of the seed, as in Cucurbita and Phaseolus ; or, secondly, the food may be stopped in transit between the nucellus and the embryo and stored in the endosperm, there to be held during the resting period of the seed and delivered over to the embryo only at the time of sprout- ing, as in Ricinus, Zea, and apparently all Gymnosperms ; or. finally, the food supply for the developing embryo may be stored in the nucel- lus itself until the time of germination, when it is passed on to the embryo through the endosperm, as in Saururus, Peperomia, Phyto- lacca, Carina, and others." The phenomenon of xenia has a direct bearing upon any discussion of the endosperm. The name was applied by Focke,5 in 1881, to the direct effect of pollen on seeds and fruits out- side of the embryo, as shown in hybrids. The case of peas has long been cited, but Giltay 14 has shown that the effects referred to occur in the cotyledons, and therefore can not be considered as xenia. So far as definitely known, the effect of foreign pollen outside of the embryo is observed only in the endosperm, 180 MORPHOLOGY OF ANGIOSPERMS as first pointed out by Komicke,3 and this has been most clearly established in the crossing of races of corn. It also appears that this influence of foreign pollen extends only to the color of the endosperm and the chemical composition of the reserve materials, the size and form of the kernels remaining un- changed, as stated by Correns.25 For example, if white or yel- low corn be crossed with pollen from a red corn, many of the resulting? kernels will be red or variously mottled ; or if sweet corn, with its wrinkled and sugary endosperm, be crossed with pollen from dent or flint corn, the result is smooth kernels with starchy endosperm. The possibility of such a direct effect of pollen was for a long time questioned, and the phenomenon remained inexpli- cable. With the discovery of " double fertilization " or triple fusion by Nawaschin 23 in 1898, the explanation of xenia oc- curred simultaneously and independently to Correns,25 De Vries,27 and Webber,40 the paper of the last investigator being a very complete resume and discussion of the subject based upon his own extensive experimental work. To claim that the phe- nomenon of xenia, as observed in corn, is due to the fusion of one of the male nuclei with the primary endosperm nucleus was an assumption, although an irresistible one, until such fusion was demonstrated by Guignard 41 in 1901. It has been proved repeatedly that when xenia occurs the embryo is a hybrid, so that we have in xenia not only a hybrid endosperm, but a gross demonstration of the occurrence and effect of the triple fusion, and also an indication of the sort of characters that can be brought into a structure by a male nucleus. In many cases of xenia following the crossing of races of different colors, the kernels are not of uniform color, but are parti-colored or variously mottled. The ingenious explanation suggested by Webber is that the male nucleus has failed to unite with the fusion-nucleus and may be able to divide independ- ently. If so, there would result two cell-races of different characters that might be variously arranged with reference to one another in the endosperm. It is entirely conceivable that under favorable conditions of nutrition a'nd physical environ- ment an independent male nucleus may begin divisions, espe- cially as this has been observed in the case of certain animals; but it seems more probable that the independent appearance of THE ENDOSPERM 181 these racial characters is due to the incompleteness of the trip fusion, since it is well known that division of the primary e^\u- sperni nucleus often begins before the constituent nuclei nave lost their identity. In fact, Webber calls attention to the begin- ning of division before complete fusion in the case of the eggs of certain animals, and the same is true of the sexual fusion- nucleus of some Gymnosperms. An alternative hypothesis sug- gested by Webber is that the male nucleus may fuse with one of the polar nuclei, the other remaining independent and dividing. These hypotheses are valuable in suggesting investigation as to whether the male nucleus ever divides independently in the em- bryo-sac, or whether it may unite with one polar nucleus, the other dividing independently. It remains to consider the morphological character of the endosperm of Angiosperms. In view of the details as to its origin and behavior given above, it is evident that it is a struc- ture peculiarly difficult to interpret. The view has long been held, dating from Hofmeister, that the endosperm is belated vegetative tissue of the female gametophyte, stimulated in a general way to develop by the act of fertilization, and in every way the morphological equivalent of the structure bearing the same name among Gymnosperms. Strasburger 37 has suggested that this postponement of the formation of endosperm is of advantage in avoiding the waste that would follow its formation and separation from the parent plant with every unfertilized ovule. Of course the serious difficulty in this view of the nature of the endosperm was that it offered no historical explanation of the fusion of the polar nuclei. It could only claim that fusions of vegetative nuclei, evidently resulting in growth- stimulus, are by no means unknown., and in fact occur in the endosperm itself. This view does not appear to have been seriouslv disturbed by the claim of Le Monnier9 in 1887, that the fusion of the polar nuclei is a sexual process, and that there- fore the endosperm is a second embryo modified to serve as food tissue. With the discovery of the fact that, at least in many cases, a male nucleus enters into the organization of the primary endo- sperm nucleus', the old view has been seriously menaced. The commonly used phrases "double fertilization" and "double fecundation '; indicate general consent to the view that this 182 MORPHOLOGY OF ANGIOSPERMS f the male nucleus is a case of true fertilization, the infer- ence ^eing that the endosperm is a second embryo or sporophyte, as Le Monnier had suggested. Strasburger 3T in discussing the whole subject concludes that the triple fusion is not real fertilization. Of course in such a discussion much depends upon the definition of fertilization. Strasburger distinguishes between " generative fertilization " and " vegetative fertilization," the former being a definite union of parental qualities and resulting in an embryo, the latter a fusion resulting merely in a growth-stimulus. He thinks that the endosperm is historically a gametophyte, and that the fusion which initiates it has no origin in an act of fertilization. Later, Miss Sargant 38 published an admirable resume of the subject, together with a clear statement of the problems involved and certain suggestions by way of interpretation. She very justly states that if the endosperm " arose from a belated formation of prothallus, we must trace the origin of the triple nuclear fusion which precedes its development " ; and if it is a modified embrvo " we have to account for the interference of the lower polar nucleus with the act of fertilization, and for the subsequent development of a body unlike a normal embryo." Her suggested interpretation of the phenomenon is that the fusion of the male nucleus with the micropylar polar nucleus, an undoubted female nucleus, both containing the reduced number of chromosomes, is a typical sexual union ; but that the antip- odal polar nucleus, with its vegetative character, and indefi- nite and usuallv increased number of chromosomes, is a disturb- ing factor, and the result is not a normal embryo but a small and short-lived mass of tissue. She aptly cites the experiments of Boveri Vi with sea-urchins, in forcing more than one sperm- nucleus to unite with a single egg-nucleus and producing mon- strous larval structures. " The presence of the third nucleus, therefore, with its redundant chromosomes, serves to secure the degeneracy of the resulting tissue." This means, of course, that the endosperm is a degenerate embryo, and that the triple fusion is a true sexual union whose normal result has been interfered with by the presence of a non-sexual nucleus in the combination. It is impossible to solve such a problem by a discussion of THE ENDOSPERM 183 the data we possess. The phylogeny of the endosperm must be traced, and the place of the polar fusion and of the triple fusion in its history determined before opinions cease to differ as to its morphological character. In view of such facts as we have, however, we are inclined to hold with Strasburger that the endosperm of Angiosperms is a gametophytic structure, and that the polar fusion and the triple fusion are interpolations in its history that do not change its essential character. The fact that endosperm sometimes forms before fertilization indi- cates that the triple fusion is not an essential prerequisite ; the fact that endosperm forms without the polar fusion points at least to the conclusion that it was once developed without it ; the indifference of the male nucleus as to which polar nucleus it fuses with (Lilium, Asclepias) does not show the selective attraction connected with sex-fusion ; and the further fact that when an undoubted fertilization occurs, whether of egg, of syn- ergid, or of upper polar nucleus, an embryo is the result, indi- cates that the presence of the male nucleus in triple fusion is of subsidiary rather than of dominating importance. That the fusing male nucleus does introduce parental characters that manifest themselves in the endosperm is proved by the phenom- enon of xenia, but this does not seem necessarily to prove the sporophytic character of the endosperm. In fact, the develop- ment and structure of the endosperm of Angiosperms is so much like that of Gymnosperms that it seems easier to regard the various fusions as merely resulting in a stimulus to growth than to imagine a degenerate embryo assuming this particular de- velopment and structure. Of course one might go to the ex- treme, and regard the endosperm as neither gametophyte nor sporophyte, but as a composite tissue involving both, but this hardlv seems to be necessary. LITERATURE CITED 1. Hofmeister. W. Neuere Beobachtungen iiber Embryobildung der Phanerogamen. Jalirb. Wiss. Bot. 1 : 82-188. pis. 7-10. 1858. 2. . Xeue Beitriige zur Kenntniss der Embryobildung der Phanerogamen. Abhandl. Konigl. Sachs. Gesell. Wiss. 6: 533- 672. pZs. 1-27. 1859. 3. Korxicke, F. Vorlaufige Mittheilungen iiber den Mais. Sitz- ungsb. Niederrh. Gesell. Nat. Heilk. Bonn. 1872. 4. Strasburger, E. Zellbildungund Zelltheilung. Ed. 3. Jena. 1880. 13 184 MORPHOLOGY OF ANGIOSPERMS 5. Focke, W. O. Die Pflauzen-Mischlinge. Berlin. 1881. 6. Guignard, L. Recherches d'embryogenie vegetale comparee. I. Legumineuses. Ann. Sci. Nat. Bot. VI. 12: 5-1(56. j>Zs. 1-8. 1881. 7. Treub, M. Notes sur l'embryon, le sac embryonnaire, et l'ovule. Ill and IV. Ann. Jard. Bot. Buitenzorg 3 : 120-128. pis. 18-19. 1883. 8. Hegelmaier. F. Untersuchungen iiber die Morphologie des Di- kotyledonen-Endospei-rns. Nova Acta Leopoldina 49: 1-104 pis. 5. 1885 ; reviewed in Bot. Centralbl. 10: 302-304. 1886. 9. Le Monxier, G. Sur la valeur morphologique de l'albumen chez les Angiospermes. Jour. Botanique 1 : 140-142. 1887. 10. Oliver, F. W. On the Structure, Development, and Affinities of Trapella, a New Genus of Pedalineae. Annals of Botany 2: 75- 115. jjls. 5-9. 1888. 11. . On Sarcodes sanguined. Annals of Botany 4: 303-326. pis. 17-21. 1890. 12. Treub, M. Sur les Casuariuees et leur place dans le systeme natu- re! Ann. Jard. Bot. Buitenzorg 10: 145-231. pis. 12-32. 1891. 13. Boveri, Th. Befruchtung. Anat. Hefte Ergebnisse 1: 386-485. figs. 15. 1891. 14. Giltay, E. Ueber den directen Einfluss des Pollens auf Frucht- und Samenbildung. Jahrb. Wiss. Bot. 25 : 489-509. pi. S3. 1893. 15. Mottier, D. M. On the Embryo-Sac and Embryo of Senecio aureus L. Bot. Gazette 18 : 245-253. j^s- 27-29. 1893. 16. Treub, M. L'organe femelle et Tapogamie du Balanophora elon- gata. Ann. Jard. Bot. Buitenzorg 15: 1-22. pis. 1-8. 1898. 17. Humphrey, J. E. The Development of the Seed in Scitamineae. Annals of Botany 10: 1-40. pis. 1-4. 1896. 18. Schaffner, J. H. Contribution to the Life History of Sagittaria variabilis. Bot. Gazette 23 : 252-273. pis. 20-26. 1897. 19. Coulter, J. M. Contribution to the Life History of Lilium philadelphicum. Bot. Gazette 23: 412-422. pis. 32-31,. 1897. 20. . Contribution to the Life History of Ranunculus. Bot. Gazette 25 : 73-88. pis. 4-7. 1898. 21. Smith, R. W. A Contribution to the Life History of the Ponte- deriaceae. Bot. Gazette 25 : 324-337. pis. 19-20. 1898. 22. Juel, H. O. Parthenogenesis bei Antennaria alpina (L.) R. Br. Bot. Centralbl. 74: 369-372. 1898. 23. Nawaschin, S. Resultate einer Revision der Befruchtungsvor- gange bei Lilium Martagon und Fritillaria tenella. Bull. Acad. Imp. Sci. St. Petersbourg 9: 377-382. 1898; reviewed in Bot. Centralbl. 78 : 241-245. 1899. 24. Caldwell, O. W. On the Life History of Lemna minor. Bot. Gazette 27: 37-66. figs. 59. 1899. 25. Correns, C. Untersuchungen iiber die Xenien bei Zea Mays. Ber. Deutsch. Bot. Gesell. 17: 410-417. 1899. THE ENDOSPERM 185 2G. Lotsy, J. P. Balanophora globosa Jungh. Erne wenigstens . ortlich-verwittwete Pflanze. Ann. Jard. Bot. Buitenzorg II. 1 : 174-186. pis. 26-29. 1899. 27. De Yries, H. Sur la fecondation hybride de Falbumen. Conipt. Rend. 129: 973-975. 1899. 28. Burns, G. P. Beitrage zur Kenntniss der Stylidiaceen. Flora 87: 313-354. pis. 13-U. 1900. 29. Campbell, D. H. Studies on the Araceae. Annals of Botany 14: 1-25. pis. 1-3. 1900. 30. Conrad, A. H. A Contribution to the Life History of Quercus. Bot. Gazette 29: '408-418. pis. 28-29. 1900. 31. Johnson, D. S. On the Endosperm and Embryo of Peperomia pellucida. Bot. Gazette 30: 1-11. pi. 1. 1900. 32. Land, W. J. G. Double Fertilization in Compositae. Bot. Gazette 30: 252-260. pis. 15-16. 1900. 33. Chodat, R, and Bernard, C. Sur le sac embryonnaire de YHelosis guayanensis. Jour. Botanique 14: 72-79. pis. 1-2. 1900. 34. Johnson, D. S. On the Development of Saururus cerniius L. Bull. Torr. Bot. Club 27 : 365-372. pi. 23. 1900. 35. Juel, H. O. Vergleichende Untersuchungen fiber typische und parthenogenetische Fortpflanzung bei der Gattung Antennaria. Handl. Svensk. Vetensk. Akad. 33 : no. 5. pp. 59. pis. 6. figs. 5. 1900; reviewed in Bot. Zeit. 59: 131. 1901. 36. Nawaschin, S. Leber die Befruchtungsvorgange bei einigen Dicotyledoneen. Ber. Deutsch. Bot. Gesell. 18: 224-230. pi. 9. 1900. 37. Strasburger, E. Einige Bemerkungen zur Frage nach der " doppelten Befruchtung " bei den Angiospermen. Bot. Zeit. 58: 293-316. 1900. 38. Sargant, Ethel. Recent Work on the Results of Fertilization in Angiosperms. Annals of Botany 14: 689-712. 1900. 39. Tischler, G. Lntersuchungen fiber die Entvvicklung des Endo- sperms und der Samenschale von Corydalis cava. Verhandl. Naturhist.-Med. Ver. Heidelberg 6: 351-380. pis. 2. 1900. 40. Webber, H. J. Xenia, or the Immediate Effect of Pollen in Maize. Bulletin 22. Div. Veg. Path, and Phys. U. S. Dept. Agric. pp. 40. pis. 4- 1900. 41. Guignard, L. La double fecondation dans le mais. Jour. Bota- nique 15: 37-50. 1901. 42. . La double fecondation dans le Naias major. Jour. Bota- nique 15: 205-213. figs. U. 1901. 43. . Double fecondation chez les Renonculacees. Jour. Bota- nique 15: 394-408. figs. 16. 1901. 44. Holferty, G. M. Ovule and Embryo of Potamogeton nutans. Bot. Gazette 31 : 339-346. pis. 2-3. 1901. 186 MORPHOLOGY OF ANGIOSPERMS 45. Lyon, H. L. Observations on the Embryogeny of Nelumbo. Minn. Bot. Studies 2: 643-655. pis. 48-50. 1901. 46. Smith, Amelia C. The Structure and Parasitism of Aphyllon uniflorum Gray. Contrib. Bot. Lab. Univ. Penn. 2: 111-121. pis. 13-15. 1901. 47. Schnegg, H. Beitriige zur Kenntniss der Gattung Gunnera. Flora 90: 161-208. figs. 28. 1902. 48. Guignard, L. La double fecondation chez les Solanees. Jour. Botanique 16: 145-167. figs. 45. 1902. 49. Strasburger, E. Ein Beitrag zur Kenntniss von Ceratophyllum submersum und phylogenetische Erorterungen. Jahrb. Wiss. Bot. 37: 477-526. pis. 9-11. 1902. 50. Hall, J. G. An Embryological Study of Limnocharis emargi- nata. Bot. Gazette 33: 214-219. pi. 9. 1902. 51. Overton, J. B. Parthenogenesis in Thalictrum purpurascens. Bot. Gazette 33 : 363-375. pis. 12-13. 1902. 52. Cook, M. T. Development of the Embryo-sac and Embryo of Castalia odorata and Nymphaea advena. Bull. Torr. Bot. Club 29 : 211-220. pZs. 12-13. 1902. 53. Johnson, D. S. The Embryology and Germination of the Genus Peperomia. Abstract. Science 15: 408-409. 1902. 54. Ikeda, T. Studies in the Physiological Functions of Antipodals and related Phenomena of Fertilization in Liliaceae. 1. Tricyr- tis hirta. Bull. Coll. Agric. Imp. Univ. Tokyo 5: 41-72. pis. 3-6. 1902. 55. Johnson, D. S. On the Development of Certain Piperaceae. Bot. Gazette 34: 321-340. p7s. 9-10. 1902. 56. Frye. T. C. A Morphological Study of Certain Asclepiadaceae. Bot. Gazette 34: 389-413. pis. 13-15. 1902. 57. Shibata, K. Experimentelle Studien uber die Entwickelung des Endosperms bei Monotropa. (Vorlaufige Mitteilung.) Biol. Centralbl. 22: 705-714. 1902. 58. Murbeck, S. Ueber die Embryologie von Ruppia rostellata Koch. Konigl. Svensk. Vetensk. Akad. Handl. 36: 1-21. pis. 1-3. 1902. 59. Frye, T. C. The Embryo-sac of Casuarina striata. To be pub- lished in Bot. Gazette 35 : 1903. 60. Wylie, R B. A Morphological Study of Elodea canadensis. To be published in Bot. Gazette 36 : 1903. CHAPTER IX THE EMBRYO It is perhaps impossible at present to formulate any defi- nite laws for the development of the embryo of Angiosperms. The details recorded are very nu- merous and confusing, the great- est amount of variation occurring in allied forms and even in the same species. Undue attention probably has been given to the succession of cell divisions in the earliest stages of the embryo, for it is at this very period that the em- bryo seems to be peculiarly respon- sive to the conditions that surround it. What the conditions are that determine that a cell- wall in a given stage of the embryo shall run now in one plane, now in an- other, or even shall fa'l to develop, are unknown ; but the study of a large series of embryos makes it evident that if there is a normal sequence of cell divisions it is being constantly interfered with. It is probable that when these minor variations are neglected, cer- tain laws of general development will appear that are concerned with the organization of the great body regions rather than with the succession of cell divisions (Fig. 81). 187 Fig. 81. — Capsella Bursa - pastoris. Photomicrograph of seed showing embryo, endosperm, and develop- ing testa ; x 125. 188 MORPHOLOGY OF ANGIOSPERMS In general, the first division of the fertilized egg is trans- verse, and this is followed by one or more divisions in parallel planes, resulting in a row of cells. This undifferentiated group of cells is conveniently referred to as the proembryo. In gen- eral, the proembryo becomes differentiated into suspensor and embryo, which eventually become very distinct, although their origin is variable. This means that in general all the product of the fertilized egg does not enter into the structure of the embryo, a fact also true of most Gymnosperms. In general, the development of the embryo is initiated by the longitudinal division of the end-cell of the proembryo, and this is followed by divisions that result in the quadrant and then the octant stage. It is in the octant stage that periclinal walls may cut off the dermatogen, but this may be deferred to a later stage, and is often irregular. The cells of the dermatogen divide only by anticlinal walls, but the inner cells continue divisions in the three dimensions, and soon the periblem and plerome become distinguishable. In general, the end-cell of the proembryo does not produce all of the embryo, but the next cell divides trans- versely, and the daughter-cell adjacent to the embryo (hypo- physis) fills out the periblem and dermatogen of the root-tip. The organization of the growing points of stem and cotyledon, in relation to the body of the embryo, are so radically different in Monocotyledons and Dicotyledons that no general statement concerning it is possible. The fact remains that every general statement given above is contradicted by well-known and by no means infrequent ex- ceptions, and even the distinction between Monocotyledons and Dicotvledons is not always clear in the embrvo. The subieet will be treated in some detail under the titles Monocotyledons, Dicotvledons, Parthenogenesis, and Polvembrvonv. t. / C? 7 tit/ MONOCOTYLEDONS The embryo of Alisma Planiagoi as described by Hanstein 7 and Famintzin,17 has long been taken as a type of the monocot- vledonous embrvo. Among recent accounts Schaffner's 43 de- scription of the embrvo of Sagittaria variabilis, following hia earlier study of Alisma™ is the most complete, and while it confirms the principal features of the earlier accounts, the great improvement in technique since the time of Hanstein made it Fig. 82. — Sagiftaria variabilis. Development of embryo. ^4,proembryo of three cells; a, basal cell (in all figures); b, middle cell (dividing); c, terminal cell from which the' cotyledon is derived; sy, synergid; B, same stage, but terminal cell dividing; C, middle cell (6j has divided, .« being the cell from whose derivatives the stem-tip arises, and terminal cell (c) dividing; I), both cells derived from b are dividing; E, terminal cell has given rise to four cells (c), and the region derived from the middle eell , /, , has developed further; F, showing further development of the middle cell region (J), while the terminal cell region has made no further progress; &, dermatogen differentiated in the terminal cell region (c), and the middle region (J further developed ; If. differentiation of dermatogen beyond the terminal region .the middle region (J) showing the differentiation between hypocotyl (h and region producing stem tip; /, more advanced stage, Bhowing satin- regions as in 77, but the dermatogen of the root-tip not yet formed, and the plerome and peri blem undifferentiated. A-F, x 400; G, x 260; II, x 400; I, x 260.— After .S, the lateral stem-tip ; x 140; £, longitudinal section of a ripe seed ; x 26. — After Schaffnek.43 191 192 MORPHOLOGY OP ANGIOS PERMS it we would cite Sparganium (Campbell55), Potamogeton (Wiegand,63 Holferty71), Zannichellia and Naias (Camp- bell 41), Triglochin (Hill 60), and Limnocharis (Hall 82). The last-mentioned form well illustrates that the general type may be maintained, and at the same time there may be no regularity in the sequence of divisions after the first two. In fact, the apical cell of the .proembryo of Limnocharis may divide by a transverse, oblique, or longitudinal wall, and in the two latter cases the cotyledon and stem-tip are both terminal, as is the case also in Z annichellia. Among the Gramineae the same general type of proembryo is formed, but if Arena fatua (Cannon65) be taken as repre- sentative of the general situation, the origin of the organs of the embryo in relation to the cells of the proembryo is quite different. In this species the cotyledon and stem-tip are both derived from the apical cell, the entire root-tip (including root- cap) from the adjacent cell, and the coleorhiza from the third cell, the suspensor consisting of only the primary basal cell. Among the Araceae a very different type of embryo is indi- cated, but so few forms have been investigated that no conclu- sion as to its prevalence in the family is safe. In 1874 Hegel- maier 9 described the absence of a suspensor in Pistia, the fertil- ized egg producing a spherical proembryo, all of which enters into the structure of the embryo. ( Jampbell 59 found the same type of embryo in Dieffenbacliia, A\g- laonema, and Lysidkiton (Fig. 84), and states that in the seg- mentation of the egg there may be two transverse divisions lie- fore any vertical division, or a regular quadrant may be formed * J. as in the ordinary fern embryo. Even if this Pistia type should prove to be characteristic of the Araceae, it is not restricted to them, for Humphrey38 has shown that the embryos of the Seitamineae have no suspensor-: and the same is true at least of certain orchids, as shown by Fig. 84. — Lysichiton hnnfxckatcense. Longitudinal section of embryo sur- rounded by endosperm, illustrating the Pistia type. — After » ampbell.55 THE EMBRYO 193 Treub 1S for Listera ovata and Epipactis palustris, and by Leavitt "3 for certain species of Goodyera and Spiranthes. It should be noted, however, that in Lemna (Caldwell54), the reduced aquatic ally of the Araceae, a mul- ticellular suspensor is formed, the embryo resembling the Lili- um type described be- low. Among the Lilia- ceae a third type of embrvo - formation seems to prevail. After the first segmentation of the fertilized egg, which is transverse, the subsequent divi- sions are very irregu- lar, being transverse, oblique, or longitudi- nal in either cell, re- sulting in a massive proembryo. The dif- ferentiation into em- bryo and suspensor is late and irregular, the suspensor being mass- ive, and inclined to continue active divi- sion until the end of the embryo-sac is oc- cupied by a spreading suspensor tissue (Fig. 85). This is characteristic of Lilian) (Coulter44), Erythronium (Schaffner 72), Tidipa (Ernst ,ll>), and probably all the allied forms, and the meristematic activity of the suspensor is apt to result in polyembryony (see below). Just how far this Li Hum type of embryo is represented among Liliales must be determined bv future investigation, but it is distinct enough to deserve separate mention. Fig. 85. — Lilium philadelphicitm. A, proembryo of two cells; x 800 ;. Ji, middle cell of filament of three cells has divided longitudinally; x 175; C, young embryo showing massive suspensor; x 300; Z>, older embryo, snowing different form of sus- pensor ; x 300. — After Coulter.44 191 MORPHOLOGY OF ANGIOSPERMS Among the Orchidaceae there is the greatest amount of variation in the formation of the embryo. In general they are characterized by very poorly developed em- bryos, the body regions not being differen- tiated, and by an extraordinary and varied development of the snspensor as an hausto- rimn. As already mentioned, however, some of them (species of List era, E pi pad is, Good- yera, Spirantlies) have no snspensor (Fig. 86). Trenb 18 in 1879 described a number of forms in which the filamen- tous snspensor grows out of the micro- pvle, often branches, and embeds it- self in adjacent nutritive tissue, such as the placenta. He found that in Phalaenopsis grandiflora branches of the snspensor not only turn toward the micropyle, but also toward the embryo and finally envelop it. Later the same investigator 24 described the suspensor of Peristylis grandis Fig. 86. — Listera ovata. Embryo at time of shedding seed. After Pfitzer in Engler and PrantlV Nat. Pflanwnfamilien. dividing transversely growing as out Fig. 87. — Gymnadenia conopaea. Section of embryo with suspen- sor protruding from micropyle. — After Marshall Wabd.w through the micropyle, and embed- ding itself by pseudopodium-like proc- esses in the placenta. The embryo of Gymnadenia conopsea, as described by Marshall-Ward,20 is probably rep- resentative. The first division of the fertilized egg is transverse, the basal cell forming a chain-like suspensor of eight to ten more or less elongated cells that pushes through the micro- pyle into the ovary cavity, and the apical cell producing a perfect octant stage, the dermatogen being cut off in the sixteen-celled stage (Fig. 87). Leavitt 73 has also described the sus- pensors of A plectrum hiemale; of THE EMBRYO 195 Corallorhiza mult i flora, in which it consists of two very long cells and embeds its tip into the placenta; of Habenaria tri- dentata, and of H. blephariglottis, in which each of the six or seven cells of the suspensor usually sends out a branch, some of them short and reaching the integument, others elongated and passing parallel with the suspensor into the tissue at the base of the funiculus. These four types of monocotyledonous embryos, which for convenience may be spoken of as Alisma, Pistia, Lilium, and Orchid types, are, of course, related to one another in ways that suggest that they are all derivatives of one general monocotyled- onous form. It is natural to assume that this primitive form is more nearly represented by the Alisma type than by any of B Fig. 8S. — Zannichellia palustris. Development of embryo. A, young embryo ; x 320 ; B. later stage, showing beginning of differentiation into stem-tip i s i and cotyledon (c), both coming from the cells derived from terminal cell of proembryo ; x 160 ; C, stem- tip (8) and cotyledon (C) clearly differentiated; x 60. — After Campbell.41 the others, not merely because it characterizes the primitive hydrophytic forms, but also because it is the simplest type, and the others may well be modifications of it. In the Pistia type the suspensor is suppressed ; in the Lilium type it becomes massive and meristematic ; in the Orchid typo it is developed as a special haustorium that passes out of the ovule on account of the lack of endosperm, and perhaps for the same reason the embrvo does not reach the stage of differentiating organs. There have been observed certain departures from the mon- ocotyledonous type of embryo that deserve special mention. 196 MORPHOLOGY OF AXGIOSPERMS In 1S78 Solms-Laubach l5 stated that in Dioscoreaceae and certain Commelinaceae the cotyledon is lateral in origin rather than terminal. The stem-tip is terminal in origin, but is later forced to one side 'by the strong growth of the cotyledon from beneath. Such a departure is, of course, fundamental, but be- fore any generalization is ventured it should be subjected to the most critical investigation. Campbell 41 finds that in Zannir chellia the terminal cell of the proembryo gives rise to both coty- ledon and stem-tip, the separation between the two organs being determined by the first vertical division of the terminal cell (Fig. 88). The same writer47 has found another suggestive variation in Lilaea subulata, one of the Juncaginaceae. The embryonic root-tip, instead of being directed toward the sus- pensor, is directed to one side, almost in continuation of the axis of the stem-tip. This lateral origin of the root is regarded by Campbell as a primitive feature, and suggestive to him of Isoetes. In other particulars the embryo is of the Alisma type. In this connection the recent results of Murbeek 95 with Ruppia are suggestive. He confirms the account of Wille that a pri- marv root is formed at the base of the embrvo, but soon dis- organizes, and that a lateral root, formed very early, is the first functional one. This is very different from the account of Ascherson in Engler and Prantl's " Die Xatiirlichen Pflanzen- familien," which is followed in Goebel's " Organography," ac- cording to which this lateral root is the primary root, its un- usual position being due to displacement. DICOTYLEDONS The best-known dicotyledonous embryo is that of Capsella, as described by Hanstein 7 and Famintzin,17 and it has been used as a basis of comparison ever since. To illustrate the earlier stages in the development of the embryo, therefore, we have made a rather complete series of camera drawings from sections of the embryo of Capsella (Figs. 89, 90; see also Fig. 81). The proembryo is a filament of cells of varying length. The apical cell divides first longitudinally, the next two divi- sions being longitudinal and transverse in either order and resulting in the octant stage. Whether the transverse division precedes or follows the second longitudinal division, it sepa- rates the cotyledonary and hypocotyledonary regions of the em- Fie. 89. — Capsella Bursa-pastoris. A, first division of terminal (embryo) cell ; 2?, quad- rant stage; C, octant stage; /^differentiation of dermatogen ; E, differentiation of periblem and plerome (latter shaded); F, completion of periblem of root; Gr beginning of differentiation of dermatogen of root-tip (indicated by mitotic figure); //, later stage, showing plerome, periblem, dermatogen, and one layer of root-cap 'plerome and dermatogen shaded); 7, two layers in root-cap (the plerome and portion of dermatogen derived from hypophysis shaded); J, young embryo sur- rounded by endosperm ; walls of ovary also shown ; x 400. 197 198 MORPHOLOGY OF ANGIOSPERMS bryo. In the octant stage the dermatogen begins to be differ- entiated, the periclinal divisions appearing first in the terminal octants and proceeding toward the root end of the embryo. The differentiation, however, is almost simultaneous, so that the dermatogen is soon completed except that of the root-tip, which is derived from the adjacent cell of the suspensor, and appears comparatively late. The periblem and plerome are differen- tiated early from the tissue within the dermato«;en. The stem- tip and cotyledons are derived from the four apical octants, and the bulk of the hypocotyl from the four basal octants. The root-tip, however, is completed by the adjacent cell of the sus- Fig. 90. — Cipsella Bursa-pastoris. Series showing contribution of upper cell of suspen- sor to embryo (plerome and dermatogen shaded) : s, upper cell of suspensor; A, hypophysis; dt dermatogen ; d\ portion of dermatogen derived from hypophysis; pi, plerome ; p, periblem ; p' , portion of periblem derived from hypophysis ; x 400. pensor (Fig. 90, s). This cell divides transversely, the basal daughter-cell taking no part in the formation of the embryo, but the other . daughter-cell (hypophysis of Hanstein) filling out the periblem and dermatogen of the root-tip. The hypophy- sis divides transversely, the daughter-cell next the embryo com- pleting the periblem of the root. The other daughter-cell by two longitudinal divisions gives rise to a plate of four cells, each of which divides transversely, the plate of four cells toward the embryo completing the dermatogen of the root-tip, and the other plate constituting the first layer of the root-cap. THE EMBRYO 109 This type of embryo, called for convenience the Capsella type, is well represented throughout the Dicotyledons, and, so far as we have the means to judge, seems to be the prevalent type, subject, of course, to variation in detail. For example, it occurs in Salix (Chamber- lain42), in which it is questionable whether the hypophysis contributes to the periblem ; in Ranunculus (Coul- ter48) and Thalictrum (Overton83), in the latter case the suspensor some- times becoming a massive and twisted organ; in Alyssum (Riddle51), which almost exactly repeats the embryogeny of Capsella ; in Sium, in which there is a very long suspensor; in Sarcodes (Oliver30); in Avicennia (Treub24); in Trapella (Oliver20), in which there is a remarkably long suspensor with an enormously elongated basal cell ; and in Senecio (Mottier34), Silphium (Mer- rell01), and Taraxacum (Schwere40). Among the Rosaeeae Pechoutre 87 has recorded a wide variation in the struc- ture of the suspensor, different genera showing everv gradation between a aim- pie filamentous suspensor (Fragaria, Geum) and one that is short and mass- ive ( Crataegus, Amygdalus). These examples represent all regions of Dicot- vledons ; and while there are differ- ences as to the division of the basal suspensor-cell, the length of the sus- pensor, and the succession of walls in the apical cell (embryo-cell) of the pro- embryo, the general type remains the same, and resembles most nearly the Alisma type among Monocotyledons. In addition to this prevailing type, there are modifications of it that sug- 14 Fig. 91. — Loranthus sphaero- carpus. A, young embryo; x 190; £, later stage, show- ing extreme lengthening of the two bulbous suspensor- cells; e, embryo; s, suspen- sor ; x 120.— After Treub." 200 MORPHOLOGY OF ANGIOSPERMS <:est as wide a range of variation as among Monocotyledons, though not so clearly related to great groups. In Geranium, as has long been known, while the Capsella type is maintained in general, there is no hypophysis, the root- tip being covered by the tissue of a massive suspensor. In Peperomia pellucida Campbell56 and Johnson62 have both observed that the first segmentation of the fertilized egg is vertical, followed bv a transverse division, and that there is 7 V ' no indication of a suspensor. In Loranthus sphaerocarpus Treub 22 has described the first division of the fertilized egg as vertical, as in Peperomia, but followed by transverse divisions, so that the proembryo resem- bles two filaments lying side by side (Fig. 91). The two basal cells elongate enormously, forming a suspensor as in Gymno- sperms, whose length is increased by the moderate elongation of the second pair of cells, and which becomes more or less tor- tuous, the cells twisting about one another. In L. pentandras (Treub26) the elongating suspensor early forces the embryo against the resistant base of the sac. where it becomes much flattened out, and for a time bears little resemblance to an em- A B Fig. 92. — Loranthus pentandrus. -4, young embryo advancing into endosperm; thick- walled tissue at base of sae deeply shaded: e, embryo; g, suspensor: x 88 ; £. later stajre. the embryo has reached the resistant base of the sac and has become flattened out: x 144.— After Treub.28 bryo (Fig. 02). In Myoporum, as described by Billings.70 the suspensor is also extremely long and filamentous, forcing the young embryo down into the principal mass of endosperm, THE EMBRYO 20 1 which is at a considerable distance from the micropylar end of the embryo-sac (Fig. 93). In Xelumbo Lyon '^ states that there is no suspensor, but that the divisions of the egg result in a large spherical body that is still undifferentiated when consisting of several hundred cells, recalling the Pistia type among Monocotyledons. In Ceratophyl- lurn demersum Strasburger 84 has found the same undifferentiated - s — e Fig. 93. — Myoporum serratum. Young embryo with very long suspensor embedded in endosperm. — After Billings.70 B Fig. 94. — Barringtonia Vriesei. A, young proembryo; £, later stage, showing differentiation into embryo (e) and suspensor (s) ; x 1<»4. — After Tkeub.27 spherical embryo of hundreds of cells and with no suspen- sor; while in Nymphaea Conard S1 finds the same type, but associated with it is a suspensor consisting of a row of throe to five cells. In Heckeria (Piperaceae) Johnson86 has described the early stage of the embryo as a globular mass '•<>mposed of several hundred cells undifferentiated except for a rudimentary suspensor: and in Cynomovium ( Balanophora- ceae) Juel93 describes the embryo as a small spherical mass of cells with no suspensor and no differentiation into body regions. In Barringtonia Vriesei, one of the. Myrtaceae, Treub 27 has described a broad mass of tissue almost filling the micropy- lar end of the embryo-sac. At first the mass is homogeneous, 202 MORPHOLOGY OF ANGIOSPERMS and it is only late that the embryo becomes differentiated from the massive suspensor (Fig. 94). In the Rubiaceae Lloyd 57, 85 has described a remarkable de- velopment of the suspensor, which in many members of the group acts as a haustori- ura (Fig. 95). In Vaillantia hispida the large suspensor cells near the em- bryo are clustered like " a bunch of grapes," while far- ther down a single elongated cell forms a point of attach- ment. In Asperula the scanty cvto- plasm and the nu- cleus are found at the distal ends of the haustorial cells of the suspensor, recalling a condi- Fio. 95. — A, Vaillantia hispida. Young embryo showing haustorial suspensor; x 375; alter Lloyd.67 B, Aspe- rula azurea. Young embryo with haustoria from sus- pensor highly developed; after Lloyd. » tion which has been described for root hairs. It is worthy of note that among the Spermacoceae and in Houstonia there is a complete absence of these striking- adaptive characters of the suspensor. It is among the Leguminosae, however, that the greatest amount of variation in embryogeny exists and the most unusual forms appear, as shown by Guignard 21 (Figs. 96-98). It is impossible to give in a brief account any adequate idea of the amount of variation displayed by the nearly forty species Guig- nard has described, involving in the main the character of the proembryo and the final condition of the suspensor. In 1880 Strasburger 19 had called attention to the fact that the cells of the very long suspensor of Lupinus separate early, leaving the embryo free and some distance from the micropylar extremity of the sac. This, however, is but one phase of the embryogeny of the Leguminosae. In every case the first segmentation of the THE EMBRYO 203 egg is transverse, but this may be followed either by longi- tudinal or transverse divisions, in the former case generallv re- suming in a massive and often globular proembryo, in the latter resulting in an extraordinarily long and conspicuous filamen- tous proembryo. In almost every case the suspensor-cells are more or less swollen and bladdery and surcharged with nutritive material, forming a conspicuous nutritive tissue for the embryo. The two types of proembryo may be illustrated as follows : As illustrations of the massive proembryo, in which the sus- pensor and embryo are gradually differentiated, but are never very distinct externally except by a constriction between them, may be cited species of Acacia and Mimosa ; Cercis siliquas- trum, in which the oblong proembryonic mass broadens at each end to form the embryo and suspensor ; Caesalpinia mimosoides, in which the embryo becomes distinct rather early as the region of more actively dividing cells; Cytisus Laburnum, in which the suspensor becomes a great mass of loose rounded cells re- sembling a globular cluster of berries ; Anthyllis tetraphylla, in which the suspensor is like that of Cytisus, but the clustered -^f — .€ Fig. 96. — Embryos of Leguminosae. A, Cercis siliquastrum, with suspensor and embryo developing about equally ; x 270 ; E-E, Spartium junceum : «, embryo ; s, suspen- sor; x 300. — After Glionard.31 cells are much fewer in number ; Spartium junceum and Trifo- lium resupinatum, in which the massive proembryo seems to constrict as in Cercis, but the suspensor as the cotyledon stage approaches is smaller than the embryo; Tetragonolobus pur- pureas, in which the larger part of the massive proembryo be- comes the suspensor; Hedysarum coronarium and Aracliis hypo- 204 MORPHOLOGY OF ANGIOSPERMS 13* gaea; Onobrychis petraea, in which the proembrvo is a globular mass of cells; and Phaseolus multiftorus and Erythrina crista- gatli, in which the massive pro- embrvo is elongated and there is no superficial separation be- tween embryo and suspensor. In case two or more of the first divisions are transverse, forming a filamentous proem- brvo, the end-cell forms the en- tire embryo, the suspensor-cells becoming relatively extremely large and bladdery inflated. Two general types may be noted. In Orobus august if olius, 0. au- reus, Pisum sativum, Lathyrus heterophyllus, L. odoratus, Er- vum Errilia, and Vicia navbon- nensis, a proembrvo consisting of a row of three cells divides longitudinally ; the two basal cells become much elongated, bladdery inflated, and multinu- cleate; the middle pair become bladdery inflated and multinu- cleate ; and at the end of such a suspensor the terminal pair of cells organize a small round, oval, or elongated embryo. In Cicer arietinum it is interesting to note that the same huge suspensor and small embryo appear, but the suspensor-cells instead of becoming multinucleate divide, forming a many-celled massive suspensor. In the other type, transverse divisions continue until the proembrvo consists of a long filament of cells, all of which, excepting the end-cell; form a suspensor, as in Medicago falcata; Galega orientalis, in which the long suspensor finally becomes massive by longitu- dinal divisions; and Ononis fruticosa, in which the suspensor- cells become very large and rounded, forming a chain that finally breaks up. In Ononis alopecuroides, however, the sus- pensor is reduced to a single cell. The genus Lupinus is espe- Fig. 97. — Embryos of Leguminosae. A, Orobus angustif olius, with suspensor of three multinucleate cells; x 336. B, Ciar arietinum, with large multi- cellular suspensor and small embryo («); x 160.— After Goignabd.m THE EMBRYO 205 eially characterized by its extensive, worm-like, and large-celled suspensors, whose cells often break apart. The suspensor may consist of twenty pairs of elongated cells, forming a tortuous filament extending the entire length of the embryo-sac, with a very small embryo at the tip, as in L. subcarnosus ; or it may be a filament of short, very broad cells, suggesting a leech in ap- pearance, as in L. pilosus; or it may be a loose, large-celled tissue lying along the cavity of the embryo-sac, actively dividing and more or less surrounding the late-forming embryo with its rounded cells, that finally break apart and become disorganized, as in L. polyphyllus, L. mutabilis, L. truncatus, etc. The degree of development of the embryo is extremely vari- able. In some cases a plumule with several leaves is formed, and Fig. 98.— Embryos of Leguminosae. A, Lupinus subcarnosus, with long sinuous sus- pensor and small four-celled embryo (e) ; x 270. B, L. luteus, with many suspensor- cells binucleate; x 160. C, L. pilosus, with some basal suspensor-eells isolated; x 80.— After Guignard." even lateral roots appear, as in Gramineae, Impatiens, Cucur- bita, Trapa, etc. ; while in many parasites and saprophytes the embryo is represented only by an undifferentiated mass of cells. 206 MORPHOLOGY OF ANGIOSPERMS Among the Monocotyledons such undifferentiated embryos ap- pear among Orchidaceae and Burmanniaeeae, in the former family the primary root never appearing; but they are even more numerous among Dicotyledons. Goebel 28 states that the embryo of Monotropa consists of five to nine cells, and that of Pyrola secunda, quoting from Hofmeister, of eight to sixteen cells. The entirely undifferentiated embryo of Aphyllon uni- florum has been noted by Miss Smith 7S ; and the embryos of Orobanchaceae (Koch14), and of Balanophoraceae and Cytina- ceae (Solms-Laubach8), consist of a very small mass of tissue. In this connection it should be noted, however, that in Cuscuta and Viscum the embryos are large and "well developed. In some non-parasitic forms also poorly developed embryos occur, as in Utricularia (Kamienski 1J), in which the embryo develops no root-tip but produces a large number of peculiar leaves. The appearance of a single cotyledon in the embryos of certain Dicotyledons has naturally attracted attention. As a prefatory illustration, it may be observed that in Trapa natans, one cotyledon is much smaller than the other, and this suggests the possibility of further abortion and even of suppression of one of the cotyledons. In Ranunculus Ficaria Irmisch1 long ago reported the occurrence of a single cotyledon sheathing below, and Erianlhis hiemalis, Gorydalis cava, and Varum (Buniiun) bulbocastanum have also been included in the list of " pseudo-monocotyledons.'' In the case of C. bulbocastanum Hegelmaier 10 discovered that the apparently single and ter- minal cotyledon is accompanied by a second almost completely aborted and lateral cotyledon. All of these forms have been investigated recentlv by Schmid,91 who discovered that in Eri- c i *J 7 anthis hiemalis the two cotyledons are of unequal size; that in Ranunculus Ficaria there is hardly a trace of a second cotyle- don, and that this trace was probably mistaken by Irmisch 1 for a sheathing base; and that in Corydalis cava there is only a slight protuberance to represent the second cotyledon, the func- tioning one in its growth gradually assuming a more terminal position and thrusting the stem-tip to an apparently lateral posi- tion, but in C. nobilis and C. lutea the normal development of cotyledons is found. In Cyclamen persicum, also. Schmid found embrvos in ripe seeds with no trace of a second cotvle- %J IT i don. From these cases it is evident that in certain dicotyled- THE EMBRYO 207 onous forms there mav be earlv abortion, which may even approach suppression, of one of the cotyledons ; and that in consequence of this the single functional cotyledon may appear terminal and the stem-tip lateral. To call such cases " pseudo- monocotyledons," however, is not consistent with the real nature of the monocotvledonous embryo. It is of interest to note, how- ever, that Miss Sargant,94 in her recent study of the " mono- cotvledonous Dicotyledons," a special case being made of Ranun- culus Ficaria, has concluded that the apparently single cotyle- don is a fusion of two. The peculiar development of the cotyledons of Nelumbo has suggested to Lyon74'75 that they represent a single two-lobed cotyledon, and that this fact, along with certain anatomical details, should place Nelumbo among the Monocotyledons. In its early stage he represents the proembryo as being a many- celled spherical body, that later becomes a flattened mass filling the micropylar extremity of the sac. The stem-tip arises from the free surface toward one side, and a cotvledonarv ridge arises behind it as a crescentic mound of tissue, whose wings finally extending around form a sheath about the stem-tip. By the development of two growing points on this cotvled- onarv sheath two lobes appear and develop rapidly, the two becoming concave and surrounding the plumule as a tube. The evidence in favor of a single cotyledon seems convincing until this embryogeny is compared with that of Nymphaea, as has been done by Conard.81 In this genus the same spherical mul- ticellular proembryo appears, two opposite and symmetrical cotyledons with the stem-tip between them arising from the free side, and the basal portion forming the hypocotyl. At maturity the cotyledons become concave and inclose the plumule, just as in Nelumbo. There can be no question that the two genera are closely related; and since the embryogeny of Nymphaea is typically dicotyledonous, it follows that that of Nelumbo must be only a modification of it, and that for some reason the stem- tip does not occupy its usual central position, and the two cotyledons arise for a time en masse, as in the case of petals in sympetaly. Conard calls attention to such behavior on the part of the cotyledons of Tropaeolum, which appear " connate- perfoliate " about the hypocotyl : and also to the fact that Hegel- maier noted the complete fusion of the cotyledons along one 208 MORPHOLOGY OF ANGIOSPERMS edge in Nuphar lutea. In his recent study of Ceratophyllum Strasburger M finds that the embryo in its earlier stages bear.-, a striking resemblance to that of Nelumbo, there being a large spherical mass of cells with no suspensor (Fig. 80). The em- brvo of Nelumbo has the rudiment of a root, although it never develops, the first functional roots coming from the stem above the cotyledon (Fig. 80, s). In Ceratophyllum the reduction due to the water habit has gone further, not even the rudiment of a root appearing in the embryo. The two cotyledons of Ceratophyllum so strongly resemble the condition in Nelumbo, that Strasburger, after examining the embryo of the latter, was forced to believe that here also, as in Ceratophyllum, there are two cotyledons. The occasional occurrence of a whorl of three cotyledons has been reported for Que reus, Amygdalus, Phaseolus, etc., and many other cases are given bv Braun.6 In this connection, recent suggestions as to the phylogeny of the cotyledon may be referred to. The current opinion re- gards it as a modified foliage leaf, and this is borne out in the majority of Dicotvledons by the assumption of the foliage func- tion. The terminal cotyledon of Monocotyledons, however, seems to belong to a different category, and to hold no relation to a foliage leaf or to a foliar member of any description. In a recent paper II. L. Lyon 88 develops the idea that the cotyle- don of Angiosperms is phylogenetically related to the sucking organ known as the " foot " among Bryophytes and Pterido- phytes. His own summary makes his position clear: (1) The typical embryos of the Pteridophytes and Angiosperms differentiate into three primary members, the cotyledon, stem, and root ; (2) cotyledons are not arrested leaves, but are primarily hausto- rial organs originating- phylogenetically as the nursing-foot in the Bryophytes and persisting throughout the higher plants ; (3) the mono- cotyledonous condition is the primitive one and prevails in the Bryo- phytes, Pteridophytes, Monocotyledons, and some Gymnosperms ; the two (sometimes more) cotyledons of the Dicotyledons are jointly the homologue of the single cotyledon of the Monocotyledons ; (4) the cotyledon always occurs at the base of the primary stem ; (5) the hypo- cotyl is a structure peculiar to the Angiosperms. being differentiated between the primary stem and root ; (6) the so-called cotyledon of the Pteridophytes and Gymnosperms, with the probable exception cf Ginkgo and the Cycads, are true foliage leaves. THE EMBRYO 209 The same general idea has been expressed by Balfour,80 as the following quotations show : '" We ought, I think, to look upon the embryo as a protocorm of embryonic tissue adapted to a seed-life. Under the influence of its heterotrophic nutrition and seed-environment it may develop organs not represented in the adult plant as we see in, for instance, the embry- onal intraovular and extraovular haustoria it often possesses. There is no reason to assume that there must be homologies between the protocorm and the adult outside an axial part with its polarity. There may be homologous organs ; but neither in ontogeny nor in phylogeny is there sufficient evidence to tsuow that the parts of the embryo are a reduction of those of the adult." " That the cotyledons, primarily suctorial organs, should change their function and become leaf-like under the new conditions after germination is no more peculiar than that the hypocotyl should take the form of an epicotylar internode, from which it is intrinsically different as the frequent development upon it of hypocotylar buds throughout its extent shows." " The protocorm has, I believe, developed along different lines in the Dicotyledons and Monocotyledons. This has been to the advan- tage of the former in the provision that has been made for rapid as opposed to sluggish further development. Confining ourselves to the general case, the axial portion of the ■ protocorm of the Dicotyledon, the hypocotyl, bears a pair of lateral outgrowths, the cotyledons, and terminates in the plumular bud and in the primary root respectively. The cotyledons are its suctorial organs, and the hypocotyl does the work of rupturing the seed and placing the plumular bud and root by a rapid elongation which commonly brings the plumular bud above ground, protected, it may be. by the cotyledons. These latter may then become the first assimilating organs unlike or like to the epico- tylar leaves. In the Monocotyledons the axial portion of the proto- corm has usually no suctorial outgrowths. Its apex and usually its base also are of limited growth. The plumular bud is a lateral devel- opment, and the primary root often an internal one. The suctorial function is performed by the apex of the protocorm, termed here also the cotyledon." " I use the term purely as an objective designation, and in the original meaning of the suctorial organ in the embryo. This terminal cotvledon in the Monocotvledons is not a leaf nor the homolosfue of the lateral cotyledons in the Dicotyledons." An explanation of the terminal cotyledon of Monocotyledons has been suggested by Miss Sargant 89 in her study of the seed- lings of Liliaceae. In Anemarvliena she finds the cotvledon 210 MORPHOLOGY OF ANGIOSPERMS traversed by two opposed vascular bundles, which suggest the fusion of two organs and a derivation from the dicotyledonous condition. This position is further strengthened by the well- known tendency among certain Dicotyledons for the cotyledons to become more or less completely fused (see Chapter XV). The whole problem, however, is too indefinite as yet, and the data are too few to permit well-grounded conclusions, but it is well worth consideration. PARTHENOGENESIS The term parthenogenesis was once very loosely applied, including all cases of the appearance of embryos without fer- tilization. Strictly, however, it includes only those cases in which the normal egg produces an embryo without fertilization, and this phenomenon has thus far been demonstrated in only three angiospermous genera, to be described below. Apogamy, being the production of a sporophyte by a gametophyte without the act of fertilization, of course includes parthenogenesis, but the production of sporophytes by gametophytic structures other than the egg may for convenience be distinguished as vegetative apogamy. In this category wyould be included all cases of embryos derived from unfertilized synergids, antip- odals, and endosperm, the last-named structure being included or not dependent upon one's view as to its morphological char- acter. When an unfertilized synergid produces an embryo, it might be claimed that it is not a case of vegetative apogamy but of parthenogenesis, since the synergid is to be regarded as a non-functioning egg. This simply serves to illustrate the fact that categories are essentially arbitrary and artificial. A third category includes those cases in which embryos are produced by the tissue of the nucellus or of the integument. This is not apogamy, although it has often been so called, for it is a case in which a sporophyte is produced by sporophytic tissue, and can be included under the general name of budding. In addition to the normal method, therefore, embryos appear among Angio- sperms in three ways, namely, by parthenogenesis, by vegetative apogamy, and by budding. In most cases vegetative apogamy and budding are associated with polyembiwony, and they will be considered under that head. The three well-authenticated cases of parthenogenesis among Angiosperms are as follows: THE EMBRYO 211 In 1898 Juel 53, 66 reported parthenogenesis in Antennaria alpina, and two years later published a very full account of this species and also of A. dioica, in the latter of which fertili- zation occurs regularly. In the parthenogenetic A. alpina usu- ally only pistillate plants are found, and in the staminate plants that do occur the pollen is either lacking or feebly developed. Juel was able to show conclusively that the embryo develops from the unfertilized egg. lie was also able to satisfy himself that the number of chromosomes (about fifty) remains un- changed throughout the entire life history, no reduction taking- place in the formation or germination of the megaspore. The first division of the nucleus of the megaspore mother-cell is like the divisions in vegetative cells, and neither in the form of chromosomes nor in the character of the spindle does it resemble the heterotypic division that is so constantly associated with the reduction of chromosomes. The mother-cell gives rise to only one megaspore, not forming a tetrad. In A. dioica, in which fertilization regularly occurs, the megaspore mother-cell gives rise to a tetrad, the first division being accompanied by a reduc- tion in the number of chromosomes ( from about twenty- four to about twelve). While the number of chromosomes was not de- termined with absolute accuracy for either species, the numer- ous countings prove the principal point, namely, that in .1. dioica a reduction occurs at the beginning of the gametophyte generation, but in the parthenogenetic A. alpina the number remains unchanged throughout the life history. In the latter also the polar nuclei do not fuse to form a primary endosperm nucleus, but each divides independently and forms a mass of endosperm, showing, like the egg, an ability to divide without previous fusion. In 1901 Murbeck 76 discovered that parthenogenesis is more or less constant in all the species of AlchemiUa belonging to Eualchemilla ; but he succeeded in finding a species (A. ar- vensis) in which fertilization regularly occurs. In the struc- ture of the nucellus AlchemiUa differs decidedly from Anten- naria, there being a large number of megaspore mother-cells, many of which form tetrads ; and it is not uncommon for sev- eral of the resulting megaspores to germinate. The general appearance of the embryo-sac is normal, and the polar nuclei usually fuse to form a primary endosperm nucleus. Since this 212 MORPHOLOGY OF ANGIOSPERMS fusion was observed in several parthenogenetic species of Al- chemilla (A. sericata, A. " hybrida" A. pubescens, A. pasto- raliSj A. acutangula, A. alpestris, and A. speciosa), its failure, as in Antennaria alpina, can hardly be regarded as character- istic of parthenogenetic forms. In the parthenogenetic species of Alchemilla, as Antennaria alpina, the number of chromo- somes remains unchanged throughout the life-history. Al- though the number was not positively established, the counting never showed less than thirtv-two or more than fortv-eight. In Alchemilla arvensis, in which fertilization regularly occurs, the numbers are sixteen and thirty-two. Aside from the more difficult cytological evidence, a convincing proof of the existence of parthenogenesis in Alchemilla alpina is found in the fact that the segmenting embryos are often obtained from unopened buds in which no pollen has been developed. In A. arvensis (Murbeck ""), in which fertilization occurs, the pollen-tube en- ters the chalaza and traverses the integument. In 1902 Overton83 discovered parthenogenesis in Thalic- trum parparascens, the investigation having been suggested by an early observation that Thalictrum Fendleri set seed freely in the absence of staminate plants. Only ovulate plants were brought into the greenhouse and forced. These set seed con- taining good embryos several weeks before the staminate plants of the vicinity had developed pollen. Investigation showed beyond a peradventure that these embryos were derived from unfertilized eggs. He also compared normal and parthenoge- netic embryos, and found that the latter are noticeablv slower in starting, though the two kinds become exactly alike at matu- rity. The cytoplasm is very dense about the unfertilized egg, and when a zone in contact with the egg changes in appear- ance the first segmentation occurs. He suggests that there is a reaction of some kind between the egg and the contiguous cytoplasm that brings about the change in the physical con- stitution of the egg that induces segmentation. This is con- ceivable from the fact that artificial parthenogenesis has been induced in the unfertilized eggs of certain low animals by changing the osmotic pressure. Overton finds that in nature this species probably produces normal and parthenogenetic em- bryos in about equal numbers. Still more recentlv Treub92 has concluded that Ficas hirta THE EMBRYO 21:) produces parthenogenetic embryos. The observation was not direct or conclusive, the inference being based upon the failure to discover pollen-tubes although embryos were common, the feeble development of endosperm, and the poorly developed synergids, all of which is negative evidence. Treub suggests that the stimulus that induces the egg to divide in this case is the puncture made by the pollinating wasp Blasfophaga. There seems to be no doubt that other cases of partheno- genesis will be discovered among Angiosperms, and that many embryos supposed to be normal are parthenogenetic. There seems to be no reason to doubt. that if an envelop of cytoplasm may result in the segmentation of the egg in Thalictrum, it may often have the same result in other cases. For example, Treub25 observed that in certain Burmanniaceae (Gonyanthes Candida and Burmannia javanicd) the egg does not segment until the embryo-sac is packed full of endosperm. Such a con- dition might well repeat the results in Thalictrum. In fact, all cases in which there is a long delay before the egg segments may be suspected of occasional parthenogenesis. POLYEMBRYOXY Polyembryony in A7igiosperms,' while not so prevalent as in Gymnosperms, is by no means a rare or recently discovered phenomenon. As early as 1719, Leeuwenhoek found two em- bryos in orange seeds. In Euonymous latifolius polyembryony was discovered three times independently; by Petit-Thouars in 1807, In- Grebel in 1820, and bv Treviranus in 1838; In this species about one-half of the ripe seeds are said to contain more than one embryo. A. Braun in 1850 gave an historical resume of the subject, and cited sixty cases as known at that time. The first demonstration of the real nature of certain cases of polyembryony was. made by Strasburger 12, 16 in lv7^. He found that in Fuul-ia ovata, Nothoscordon fragrans, Citrus Aurantium, and Coelebogyrie ilicifolia the cells of the nucellns above the apex of the embryo-sac become rich in contents, divide and grow, and form several embryos that push the sac wall before them and become placed in the seed like normal em- bryos. Tn Funkia the egg is fertilized, but seldom or perhaps never produces an embryo, dividing a few times and then disor- ganizing (Fig. 99). When pollination is prevented artificially, 2U MORPHOLOGY OF ANGIOSPERMS Fio. 99.— Funkia ovata, showing adventitious embryos; fer- tilized egg has given rise to weak proembryo of three cells; x 190. — After Strasbirgeb.18 the adventitious embryos begin to develop but never mature. In Citrus the embryos are derived not onlv from the cells of the nucellus capping the sac, but also from those lower down, which may be separated from the sac by several cells. In Coele- bogyne, long sup- posed to be par- thenogenetic, fer- tilization never occurs in Europe, since only pistil- late plants are cultivated. These are not cases of apogamy, as often stated, but are evidently cases of vegetative multiplication or budding, since the em- bryos arise from sporophytic tissue. In Opuntia vulgaris (Ganong49) the ripe seed contains one large embryo and sev- eral smaller ones pressed to one side. Half ripe seeds generally show that the large embryo comes from the micropylar end of the sac, while the small ones arise from nucellar tissue. Among Cactaceae the only previously known case of polyembryony is that of Opuntia tortispina. The multiplication of em- bryos bv budding from a mass- ive suspensor also occurs, and is especially common in the Lilium type of embryogeny, in which the suspensor is strongly meristematic. In 1805 Jef- frey 35 called attention to the fact that in Erythronium ameri- canum the suspensor is a mass- ive and lobed tissue on whose free surface two to four embryos appear, only one persisting (Fig. 100). As in Funkia, the cells of the nucellus are Fig. 100. — Erythronium americanum . Four embryos derived from fertilized x 144. — After Jeffrey.3* ezs; THE EMBRYO 215 rich in protoplasmic contents, and this led Jeffrey to sus- pect that a reinvestigation of Funkia with the aid of modern technique would reveal a similar condition. The examination, however, confirmed Strasburger's account, so that while the general appearance of sections is much the same in the two cases (cf. Figs. 99 and 100), it is established that in Funkia the embryos come from the nucellus, while in Erythronium they come from the fertilized egg. In Erythronium albidum Schaff- e in em cm D Fig. 101. — Limnocharis emarginata. A-C, three sections of one embryo, showing em- bryo proper (e) and embryo-buds from suspensor (em) ; Z>, appearance of growing point of stem {gp). — After Hall.82 ner 72 found the same large, irregular, and much-lobed sus- pensor, but it was associated with only one embryo. In Tulipa Gesneriana Ernst 69 also observed the phenomenon of a massive suspensor associated with one to six embryos, only one of which usually persists. In these cases the Lilium type of embryogeny is obscured by the early and rapid growth of the suspensor region of the proembryo. the embryonal cell appearing hardly more than one of the cells of its free surface. In these cases 15 216 MORPHOLOGY OF ANGIOSPERMS of polyembryony, therefore, one of the embryos is to be regard- ed as normal, and the others as secondary or adventitious. Ex- actly the same thing sometimes occurs in Limnocharis emargi- nata, one of the Alismaceae, as observed by Hall S2 (Fig. 101). In this species the basal suspensor-cell may increase very much in size and remain un- divided, as is most common in the Alisma type ; or it may divide extensively, forming a massive tissue from which several embryos bud. It was not observed whether more Fig. 102.— Mimosa Den- than one embryo matures, but presumably hartu. Tiiree embryos not. This case is interesting not onlv on occupying position of aeeoimt of the polyembryony, but also be- egg-apparatus; x 384. • .1 w 1 Ai —After Gdionabd.« cause it emphasizes the relation between the Alisma and Lilium types of embrvogenv. Illustrations of ordinary apogamy are relatively numerous, apparently every cell within the embryo-sac being able under certain conditions to produce an embryo. In some cases a synergid is fertilized, and then the resulting embryo should probably be regarded as normal ; it certainly is not apogamous. For example, Schwere 40 discovered synergid fertilization in Taraxacum officinale ; and Guignard 6S has ob- served that in Xaias major the per- sistent synergid instead of the pri- mary endosperm nucleus may be fer- tilized bv the second male nucleus, resulting in two embryos lying side by side (Fig. 103). An embryo from a synergid in addition to a normal embryo from the egg has been re- ported by several observers. In Mi- mosa Denhartii Guignard 21 has found cases which suggest the development of embryos from all three cells of the egg-apparatus. Sometimes two simi- lar embryos appear, one in the position of the egg and the other in that of a synergid; sometimes a group occurred con- sisting of one unchanged svneraid, one embryo in the egg position, and a second embryo in the position of the second Fig. 103. — Xaias major. Two embryos, one from fertilized egg, the other from fertilized synergid, a male nucleus hav- ing fused with nucleus of synergid instead of polar nu- cleus ; e, endosperm nucleus ; x 176. — After Guignard.** THE EMBRYO 21 Y synergid ; and in one case three embryos were seen occupying the position of the egg-apparatus (Fig. 102). Although favor- ing this interpretation, Guignard mentions the possibility that the extra embryos may have come from the separation of early -i gments of the egg. a view doubtless suggested by the separa- tion of the cells of the suspeusor in certain of the Legu- minosae. In Vincetoxicum nigrum and V. medium ( hauveaud 33 finds that polyembryony is a regular phenomenon, one, two, three, four, and even five embryos appearing, more than one of which may reach maturity. The svnergids are doubtless involved. Chauveaud found four or five bodies in the pollen- tube which lie thought might be interpreted as male nuclei, and responsible for polyembryony. He also concludes that poly- embryony is a primitive feature of Angiosperms, the number having been reduced in the interest of one strong embryo. In describing synergid fertilization in Iris sibirica, Dodel 31 im- plies a somewhat similar view, when he interprets the synergids as partially aborted eggs. In this form more than one pollen- tube may enter the micropyle. In certain orchids, as Gymnadenia conopsea (Stras- burger16), two embryos sometime occur in the same sac, but their origin is uncertain, although it is very probable that one of them is derived from a synergid, either apogamously or by fertilization. In a preliminary paper, Hegelmaier 79 states that polyem- bryony is habitual in Euphorbia dulcis, two to nine embryos appearing at the micropylar end of the sac. One of the em- bryos, which certainly comes from the egg and mav be dis- tinguished from the others by the presence of a suspensor, becomes the functional embryo. Fertilization was not studied, and so the origin of some of the embryos is in doubt, although it is certain that some come from the nucellus. Two embrvos t/ often reach the cotyledon stage, with tissue systems differen- tiated, while the others appear as irregular masses. Allium odorum presents a remarkable case of polyembryony. In 1895 Tretjakow 3fl reported one to three embrvos from the antipodal cells (Fig. 104), the fertilized e^ix and sometimes a synergid forming additional embryos. In the same species Hegelmaier 45 observed five embryos in a single embryo-sac ; 21S MORPHOLOGY OF ANGIOSPERMS one normal, one from a synergid, two from antipodal cells, and one from the inner integument (Fig. 105). It is interesting to note that while polyembryony is so frequent in Allium odo- rum, it has not been observed in other species of the genus. Hegelmaier exam- ined A. fistulosum and J., ursinum, and Elmore50 made a thorough study of A. cernuum, A. tricoccum, and A. canadense, without discovering a single extra em- bryo, reporting also very small and eva- nescent antipodals. In parthenogenetie species of Alchemilla Murbeek76 found embryos from the egg, from the synergids, and from the nueellar tissue (Fig. 106). In Balanophora elongata and B. glo- bosa fertilization is known not to occur, and both Treub 46 and Lotsv 58 state that the embrvo is formed by the upper polar nucleus. In addition to this, a cell in the midst of the endosperm is said to develop into a five to ten- celled " pseud-embryo,*' whose significance and history we are Fig. 104. — Allium odorum. Three embryos derived from the three antipo- dal cells; x 118.— After Tretjakow.38 t o A Fig. 105. — Allium odorum. A, section of ovule with four embryos, one from egg, one from a synergid, ore from an antipodal cell, and one from the wall : x 15; B, tv o embryos, one from egg and one from a synergid: the other synergid Bomewhat enlarged and lying between the two embryos : x 24rt : C. embryo derived from inner integument: /, inner integument ; o, outer integument; x 246. — After Hegelmaier. 15 at a loss to understand (Fig. 107). In the allied Helosis guaya- nensis, also, Chodat and Bernard 64 think that fertilization does not occur, and that the embrvo arises apogamously from the endosperm. It is evident that polvembrvonv is bv no means so rare a THE EMBRYO 219 phenomenon as many may have supposed. The cases on record are already sc numerous that only an exhaustive study of the literature would make it safe to venture an estimate of the number. Since in nearly all the cases described this phenome- non is rare rather than habitual, it is probable that under con- ditions not yet understood a large number of plants may exhibit polyembryony occasionally. Fig. 106. — Embryos in parthenogenetic species of Alchemilla. A, A. sericata, one par- thenogenetic embryo from egg and one from synergid, the other synergid breaking down : the two polar nuclei and antipodal cells also shown ; x 284; B, A.pagtoralis, showing one synergid partly disorganized, one embryo of four cells from unfertilized vgg, one embryo from nucellus, two polar nuclei and one synergid nucleus forming group at middle of sac, also three disorganizing antipodal cells; x 190. After MUBBECK.*0 The scattered literature of the subject is admirably sum- marized by Ernst 69 in his presentation of polyembryony in Tirfipa Gesneriana. The following synoptical statement is taken from Ernst, and supplemented by the more recent addi- tions. Tn case the same form is treated in several account^, there is no attempt to cite all of them or even the first refer- ence, but a selection is made of those citations that direct to Fig. 107. — Balanophora elongata. Stages in development of embryo-sac, endosperm, and embryo. A, archegonium-like megasporangium with mother-cell that becomes megaspore directly without forming tetrad: x 145; B. quadrinucleate stage of embryo-sac; x 200; C, nearly mature sac showing above the two synenrids and oosphere, just beneath the micropylar polar nucleus, and at opposite end of sac a group of four nuclei, the three antipodals. and the lower polar nucleus; x 280: D, at upper end the synergids and egg are disorganizing, just beneath are two cells resulting from first division of upper polar nucleus ; x 280; E. six cells of endosperm shown ; synergids and egg still visible at upper end of sac ; x 300 ; F. two-celled embryo formed from an inner cell of the endosperm ; x 300.— After Treub.«« 220 THE EMBRYO 221 3 the most complete descriptions. The forms that Ernst include under " pseudo-polyeinbryony " are not treated in our discus- sion of the subject. Pseudo-poly embryony . 1. Ovules Grown Together. Pirns Mains, Loranthus euro- paeus, Viscum album (all A. Braun 4). 2. Division of Nucellus. Morns albus (Hofmeister2). Orchis Mono (Braun4), Gymnadenia conopsea (Strasburger "), Coffea aro- bica ( Hanausek "). 3. Development of Several Embryo Sacs in the same Nu- cellus. Cheiranthus Cheiri (Schacht3), Rosa sp. (Hofmeister2), Eosa livida (Strasburger "), Tri folium pratense (Jonsson23), Taraxa- cum officinale (Schwere40). True Polyembryony. A. Embryos derived from cells outside the sac, hence from sporo- phytic tissue (vegetative multiplication or budding). 1. Embryos Derived from Cells of the Nucellus. Funk in ovata (Strasburger1-). Nothoscordon fragrans (Strasburger"), Citrus Aurantium (Strasburger 16), Mangifera inclica i Strasburger w), Euony- mus americanus (Braun4), Coelebogyne ilicifolia (Braun.4 Stras- burger16), Clusia alba (Goebel"), Opuntia vulgaris (Ganong49), Al- chemilla pastoralis (Murbeck90). 2. Embryos from Cells of the Integument. Allium odorum (Tretjakow,36 Hegelmaier 45). B. Embryos derived from cells within the sac (parthenogenesis and vegetative apogamy) ; although not in the same morphological category, embryos from the suspensor are also included in the list (vegetative multiplication or budding). 1. Normal Occurrence of Two Eggs. Santalum album and Sinningia Lindleyana (both Strasburger12). 2. Embryos from Synergids. Glaucium luteum (Hegelmaier12), Mimosa Denhartii and Schrankia uncinata (Guignard21), Iris sibi- rica (Dodel:1). Lilium Martagon (Overton32), Vincetoxicum nigrum and V. medium ( Chauveaud 3S), Allium odorum (Tretjakow,38 Hegel- maier45). Taraxacum officinale (Schwere4"), Aconitum Napellus (Osterwalder69), Alchemilla sericata (Murbeck90), Naias majcr (Guignard M). 3. Splitting of Embryo Derived from Egg. Loranthus euro- paeus (Braun *). 4. Embryos from Antipodal Cells. Allium odorum (Tretja- kow.38 Hegelmaier13). 5. Embryos from Endosperm Cells. Balanophora elongata (Treub 48). 222 MORPHOLOGY OP ANGIOSPERMS 6. Embryos from the Suspensor. Erythroniwm dens-canis (Hofmeister6), E. americanum (Jeffrey '*), Tulipa Gesueriana (Ernst"), Limnocharis emarginata (Hall"). LITERATURE CITED 1. Irmisch, T. Beitrage zur vergleichenden Morphologie der Pflan- zen. Abhandl. Natur. Gesell. 2: 30-43. 1854; 3: 63-102,107-137. 1855. 2. Hofmeister, W. Neuere Beobachtungen iiber Ernbryobildung der Phanerogainen. Jahrb. Wiss. Bot. 1 : 82-188. pis. 7-10. 1858. 3. Schacht. H. Ueber Pflanzen-Befruchtung. Jahrb. Wiss. Bot. 1 : 193-232. pis. 11-15. 1858. 4. Braun, A. Ueber Polyembryonie und Keimung von Coelebogyne, ein Naehtrag zur der Abhandluug iiber Parthenogenesis bei Pflanzen. Abhandl. Konigl. Akad. Wiss. Berlin, pp. 109-20J. 1859. 5. Hofmeister, W. Neue Beitrage zur Kenntniss der Embryobildung der Phanerogamen. II. Monocotyledonen. 1861. 6. Braun. A. Pflanzenmissbildungen. 1869. 7. Hanstein, J. Entwickelungsgeschichte der Keime der Monocoty- ledonen und Dicotyledonen. Bot. Abhandl. Bonn, pp. 112. 1870. 8. Solms-Laubach, H. Graf zu. Ueber den Bau der Saraen in den Familien der Rafflesiaceae und Hydnoraceae. Bot. Zeit. 32: 337-342, 353-358, 369-374, 385-389. pi. 8. 1874. 9. Hegelmaier, F. Zur Entwickelungsgeschichte monocotyledoner Keime nebst Bemerkungen iiber die Bildung der Samendeckel. Bot. Zeit. 32: 631-639, 648-671, 673-686, 689-700, 705-719. 1874. 10. . Vergleichende Untersuchungen. Stuttgart. 1875. 11. Kamienski, F. Vergleichende Untersuchungen iiber die Ent- wickelungsgeschichte der Utricularien. Bot. Zeit. 35: 761-775. pi. U. 1877. 12. Strasburger. E. Befruchtung und Zelltheilung. Jena. 1877. 13. Hegelmaier. F. Vergleichende Untersuchungen iiber Entwicke- lung dicotyledoner Keime. Stuttgart. 1878. 14. Koch, L. Ueber die Entwickelung des Samens der Orobanchen. Jahrb. Wiss. Bot. 11: 218-261. pis. 8-10. 1878. 15. Solms-Laubach, H. Graf zu. Ueber monocotyle Embryonen mit scheitelbiirtigen Vegetationspunkt. Bot. Zeit. 36: 65-74, 81-93. figs. 37. 1878. 16. Strasburger, E. Ueber Polyembryonie. Jenaisch. Zeitsch. Naturwiss. 12: 647-670. 1878. 17. Famintzin. A. Embryologische Studien. Mem. Acad. Imp. Sci. St. Petersbourg 26 : 10. 1879. 18. Treub, M. Notes sur Tembryogenie de quelques Orchidees. Na- turkund. Verhandl. Konigl. Akad. Deel 21 : 1879. v- THE EMBRYO 223 19. Strasburger, E. Einige Bemerkungen iiber vielkernige Zellen uud iiber Embryogenie von Lupinus. Bot. Zeit. 38 : 845-854, 857-868. figs. 64. 1880. 20. Ward, H. Marshall. On the Embryo-sac and Development of Gymnadenia conopsea. Quart. Jour. Micr. Sci. 20 : 1-18. pis. 1-3. 1880. 21. Guignard, L. Recherches d'embryogenie vegetale eomparee. I. Legumineuses. Ann. Sci. Nat. Bot. VI. 12: 5-166. pis. 1-8. 1881. 22. Treub, M. Observations sur les Loranthacees. Ann. Sci. Nat. Bot. VI. 13: 250-282. pis. 13-20. 1882; reprinted in Ann. Jard. Bot. Buitenzorg 3: 1-12. pis. 1-2. 1883, and 2: 54-76. pis. 8-15. 1885. 23. Jonsson, B. Polyembryonie hos Trifolium pratense L. Bot. Central bl. 16: 171. 1883. 24. Treltb, M. Notes sur l'embrj-on, le sac embryonnaire. et Tovule. I and II. Ann. Jard. Bot. Buitenzorg 3 : 76-87. pis. 13-15. 1883. 25. . Notes sur Tembryon, le sac embryonnaire, et l'ovule. Ill aud IV. Ann. Jard. Bot. Buitenzorg 3: 120-128. pis. 18-19. 1883. 26. . Observations sur les Loranthacees. IV. Ann. Jard. Bot. Buitenzorg 3: 184-190. pis. 28-29. 1883. 27. . Notes sur I'embryon, le sac embryonnaire. et Tovule. V. Ann. Jard. Bot. Buitenzorg 4: 101-106. pi. 8. 1884. 28. Goebel, C. Outlines of Classification and Special Morphology. English translation. Oxford. 1887. 29. Oliver, F. W. On the Structure, Development, and Affinities of Trapella, a New Genus of Pedalineae. Annals of Botany 2 : 75- 115. pis. 5-9. 1888. 30. . On Sarcodes sanguined. Annals of Botany 4: 303-326. pis. 17-21. 1890. 31. Dodel, A. Beitrage zur Kenntniss der Befruchtungserscheinungen bei Iris sibirica. Festschrift (Nageli und Kolliker). Wiirzburg. 1891. 32. Overton, E. Beitrag zur Kenntniss der Entwickelung und Ver- einigung der Geschlechtsprodukte bei Li li urn Martagon. Fest- schrift (Nageli und Kolliker). Wiirzburg. 1891. 33. Chauveaud, G. L. Sur la fecondation dans les cas de polyem- bryonie. Compt. Rend. 114: 504-506. 1892. 34. Mottier, D. M. On the Embryo-sac and Embryo of Senecio au- reus L. Bot. Gazette 18 : 245-253. pis. 27-29. 1893. 35. Jeffrey, E. C. Polyembryony in Erythronium americanum. Annals of Botany 9: 537-542. pi. 19. 1895. 36. Tretjakow, S. Die Betbeiligung der Antipoden in Fallen der Polyembryonie bei Allium odorum L. Ber. Deutsch. Bot. Gesell. 13 : 13-17. pi. 2. 1895. 37. Hanausek. T. F. Ueber symmetrische und polyembryonische Samen von Coffea arabica L. Ber. Deutsch. Bot. Gesell. 13 : 73-78. pi. 6. 1895. 224 MORPHOLOGY OF ANGIOSPERMS 38. Humphrey, J. E. The Development of the Seed in Scitaniineae. Annals of Botany 10: 1-40. pis. 1-4. 1896. 39. Schaffxer. J. H. The Embryo-sac of Alisma Plantago. Bot. Gazette 21 : 123-132. pis. 9-10. 1896. 40. Schwere, S. Zur Entvvickelungsgeschichte der Frucht von Ta- raxacum officinale Web. Ein Beitrag zur Embryologie der Compositen. Flora 82 : 32-66. pis. 2-5. 1896. 41. Campbell, D. H. A Morphological Study of Naias and Zanni- chellia. Proc. Calif. Acad. Sci. III. 1: 1-62. pis. 1-5. 1897. 42. Chamberlain. C. J. Contribution to the Life History of Salix. Bot. Gazette 23 : 147-179. pis. 12- 18. 1897. 43. Schaffxer. J. H. Contribution to the Life History of Sagittaria variabilis. Bot. Gazette 23 : 252-273. pis. 20-26. 1897. 44. Coulter, J. ML Contribution to the Life History of Lilium phil- adelphicum. Bot. Gazette 23: 412-422. pis. 32-34. 1897. 45. Hegelmaier. F. Zur Kenntniss der Polyembryonie von Allium odovum. Bot. Zeit. 55 : 133-140. 1897. 46. Treub, M. L'organe feinelle et 1'apogamie du Balanophora elongata. Ann. Jard. Bot. Buitenzorg 15: 1-22. pis. 1-8. 1898. 47. Campbell, D. H. The Development of the Flower and Embryo in Lilaea subulata HBK. Annals of Botany 12: 1-28. pis. 1-3. 1898. 48. Coulter. J. M. Contribution to the Life History of Ranunculus. Bot. Gazette 25 : 73-88. pis. 4~'<. 1898. 49. G-ANONG, W. F. Upon Polyembryony and its Morphology in Opuntia vulgaris. Bot. Gazette 25: 221-228. pi. 16. 1898. 50. Elmore, C. J. Some Eesults from the Study of Allium. Bot. Gazette 26: 277-278. 1898. 51. Riddle, Lumina C. The Embryology of Alyssum. Bot. Gazette 26 : 314-324. pis. 20-28. 1898. 52. Osterwalder. A. Beitrage zur Embryologie von Aeon it um Xa- pellus. Flora 85 : 254-292. jjIs. 11-15. 1898. 53. Juel, H. O. Parthenogenesis bei Antenuaria alpina (L.) R. Br. Bot. Centralbl. 74: 369-372. 1898. 54. Caldwell. O. W. On the Life History of Lemna minor. Bot. Gazette 27 : 37-66. figs. 59. 1899. 55. Campbell, D. H. Notes on the Structure of the Embryo-sac in Sparganium and Lysichiton. Bot. Gazette 27 : 153-166. pi. 1. 1899. 56. . Die Entwickelung des Embryosackes von Peperomia pel- lucida Kunth. Ber. Deutsch. Bot. Gesell. 17: 452-456. pi. 31. 1899; also, A Peculiar Embryo-sac in Peperomia pellucida. Annals of Botany 13: 626. 1899. 57. Lloyd. F. E. The Comparative Embryology of the Rubiaceae. Bull. Torr. Bot. Club 28: 1-25. p/.s. 1-3. 1899. 58. Lotsy. J. P. Balanophora globosa Jungh. Eine wenigstens ortlich-verwittwete Pflanze. Ann. Jard. Bot. Buitenzorg II. 1 : 174-186. pis. 26- 29. 1899. THE EMBRYO 225 59. Campbell, D. H. Studies on the Araceae. Annals of Botany 14: 1-25. pis. 1-3. 1900. 60. Hill, T. G. The Structure and Development of Triglochin mart- timum L. Annals of Botany 14: 83-107. pis. 0-7. 1900. 61. Merrell, W. D. A Contribution to the Life History of Silphium. Bot. Gazette 29 : 99-133. pis. 3-10. 1900. 62. Johnson, D. S. On the Endosperm and Embryo of Peperomia pellucida. Bot. Gazette 30: 1-11. pi. l. 1900. 63. Wiegand, K. M. The Development of the Embryo-sac in some Monocotyledonous Plants. Bot. Gazette 30: 25-47. pis. 6-7. 1900. 64. Chodat, R., and Bernard, C. Sur le sac embryonnaire de VHe- losis guayanensis. Jour. Botanique 14: 72-79. pis. 1-2. 1900. 65. Cannon, W. A. A Morphological Study of the Flower and Em- bryo of the Wild Oat, Avena fatua. Proc. Calif. Acad. Sci. III. 1 : 329-364. pis. 49-53. 1900. 66. Juel, H. O. Yergleichende Untersuchungen iiber typische und parthenogenetische Fortpflanzung bei der Gattung Anteunaria. Handl. Svensk. Vetensk. Akad. 33 : no. 5. pp. 59. p/s. 0. figs. 5. 1900 ; reviewed in Bot. Zeit. 59 : 131. 1901. 67. Goebel, C. Bemerkung zu der vorstehenden Mittheilung (Mo- bius : Parasitismus und sexuelle Reproduktion im Pflanzen- reiche). Biol. Centralbl. 20: 571-572. 1900. 68. Guignard. L. La double fecondation dans le Naias major. Jour. Botanique 15: 205-213. figs. 11 1901. 69. Ernst, A. Beitrage zur Kenntniss der Entwickelung des Embryo- sackes und des Embryo (Polyembryonie) von Tulipa Gesne- riana L. Flora 88: 37-77. pis. 4~S. 1901. 70. Billings. F. H. Beitrage zur Kenntniss der Samenentwickelung. Flora 88: 253-318. 1901. 71. Holferty, G. M. Ovule and Embryo of Potamogeton natans. Bot. Gazette 31 : 339-346. p>ls. 2-3. 1901. 72. Schaffner, J. H. A Contribution to the Life History and Cytol- ogy of Erythronium. Bot. Gazette 31 : 369-387. pis. 1,-9. 1901. 73. Leavitt. R G. Notes on the Embryology of some New England Orchids. Rhodora 3 : 202-205. pi. 33. 1901. 74. Lyon, H. L. Preliminary Note on the Ernbryogeny of Nelumbo. Science 13: 470. 1901. 75. . Observations on the Ernbryogeny of Nelumbo. Minn. Bot. Studies 2: 643-655. pis. 48-50. 1901. 76. Murbeck. S. Parthenogenetische Embrvobildung in der Gattung Alchemilla. Lunds Univ. Arsskrift 36- : no. 7. pp. 46. pis. 0. 1901; reviewed in Bot. Zeit. 59: 129. 1901. 77. . Ueber das Verhalten des Pollenschlauches bei Alchemilla arvensis und das Wesen der Chalazogamie. Lunds Univ. Arsskrift 36 : pp. 19. pis. 2. 1901. 220 MORPHOLOGY OF AXGIOSPERMS 78. Smith, Amelia C. Tlie Structure and Parasitism of Aphyllon uniflorum Gray. Contrib. Bot. Lab. Univ. Penn. 2: 111-121. pis. 13-15. 1901. 79. Hegelmaier, F. Ueber einen neuen Fall von liabitueller Poly- embryonie. Ber. Deutscli. Bot. Gesell. 10: 488-499. 1901. 80. Balfour, I. B. Tbe Angiosperms. Address to the Botanical Sec- tion, Brit. Assn. Adv. Sci. Glasgow. 1901. 81. Conard, H. S. Note on the Embryo of Xymphaea. Science 15: 316. 1902. 82. Hall, J. G. An Embryological Study of Limnocharis emargi- nata. Bot. Gazette 33 : 214-219. pi. 9. 1902. 83. Overton, J. B. Parthenogenesis in TJialictrum purpurascens. Bot. Gazette 33 : 363-375. pis. 12-13. 1902. 84. Strasburger, E. Em Beitrag zur Kenntniss von Ceratophyllum submersum und phylogenetische Erorterungen. Jahrb. \Viss. Bot 37 : 477-526. pis. 9-11. 1902. 85. Lloyd, F. E. The Comparative Embryology of the Rubiaceae. Mem. Torr. Bot. Club 8: 27-112. pis. 5-15. 1902. 86. Johnson, D. S. On the Development of Certain Piperaceae. Bot. Gazette 34 : 321-340. pis. 9-10. 1902. 87. Pechoutre, F. Contribution a l'etude du developpement de l'ovule et de le graine des Eosacees. Ann. Sci. Nat. Bot. YIIL 16: 1-158. figs. 166. 1902. 88. Lyon, H. L. The Phylogeny of the Cotyledon. Postelsia 1901 : 55-86. 1902. 89. Sargant, Ethel. The Origin of the Seed-leaf in Monocotyledons. The New Phytologist 1 : 107-113. pi. 2. 1902. 90. Murbeck, S. Ueber Anomalien im Baue des Nucellus und des Embryosackes bei parthenogenetischen Arten der Gattung Al- chemilla. Lunds Univ. Arsskrift 383 : no. 2. pp. 10. pis. 13. 1902. 91. Schmid, B. Beitriige zur Embryo-Entwickelung einiger Dicotylen. Bot. Zeit. 60 : 207-230. pis. 8-10. 1902. 92. Treub, M. L'organe femelle et Tembryogenese dans le Ficus hirta Vahl. Ann. Jard. Bot. Buitenzorg II. 3 : 124-157. pis. 16- 25. 1902. 93. Juel, H. O. Zur Entwicklungsgeschichte des Samens von Cyno- morium. Beih. Bot. Centralbl. 13 : 194-202. figs. 5. 1902. 94. Sargant, Ethel. A Theory of the Origin of Monocotyledons, founded on the Structure of their Seedlings. Annals of Botany 17: 1-92. pis. 1-7. 1903. 95. Murbeck, S. Ueber die Embryologie von Ruppia rosteWata Koch. Handl. Svensk. Vetensk. Akad. 36: pp. 21. pis. 3. 1902. CHAPTER X CLASSIFICATION OF MONOCOTYLEDONS A satisfactory classification of Angiosperms still remains an impossible task. The immense number of species and their entanglement of relationships, as well as our merely superficial knowledge of the great majority of forms, have made progress toward a natural classification very slow. Since the time of John Ray (1703) steps in this progress have been taken by De Jussieu (1789), De Candolle (1819), Endlicher (1836- 1840), Brongniart (1813), Braun (1861), Bentham and Hooker (1862-1883), Eichler (1883), Engler (1892), and others. Xaturally, the increasing knowledge of morphology and the changed conception of species have gradually broken up artificial assemblages, but much of classification is still arti- ficial. It does not lie within the purpose of this book to trace the historical development of classification, nor to present an- other scheme for consideration. We merely adopt the classi- fication of Eichler as modified by Engler, and elaborated in Engler and Prantl's Die Natiirlichen Pflanzenfamilien, as the best expression of our present knowledge of morphology as applied to the whole of Angiosperms. The special student of morphology must have enough knowledge of general relation- ships to enable him to select critical forms for investigation and to appreciate the bearings of his results. The purpose of the following presentation. Therefore, is to trace in a general way the evolution of Angiosperms and to point out the greatest gaps in knowledge, using the classification mentioned as the besl available basis. Xo attempt is made to use the varying termi- nology of the larger groups of classification, but coordinate groups are indicated by common endings. According to Engler, the general tendency among Monocot- 327 228 MORPHOLOGY OF ANGIOSPERMS yledons is to advance from naked flowers with parts spirally arranged and indefinite in number to pentacyclic trimerous flowers. There are also such lines of advance as from apocarpy to syncarpy, from hypogyny to epigyny, from actinomorphy to zygomorphy, etc. These tendencies are often very unequally expressed even by different groups of the same alliance, one group developing chiefly along one line, and another group along another line, so that the results are very different. It is also often a question whether a simple floral structure is primi- tive or reduced. In the older morphology there was a typical floral structure, and all simpler ones were regarded as reduced forms. There can be no doubt that there are reduced floral structures, as in Lemuel ; but the great majority of simple flowers are probably primitive. Upon these and other considerations, Engler has subdivided the Monocotvledons into ten great alliances. The first six con- stitute the more primitive Spiral series, and although the trim- erous habit appears among them, the spiral arrangement and indefinite numbers occur in one or more sets. The remaining; four alliances constitute the Cyclic series, the highly specialized Monocotyledons. I. Paxdaxales. — This includes the Pandanaceae, Typha- ceae, and Sparganiaceae, together containing a little more than 100 species. The Pandanaceae (about SO species), or screw- pines, belong to the oriental tropics, chiefly the coasts and is- lands of the Indian and Pacific oceans ; while the other families are mainly represented in temperate regions. That these forms are primitive Monocotyledons is indicated by the following facts: there is nothing to represent a perianth unless the floral bracts of Sparganium be regarded as one: the sporophylls are mostly spiral and indefinite in number, the sta- mens of Pandanaceae often being very numerous and exhibiting the greatest variation in arrangement ; the species are all hydro- phytic; and the plants are anemophilous. Such flowers as those of the Pandanaceae and Typhaceae are extremely simple, the peculiar hairs accompanying the sporophylls of the latter ap- parently representing sterile sporophylls: while the Spargania- ceae are the most advanced members of the alliance, a perianth probably being represented by a set of small bracts, and the trimerous character appearing. CLASSIFICATION OF MONOCOTYLEDONS 229 A well-marked feature of the group is the protection of the flower-clusters by a prominent leaf-sheath. The development of this sheath as a protecting organ before the appearance of a fully developed perianth is one of the constant features of the more primitive Monocotyledons, and in some of the following groups it becomes highly specialized. The hydrophytic Pandanales, therefore, begin in the great- est simplicity, so far as floral structures are concerned, the Pandanaceae being the most primitive forms on account of the indefinite number of the sporophylls and the spiral arrangement of the stamens, and the series has not advanced verv far. It should be remembered, however, that the three existing families probably represent fragments of a formerly much larger alli- ance, so that the association of the temperate Typha and Spar- ganium with the tropical Pandanaceae may not be so unnatural in reality as it appears at present. It is extremely desirable to obtain some accurate knowledge of the essential morphology of the Pandanaceae. II. Helobiales. — This includes the Potamogetonaceae, Xaiadaceae, Aponogetonaceae, Juncaginaceae, Alismaceae, Bu- tomaceae, and Hydrocharitaceae, together containing about 235 species. Engler has set apart the small family Triuridaceae, containing about 18 species, as representing a distinct series, Triukidales, but this can be disregarded in this very general presentation. This is one of the most remarkable of the monocotyledonous lines in its extent, reaching from the greatest floral simplicity in Potamogetonaceae to highly developed flowers in Hydro- charitaceae. It has been called an unstable or plastic line, and may have given rise to higher forms ; in any event it is probably to be regarded as one of the most important phylogenetic lines among the Monocotyledons. For this reason morphological investigation in recent years has specially cultivated this series of forms, particularly the more primitive families. About the only taxonomic character that holds these diverse forms together is the fact that they are exceptional among Monocotyledons in the feeble development of endosperm. They are characteris- tically aquatic, and sheathing bracts enclosing the flower-clus- ters are largely developed. In most of the forms the spiral arrangement and indefinite number of floral parts is very appar- 230 MORPHOLOGY OF AXGIOSPERMS ent, but the line as a whole presents almost a complete series from the simplest floral structure to one of the most highly developed. The series of floral changes may be broadly indicated as follows. In Potamogetonaceae and ^saiadaeeae there is no peri- anth, and the stamens and carpels are indefinite in number ; in Juncaginaceae a bract-like perianth is present, there is a dis- tinct tendency toward the trimerous habit, and svncarpv may occur ; in Alismaceae the perianth is differentiated into calyx and corolla, and the trimerous tendency is very clear, though the carpels are usually indefinite in number ; in Hydrocharita- ceae, in addition to a differentiated perianth and a strong ex- pression of the trimerous tendency (although the stamens and carpels are often indefinite in number), the flowers are epigy- nous. The plants are chiefly anemophilous or hydrophilous, but the appearance of a differentiated perianth in the Alisma- ceae is probably associated with a certain amount of ento- mophily. Heliobales, therefore, begin with as great simplicity of floral structure as do the Pandanales, but thev have advanced much further in floral development. That such an extensive line comprises so few species is probably associated with the uniformity of aquatic conditions. In the whole series, how- ever, there is no distinct settling into a complete trimerous habit, which is intimated rather than established. III. Glumai.es.* — In this alliance are the two great fami- lies Gramineae and Cyperaceae, the former including about 351 genera and 4,700 species, the latter 76 genera and about 2,300 species. In point of species this is one of the greatest of angiospermous alliances, and in display of individuals it is un- questionably the greatest. The common features of the two families are the absence of a perianth, the protection of the flowers by special bracts, the fluctuating of the stamens between one and many, the solitary carpel, and anemophily. It is not probable that the two families are related to one another genet- ically, but they represent approximately the same stage of floral development. The peculiar features of the bract-protection, as contrasted * G-lumiflorae of Engler. CLASSIFICATION OF MONOCOTYLEDONS 231 with the preceding alliances, is that the bract Joes not ensheath a whole flower-cluster but individual flowers. It is this charac- teristic bract {glume, palet) that gives name-to the alliance. The lodicules of Graniineae and certain hairs and bracts of Cyperaceae are regarded by some as representing a perianth. Even if this doubtful claim be allowed, such a perianth is better regarded as one that is very primitive rather than re- duced. The primitive character of Glumales is indicated by the characters given above, but contrasted with the Helobiales it is a rigid group that has not advanced far in floral development, but has proved to be a remarkably successful type of vegeta- tion. Moreover, it is the primitive group of Monocotyledons that seems to have been the first to establish itself upon the drier and more diversified land surface, and this fact may hold some relation to its structural stability and its great display of species. Evidence of its aquatic origin may be obtained not only from the numerous hydrophytic forms, but also from ana- tomical characters that relate it to Helobiales and Pandanales rather than to the terrestrial alliances. Pandanales, Helobiales, and Glumales are the only three alliances of Monocotyledons that -include the most primitive type of monocotyledonous floral structure. Their possible ge- netic relation to one another is entirely obscure, and in their present display they seem to emerge from the beginnings of the history of Monocotyledons as independent lines. The remain- ing seven alliances are either derived from these three, or their primitive members have disappeared. IV. Palmales.* — The palms are the chief representatives of monocotyledonous trees, and are characteristic of all tropical regions. The single family Palmaceae includes about 150 gen- era and 1,100 species, though these numbers will doubtless be much increased when the palms are studied in their habitats. A knowledge of the essential morphology of this group is also much to be desired. A perianth is always present, although it is very " rudimen- tary " and hence doubtful in Phytelephas and Coryphanthe, but it is not differentiated into a distinct calyx and corolla. * Principes of Engler. 16 232 MORPHOLOGY OF ANGIOSPERMS As there are no naked flowers, this group does not have as primitive members as do the three preceding ones. The sta- mens are extremely variable in number, ranging from three to indefinitely numerous, showing the primitive spiral charac- ter; while the carpels are usually three and sometimes form a syncarpous pistil. The enormous flower-cluster is ensheathed by a great bract (spathe) that is more or less tough and even woody, a feature recalling the same tendency in Pandanales and Helobiales. As the axis of inflorescence is sometimes thickened and the flowers more or less embedded in it, the inflorescence is often spoken of as a branching spadix. These characters indicate a group as a whole considerably further advanced than the preceding ones in the constant pres- ence of a definite perianth, although it is undifferentiated. The association of floral envelops with a spathe is of interest, but in such conditions a highly developed perianth could not be expected. While there is doubtless anemophilous pollination, entomophily must exist to a certain extent. The whole struc- ture suggests one that is intermediate between the dominance of bract and perianth, between anemophily and entomophily. Palmales, therefore, differ from Glumales in the definite trimerous perianth, as well as in numerous other features ; from the Helobiales in that the number of carpels is constant; but through Phytelephas and Coryphanthe, with their rudimen- tary perianth, as well as through general habit, the connection with Pandanales seems clear. It seems probable) therefore, that the Palmales have been derived from the Pandanales, sur- passing the Glumales in floral development, but not reaching the differentiation of calyx and corolla and epigyny attained by the higher members of the Helobiales. V. Syxaxtiiai.ks." — Thi> includes a small family (Cy- clanthaceae) of the American tropics, represented by about 45 species, and usually and naturally associated with the screw- pines and palms. The flowers are in an unbranched spadix, either scattered or in a close spiral, and there is generally an evident bract-like perianth in one or two cycles. The stamens range from six to indefinitely numerous, and the carpels are one to four. Tn the staminate flowers there is no trace of car- * Syxaxthae of Enpler. CLASSIFICATION OF MONOCOTYLEDONS 233 pels and the stamens are connate ; while in the carpellate flowers there are very conspicuous and often branching staminodia. There is a strong tendency to " coalescence " in all the members, the perianth often being tubular, the stamens usually connate, and the carpels (if more than one) always forming a syncar- pous pistil. The group is also peculiar in the very numerous ovules upon a single parietal placenta. Too little is known of the morphology of the group to speak of its relationships with any definiteness, but it seems safe to regard it as another branch of the Pandanales stock. The Pan- danales, Palmales, and Synanthales are thus referred to a com- mon origin, with the Pandanales as the most primitive repre- sentative of the stock. This tropical association seems to be a strange one for Typha and Sparganium, but otherwise it seems to be entirely natural, and not clearly related to any other ]\Iono- cotvledons. VI. Aeales.* — This includes the Araceae with about 1,000 species, and the Lemnaceae with about 25. The Aroids form one of the most distinct and also diversified groups of Monocot- yledons. The characteristic features are the spadix, the highly developed spathe, and the broad net-veined leaves. There is also probably greater anatomical differentiation than in any other monocotyledonous group, which is taken advantage of in their classification. The floral structure is of three general types: (1) the Calamus type, in which the flowers are bisporan- giate, pentacyclie, 2 to 4-merous, and syncarpous ; (2) the Calla type, in which the flowers are bisporangiate, with no perianth, 6 to '•» stamens, and 1 carpel; (3) the Arum type, in which the flowers are monosporangiate (staminate flowers above and • •arpellate flowers below on the same spadix), and with no perianth. It is evident that the floral structure is extremely fluctua- ting, and that this is probably associated with the extreme spe- cialization of the spathe. Engler has called attention to the fact that the flowers with a perianth are associated with a bract-like spathe; while those without a perianth (the great majority) are associated with a petaloideous spathe. In any event, the bract reaches its highest specialization in this group, * Spathiflorae of Engler. 234 MORPHOLOGY OP ANGIOSPERMS being not merely a protecting organ, but immensely varied in form, texture, and color to secure entomophily. In other words, the conspicuous function of the perianth in the petaloideous groups is here assumed by the spathe, and the flowers retain for the most part the primitive character. There are many features of the Aroids that suggest the He- lobiales, especially the Potainogetonaceae, so that Engler inclines to the belief that they have been derived from that stock. If this be true, they represent a strong terrestrial branch from the aquatic Helobiales, that in tropical conditions has become extremely varied in form and structure, and that has assumed the erect, climbing, and epiphytic habits. It does not seem probable that any other monocotyledonous alliance is asso- ciated with these two in origin ; but the suggestion has been made that from the Aroids the Dicotyledons, or at least some of their phyla, may have been derived. One of the most prom- ising fields of morphological research is among the tropical Aroids. The Lemnaceae represent a distinct reduction series, being Aroids adapted to the free-swimming habit, and remarkably reduced in structure, Wolffia being the smallest known seed- plant. The six great alliances just considered constitute the Spiral series of Engler, with inconstant number of floral members, with mostly no perianth or one not adapted to entomophily, and with a striking development of sheathing leaves or bracts in connection with the inflorescence or the individual flowers. The four remaining alliances constitute the Cyclic series, in which the almost constant floral formula is perianth 3 + 3, stamens 3 + 3, carpels 3 and forming a syncarpous pistil. The two perianth sets may be variously modified, but there runs through the series an increasing specialization of the perianth for entomophily, which reaches its extreme expression in the Orchidaceae. As a consequence, the perianth rather than bracts becomes the conspicuous floral feature. The pentacy- clic trimerous habit having become established, the cyclic groups have largely differentiated in the direction of a conspicuous perianth, epigyny, and zygomorphy. The number of species involved is so great that onlv the broadest outlines can be con- sidered. CLASSIFICATION OP MONOCOTYLEDONS 235 VII. Fablnales.* — The eleven families of this alliance are Flagellariaceae, Restionaceae, Centrolepidaceae, Mayacaceae, Xyridaceae, Eriocanlaceae, Rapateaeeae, Bromeliaceae, Com- melinaceae, Pontederiaceae, and Philydraceae, together contain- ing a little more than 2,000 species. The large families are Bromeliaceae with over 900 species, Eriocaulaceae with 460, Commelinaceae with more than 300, and Restionaceae with nearly 250. The chief character that holds these diverse fami- lies together and separates them from the Liliales is the thin- walled endosperm rich in starch, whose cells become easily broken up and dissociated, resulting in a " mealy " or " crum- bly " endosperm. From the evolutionary standpoint the following facts are of importance: for the most part the forms are grass-like herbs, with all habits from aquatic to xerophytie and epiphytic ; they are mostly bracteate forms, the upper bracts showing a decided tendency to ensheath the inflorescence ; thev are mostlv ane- mophilous, but some forms have a perianth adapted to ento- mophily ; the perianth ranges from scarious to petaloid, from undifferentiated to a distinct calyx and corolla, from polypetaly to sympetaly ; the flowers are syncarpous and, with the excep- tion of a fewr Bromelias, hypogynous. Such evidence indicates a relatively primitive cyclic alli- ance with many characters recalling the spiral forms, the bract-protection and anemophily not being definitely replaced by a highly developed perianth and entomophilv. The origin of the series is of course obscure, but the evidence seems to favor the Glumales as the original stock. As illustrating the construction of a natural sequence of families, those of this alliance may be used as follows: The Flagellariaceae, Restionaceae, and Centrolepidaceae, belonging to the oriental tropics chiefly of the Southern Hemi- sphere, have a bracteate undifferentiated perianth and are ane- niuphilous, in habit and general character resembling the Spiral series. The Mayacaceae, Xyridaceae, and Eriocanlaceae have a dif- ferentiated calyx and corolla, and orthotropus ovules with very small embryos. These three famili* -. together with Restiona- * Farixosae of Engler. 236 MORPHOLOGY OF ANGIO!>PERMS ceae and Centrolepidaceae, constitute the main part of the old group Enantioblastae, characterized by the orthotropous ovules. The Rapateaceae, chiefly South American, have a distinct calyx and corolla, anatropous ovules, and small embryos. The Bromeliaceae, the great epiphytic family of the Ameri- can tropics, have a distinct calyx and corolla, anatropous ovules, and larger elongated embryos. The Commelinaceae, in addition to the distinct calvx and corolla, show a tendency to zygomorphy. This family has the orthotropous ovules and small embryos of the Enantioblastae, but the characters given, as well as the habit and inflorescence, seem to forbid that alliance. The Pontederiaceae and the Australasian Philvdraceae have long cylindrical embryos, a general tendency to a reduced number of stamens and carpels, and in the latter family sym- petaly. VIII. Liliales.* — The nine families of this alliance are Juneaceae, Stemonaceae, Liliaceae, Haemodoraceae, Amarylli- daceae, Velloziaceae, Taccaceae, Dioscoreaceae, and Iridaceae, together comprising almost 5,000 species. The largest families are Liliaceae with nearly 2,500 species, Iridaceae with more than 1,000, and Amaryllidaceae with nearly 900. This great alliance may be regarded as containing the typ- ical highly developed Monocotyledons. It is characterized by a conspicuous development of the perianth and a prevailing entomophilous hain't. The endosperm cells are thick-walled and in general contain oil rather than starch, resulting in an endo- sperm that is not " mealy," as in the Farinales. The Junea- ceae, Haemodoraceae, and Velloziaceae are exceptions in pro- ducing a starch-containing endosperm, but the cells do not be- come dissociated. In passing from the lower members of the series to the higher there is a transition from an undifferenti- ated scarious perianth to a differentiated and petaloideous one; and from hypogyny to epigyny, the four lower families being hypogynous and the five higher epigynous. The sequence of families begins with the Juneaceae, which with their grass-like habit, scarious perianth, and starchy en- dosperm, may be fairly regarded as intermediate between Fari- * Liliiflorae of Engler. CLASSIFICATION OF MONOCOTYLEDONS 237 nales and Liliales. The Liliales are midway in the series, hav- ing attained a petaloideons perianth and entomophily, and having become so diversified in structure and habit as to raise a question as to their monophyletic origin. The Amaryllida- ceae introduce epigyny, and the highly specialized Iridaceae complete the series. The last six families are in great need of morphological investigation in the tropics where they are chiefly massed. The genetic connection between Liliales and Farmales seems clear, so that if the latter are regarded as derived from the Glumales, the former must be referred to the same stock, probably dissociating early from the Farinales. The two remaining alliances are characterized by epigyny and zygomorphy, highly specialized entomophilous structures, reduction and modification of stamens, and very small and un- differentiated embryos. In all probability they are not genet- ically related, but they resemble one another more than they do the other alliances. IX. Scitamixales.* — The four families of this alliance are Musaceae, Zingiberaeeae, Cannaceae, and Alarantaceae, to- gether comprising nearly 800 species, 500 belonging to the Ziiiiiiberaceae. The four families are undoubtedlv 2,-eneticallv related, although the first two are restricted to the oriental tropics, and the last two to the occidental. In addition to the characters mentioned above, the replacing of functional sta- mens by petaloid staminodia is very characteristic, commonly only one stamen being functional and even this one being peta- loid. In nearly every case, also, there is a labelluin, formed either by the perianth or the staminodia. The habit of the vege- tative body, however, is most peculiar. The real stem is a rhi- zome, but the enormous leaves, differentiated into sheath, peti- ole, and pinnately veined blade, build up a false stem by means of their very large and closely overlapping sheaths. The temptation is to derive this alliance from the Dracaena region of the Liliaceae, but important anatomical features that are common to all four families are opposed to this view. That it is connected in some way with the Glumales-Farinales-Lili- ales stock seems most probable; and if so the general structures * Scitamineae of Engler. 238 .MORPHOLOGY OF ANGIOSPERMS indicate a separate origin from Glnmales. A morphological investigation of these families in the tropics is greatly to be desired. X. Orchidales.* — The two families of this alliance are Bnrmanniaceae and Orchidaceae, all but about 55 of the 7,000 species belonging to the latter family. These two unequal fami- lies are held together by the very numerous and small ovules and by the extreme zygomorphism of the flower, but the Bur- manniaceae have endosperm, often six stamens, and frequently connate perianth-segments, approaching the Amaryllidaceae. The chief interest of the alliance centers about the Orchi- daceae, the greatest monocotyledonous family in point of spe- cies and the most highly specialized. The epiphytic habit is extensively developed, and the terrestrial forms are mostly saprophytic or parasitic. These habits have resulted in the development of certain special structures, such as the bulbous leaf-bases and velamen of the epiphytic forms ; and in the sup- pression of some normal structures, as the primary root, and sometimes all roots. The absence of endosperm, the poorly developed embryo, and the extensive use of the suspensor as a remarkably developed haustorial organ are probably but addi- tional results of the unusual habits of the family. The notable floral structures are the modification of one of the petals to form the labellum and spur, the remarkable " gynostemium," the twisted ovary, and the pollinium-mechanism. As an illustration of the varying modifications of floral structure, the ordinary orchid may be compared with the Cy- pripedium type. The flowers are pentacyclic, and the cycles are developed in the two types as follows, beginning with the outermost. In both types the first cycle consists of three sepals, and the second of three petals, the posterior (made anterior by the twisting of the ovary) forming the labellum and spur. In the third cycle two lateral stamens are suppressed in both types, but in ordinary orchids the anterior one is functional, while in Cypripedium it is replaced by a staminodiuin. In the fourth cycle the posterior stamen is suppressed in both types, but in ordinary orchids the two laterals are replaced by staminodia, while in Cypripedium they are functional stamens. In the * MrcRosPERMAE of Engler. CLASSIFICATION OP MONOCOTYLEDONS 239 fifth cycle in ordinary orchids the two lateral carpels form the stigma, the anterior producing the disk-bearing " rostelhim," while in Cypripedium all three carpels form the stigma. The origin of the Orchidaceae is very obscure. It is com- mon to regard them as derived from the Liliales, but there are many objections to this hypothesis. In any event, it seems most natural to refer them to the same general stock. According to the views presented in this chapter, there are three primitive monoeotyledonous stocks — Pandanales, Helobi- ales, and Glumales — and thev are connected with the other alliances as follows: Pandanales-Palmales-Synanthales; Helo- biales-Arales; Glumales-Farinales-Liliales-Scitaminales-Orchi- dales. CHAPTER XI CLASSIFICATION OF ARCHICHLAMYDEAE Two great divisions of Dicotyledons are evident, the Arehi- chlamydeae and Sympetalae, although there is no sharp distinc- tion between them. Sympetalous forms among the former and polypetakms forms among the latter occur, bnt in the main apetaly or polypetaly is a distinctive feature of the Arehichla- mydeae, and sympetaly of the Sympetalae. That the Archi- chlamydeae include the most primitive Dicotyledons is clear, but what forms are to be regarded as the most primitive is open to discussion. The classification of the Arehichlamydeae is an exceedingly puzzling problem, and the current schemes are far less definite and satisfactory than those for the classification of Monocotvle- dons and Sympetalae. Questions of primitive and reduced characters, and of relative rank on the basis of combination of characters, are particularly involved among Arehichlamydeae, and hence opinions vary widely as to the details of their classi- fication. The difficulties arise from tfie fact that the characters of the group are extremely fluctuating, not being established as among the Sympetalae. Add to this that more than 60,000 species * are recognized, over three times as numerous as the species of Monocotyledons, included in 180 families, and it becomes evident that the confusion of relationships is bewil- dering. Engler has arranged the Arehichlamydeae in twenty-six al- * The numbers of species given in this chapter must be regarded as approx- imate and conservative. They will vary with the increase of knowledge and t he conception of species, but in this chapter they are only intended to indi- cate the relative display of different types of structure. 240 CLASSIFICATION OF ARCHICHLAMYDEAE 241 liances, coordinate with the ten series of Monocotyledons. The general sequence of these alliances is based, as in Monocotyle- dons, upon the development of the perianth and of the floral axis, and the arrangement of floral members ; but other characters, chiefly those derived from the ovules, are also used to disen- tangle relationships. Of course there is no real sequence of these twenty-six alliances, for they represent, for the most part, parallel or divergent lines of development. The sequence of presentation is determined in the main by the relative advance- ment of the lower members of each alliance, whose higher mem- bers may or may not have made great advancement and in manv directions. Such an assemblage of forms may be conceived of as a tangled thicket, through which certain paths may be more or less evident, but in which no orderly arrangement is apparent. It would be confusing, even were it possible, to discuss the relationships of each of the twenty-six series. They can only be presented as assemblages of families that seem to be natural, perhaps not so much on account of their common origin as on account of their approximately equal grade of advancement, and hence " form-groups " rather than necessarily genetic groups. The following presentation of the alliances of Archichlamy- deae is largely based upon Engler's " Uebersicht iiber die Unter- abteilungen, Klassen, Reihen, Unterreihen, und Familien der Embryophyta siphonogama," published in Engler and Prantl's Die Natiirlichen Pflanzenfamilien in 1897 (Lieferung 165). The first twelve alliances are especially puzzling. Among them are evidently the most primitive forms in floral structure. They also include the chalazogamic forms, and ovules whose structure is unusual among Angiosperms. The families are practically those that were disposed of by Eichler as Amen- tiferae, together with miscellaneous groups of uncertain affinity. That the so-called Amentiferae or Amentaceae represented a heterogeneous assemblage of forms has lona; been evident. It is a question whether Engler's splitting up into alliances has not been excessive in this part of his scheme, certain morphological characters sometimes being used that may not prove to be of first importance. In any event, the splitting up will serve to keep apart distinct groups until they can be recombined natu- rally. There is no region of the Archichlamvdeae which has 212 MORPHOLOGY OF ANGIOSPERMS recently received more deserved attention from morpholbgists, and which still so greatly needs investigation. I. Casuarinales.* — This includes the single family Casu- arinaceae, containing about 25 species. Engler regard- the al- liance as the most primitive because the ovule develops numer- ous megaspores. This particular character can not be regarded as distinctive, since among the Fagales the same character, associated also with chalazogamy, occurs, and numerous mega- spores are found among the liana les, Rosales, etc. The low position, however, is justified by the primitive flowers, which are either naked or with a bract-like perianth. The next two alliances are regarded as relatively primitive on account of their naked flowers, together with the Casuari- nales being the only naked alliances. II. Pipekales. — This includes the Saururaceae, Pipera- ceae, Chloranthaceae, and Lacistemaceae, together containing about 1,150 species, of which about 1,100 belong to the Pipera- eeae. The results of the investigation of Peperomia pellucida by Campbell and by Johnson indicate that the tropical Pipera- ceae are probably most promising forms for morphological investigation, and are to be considered in any discussion as to the most primitive Dicotyledons. III. Salicales. — This includes the single family Salica- ceae, containing about 180 species. IV. Myricales. — This includes the single family Myrica- ceae, containing 10 species. The advance in floral structure is shown by the fact that the several bracts near the flower may be regarded as an extremely primitive perianth. V. Balaxopsidales. — This includes the single family Ba- lanopsidaceae, containing 7 species. This is an uncertain type, and raises the question of reduction. The staminate flowers have a rudimentary perianth and an indefinite number of sta- mens; and the carpellate flowers have a bracteate perianth. Engler calls attention, however, to the fact that there are no intermediate forms for a reduction series, and that the indefi- nite number of stamens is a primitive character. VI. Leitneriales. — This includes the single family Leit- neriaceae, containing 2 species. The primitive character' of this * Vi:RTiriLLATAE of Engler. CLASSIFICATION OF ARCHICHLAMYDEAE 243 type, with its flowers naked or with a bracteate perianth, is very doubtful. Engler states that if any evidence of reduction is obtained, this family would be included among the Rosales, near the Hamamelidaceae. VII. Juglaxdaees. — This includes the single family Ju- glandaceae, containing about 30 species. This alliance is dis- tinctly higher than the preceding ones in that there is nearly always a distinct perianth, which in the carpellate flowers is coalescent with the ovary, so that there is a resemblance to epigyny. Disregarding the Balanopsidales and Leitneriales as doubtful and possibly reduction alliances, the Juglandales are to be compared directly with the Myricales. The two were for- merly associated in a single alliance, but the distinct perianth, as well as chalazogamy, serve to distinguish the Juglandales. It is a question whether such differences, and the others asso- ciated with them, are incompatible in a single alliance. VIII. Fagales. — This includes the Betulaceae and Fa°a- ceae, together containing about 420 species, nearly 350 of which belong to the Fagaceae. This is a parallel alliance with Juglan- dales, having a distinct but bracteate perianth, which in the carpellate flowers is more or less coalescent with the ovary. Among Betulaceae, also, chalazogamy occurs, as in Juglanda- ceae and Casuarinaceae. IX. Ukticai.es. — This includes the Ulmaceae, VEoraceae, and Urticaeeae, together containing about 1,560 species, the large families being Aloraceae with about 920 species, and the Urticaeeae with about 520. This is an alliance parallel with the Juglandales and Fagales, with the distinct and bracteate perianth, which, as in Fagales, is definitely cyclic. X. Proteales. — This includes the single great Australasian family Proteaceae, with about 950 species. In this alliance the next stage in the development of the cyclic perianth become- evident. Although it is sometimes green and bract-like, in the majority of cases it is petaloid, but there is no differentiation of calyx and corolla. A character used to distinguish this alliance from the following is the single carpel with well-developed ovule. XL Saxtai.ai.ks. — This includes the Loranthaeeae, aIvzo- dendraceae, Santalaceae, Grubbiaceae, Opiliaeeao, Olacaceae, and Balanophoraceae, together containing about 1,260 specie-. 244 MORPHOLOGY OF ANGIOSPERMS the large families being Loranthaceae with 800 species, Santa- laeeae with 246, and Olacaceae with 150. In this alliance, also, the cyclic perianth is for the most part petaloid, but there is advancement in the general differentiation of a calyx and co- rolla. For the most part, there is a syncarpous pistil of three carpels, but the carpels may be two or one ; and a free central placenta develops ovules without an integument or no distinct ovules at all. There is much diversity within the alliance, at least three distinct lines being evident ; but the rather remark- able morphological structures found in the alliance are prob- ably related to their general parasitic or semi-parasitic habits. XII. Aristolochiales. — This includes the Aristolochia- ceae, Rafflesiaceae, and Hydnoraceae, together containing about 235 species, of which 205 belong to the Aristolochiaceae. The members of this series are distinctly in advance of the preceding in the coalescence of the petaloid segments of the perianth, and especially in epigynv. The indefinite number of ovules is also a distinguishing feature. The preceding twelve alliances represent a primitive com- plex, in which reduced forms may have been included. How they may be related to one another in origin is too obscure for profitable discussion, but it seems probable that they are not at all related to the following alliances. In other words, whether they represent a single genetic stock or several, they appear to be isolated from the higher alliances. XIII. Polygonales. — This includes the single family Polygonaceae, with about 750 species. Its mostly cyclic flowers, with undifferentiated perianth or distinct calyx and corolla, puts it upon about the plane of advancement attained by the preceding alliances ; while its strong trimerous tendency and peculiar habit set it well apart. This is sometimes re- garded as a transition group between the preceding alliances and the Centrospermales. In any event, it may be regarded as fairly associated with the latter. XIV. Centrospermaees.* — This includes Chenopodia- ceae, Amarantaceae, Xyctaginaceae, Batidaceae, Cynocramba- ceae, Phytolaccaceao, Aizoaceae, Portnlacaceae, Basellaceae, and Caryophyllaceae, together containing about 3,320 species, the * Centrospermae of Engler. CLASSIFICATION OF ARCHICHLAMYDEAE 245 large families being Caryophyllaceae with 1,420 species, Aizoa- ceae with 575, and Chenopodiaceae and Amarantaceae each with about 435. In this alliance the floral characters range from the bracteate undifferentiated perianth of Chenopodiaceae to the distinct calyx and corolla of many Caryophyllaceae. In the alliance as a whole calyx and corolla are frequently rather than prevailingly distinct, and only the highest family has at- tained the conspicuous corolla associated with entomophily. A feature of the alliance is the conspicuous perisperm. The Polygonales and Centrospermales may possibly have a closely related origin, but it does not seem probable that they are related in any way to the following alliance, but that they represent a general line of development whose highest expres- sion is among the Caryophyllaceae. XV. Raxales. — This includes Xymphaeaceae, Ceratophyl- laceae, Trochodendraceae, Ranunculaceae, Lardizabalaceae, Ber- beridaceae, Menispermaceae,Magnoliaceae, Calycanthaeeae, Lac- toridaceae, Anonaceae, Myristicaceae, Gomortegaceae, Monimia- ceae, Lauraceae, and Hernandiaceae, together containing about 4,050 species, the large families being Lauraceae with 1,015 species, Ranunculaceae with 990, Menisperrnaceae with 390, Anonaceae with 345, Monimiaceae with 245, Alyristicaceae with 235, and Berberidaceae with 135. This great alliance introduces the prevailing habit of a dis- tinct calyx and corolla, and is characterized by the prevalence of apocarpy and hypogyny. The primitive character of the flower is indicated not only by apocarpy and hypogyny, but also by the strong tendency to the indefinite repetition and spiral arrangement of the floral members. Were it not for the preva- lence of a distinct calyx and corolla the alliance would not hold so high a rank. At least three prominent developmental lines are evident, viz., Xymphaeaceae to Ceratophyllaceae, Ranun- culaceae to ^Menispormaceae, and Magnoliaceae to Hernandia- ceae. In each of these lines there is an advance from the spiral to the cyclic arrangement, and in the last line epigyny is reached. As is also known, zvgomorphy occasionally occurs, being present in no preceding alliance except the Aristolochiales. It seems probable that the higher alliances of the Archichla- nivdeae are related in some way to the Ranalcs, whose numerous lines of development seem to have been taken up by other 216 MORPHOLOGY OF ANGIOSPERMS alliances. It follows that the subsequent alliances will touch the Ranales in various ways, the latter representing a plexus out of which various divergent lines have become distinct. This con- ception of the genetic position of Ranales among Archichlamy- deae has brought to them the attention of morphologists, and the results thus far have more than justified their investigation. XVI. Rhoedales. — This includes Papaveraceae, Cruei- ferae, Tovariaceae, CajDparidaceae, Resedaceae, and VEoringa- eeae, together containing about 2,615 species, the large families being Crueiferae with 1,800 species, Capparidaceae with 125, and Papaveraceae with 2 SO. There seems to be no question that this alliance is closely related to the Ranales. The connec- tion seems to be through the Papaveraceae, which exhibit struc- tures resembling those of Xymphaeaeeae ; while the transition from Papaveraceae to Crueiferae through the Finn arid forms is plain, and the affinity of Crueiferae and Capparidaceae is unquestioned. XVII. Sarracexiales. — This includes Sarraceniaeeae, Xepenthaceae, and Droseraceae, together containing 115 spe- cies, nearly 100 of which belong to the Droseraceae. The alliance is evidently parallel with Rhoedales, and both are cer- tainly related to the Xymphaeaceae-region of the Ranales. In fact, the Xymphaeaeeae, Papaveraceae, and Sarraceniales have many things in common in the arrangement of floral members and the spirocyclic character of the flowers. The distinctive character of Sarraceniales as compared with Rhoedales is the prevalence in the former of central placentation. XVIII. Rosales. — This includes Podostemonaceae, Hy- drostachyaceae, Crassulaceae, Cephalotaceae, Saxifragaeeae, Pittosporaceae, Brunelliaceae, Cunoniaceae, VTyrothanmaceae, Bruniaceae, Ilamamelidaceae, Platanaceae, Crossosomataceae, Rosaceae, Connaraeeae, and Leguminosae, tos-ether containing about 11.27<) species, the large families being Leguminosae with over 11,000 species, Rosaceae with 1,525^ Saxifragaeeae with 630, and Crassulaceae with 190. Since this alliance contains bv far the greatest family of Archichlamvdeae, in fact, with a single exception, the greatest family of Angiosperms, it may be regarded as the most representative and dominant alliance. The beginnings of this great alliance, with apocarpy, hypo- gyny, and indefinite repetition of certain floral members, have CLASSIFICATION OF ARCHICHLAMYDEAE 247 much in common with the Ranales, especially the line con- taining Ranunculaceae. However, it has reached a much higher development in the more frequent occurrence of syncarpy, and also of perigyny and epigyny, and especially in the remarkable development of zygomorphy among the Leguminosae. Disre- garding the smaller families, the Saxifragaceae may be regard- ed as the beginnings of the alliance, originating in the Ranales, and diverging toward Podostemonaceae in one direction and Rosaceae-Leguminosae in the other. It has long been known that there is no real distinctive character separating Saxifraga- ceae and Rosaceae ; and the transition from the latter family to the Leguminosae is easy. Rosaceae are characterized by actino- morphic flowers and several carpels; while Leguminosae have zygomorphic flowers and a single carpel ; but there are members of the two families that exactly reverse these distinctions. There seems to be a general plexus formed by the Rosa tribe of Rosa- ceae and the Mimosa tribe of Leguminosae, which is not very far removed from the Ranunculaceae among Ranales. Out of the Rosa tribe the two very distinct lines of drupe-forms and pome-forms have diverged; while the Mimosa tribe, with its actinomorphic flowers and numerous usually free stamens, leads through the Caesalpinia tribe, with its actinomorphic or zygo- morphic flowers and free stamens, to the Papilio tribe with its strongly zygomorphic flowers and coalescent stamens. The culmination of the alliance is of course the elaboration of zygomorphy, the Leguminosae dominating in this regard among Archichlamydeae, as do the Orchidaceae among Monoco- tyledons, and the Personales among Sympetalae. In the preceding related alliances, from Ranales to RosaL ■-, the cyclic character of the flower is not fully established, every line of development having spiral members. In the following alliances, however, the cyclic character is fully established. XIX. Geraxiales. — This includes Geraniaceae, Oxalida- ceae, Tropaeolaceae, Linaceae, Humiriaceae, Erythroxylaceae> Zygophyllaeeae, Cneoraceae, Rutaceae, Simarubaceae, Bursera- ceae, Meliaceae, Malpighiaceae, Trigoniaceae, Vochysiaceae, Tremandraceae, Polygalaceae, I Hchapetalaceae, Euphorbiaceae, and Callitrichaceae, together containing about 9,160 species, the large families being Euphorbiaceae with 4,140 species, Ru- taceae with 910, Meliaceae with 753, Malpighiaceae with 700, 17 24S MORPHOLOGY OF ANGIOSPERMS Polygalaceae with 607, Geraniaceae with 455, Oxalidaceae with. 330, and Burseraceae with 320. This cyclic alliance begins with those families that are iso- carpic and extends to those in which a reduction in the number of carpels is prevalent. It is chiefly distinguished from the Sapindales, with which it is parallel and very closely allied, by the orientation of the ovules, the raphe of the anatropous ovules being ventral in Geraniales and dorsal in Sapindales. Just the significance of such a character in distinguishing great genetic alliances is not clear, but its constancv is in its favor. Three lines of development are evident, the most prominent beginning with Geraniaceae, including the zygomorphic and anisocarpic Tropaeolaceae and the completely syncarpic Lina- ceae and its allies, and ending in Cneoraceae to Meliaceae with oil-cells and highly differentiated tissues. Another line is Mai- pighiaeeae to Vochysiaceae, characterized by oblique zygomor- phy ; while Polygalaceae with its strongly zygomorphic flowers, Dichapetalaceae, and Euphorbiaceae, show no surviving fea- tures in common. The affinities of these last three families are extremely doubtful, and those of Callitrichaceae are even more so. XX. Sapindales.* — This includes Buxaceae, Empetra- ceae, Coriariaceae, Limnanthaceae, Anacardiaceae, Cyrillaceae, Pentaphylaeeae, Corynocarpaceae, Aquifoliaceae, Celastraceae, Hippocrateaceae, Stackhousiaceae, Staphyleaceae, Icaeinaceae, Aceraceae, Hippocastanaceae, Sapindaceae, Sabiaceae, Meli- anthaceae, and Balsaminaceae, together comprising about 3,125 species, the large families being Sapindaceae with 1,040 species, Celastraceae with 425, Anacardiaceae with 395, Balsaminaceae with 300, and Aquifoliaceae with 285. As among Geraniales, the alliance begins with isocarpic forms and passes to those in which the number of carpels is reduced, and in the higher families zygomorphy is attained. The orientation of the ovules that separates this alliance from the Geraniales was referred to under that alliance. Engler rec- ognizes so many lines of development among Sapindales that the alliance seems to be well broken up. and the different mem- bers not clearlv related to one another. * Sometimes called Celastrales. CLASSIFICATION OF ARCH1CHLAMYDEAE 249 XXI. Rhamnales. — This includes Rhamnaceae and Vita- ceae, together containing about 955 species, almost exactly equally distributed between the two families. The alliance is clearly parallel with the preceding one, but is distinctly set apart by its tetracyclic flowers with opposite stamens. XXII. Malvales.— This includes Elaeocarpaceae, Chlae- naceae, Gonystylaceae, Tiliaceae, Malvaceae, Triplochitonaceae, Bombacaceae, Sterculiaceae, and Scytopetalaceae, together con- taining about 1,740 species, the large families being Malvaceae with about 800 species, and Sterculiaceae with 780. This alliance is very uneven in the advancement of its characters, and in certain features would seem to precede Geraniales and Sa- pindales in any sequence ; but it is so closely related to Parie- tales through Elaeocarpaceae and Chlaenaceae that it seems clear it should be placed near them. Distinct or slightly united carpels are found, as among the Geraniales and Sapindales, but complete syncarpy prevails. The inequality of advancement is shown in such families as Tiliaceae, in which there is complete syncarpy associated with indefinite stamens ; and Sterculiaceae, in which there is a com- plexity in the arrangement of stamens approaching that in Malvaceae, associated with a more or less incomplete union of carpels. XXIII. Parietales. — This includes Dilleniaceae, Eu- cryphiaceae, Ochnaceae, Caryocaraceae, Marcgraviaceae, Qui- inaceae, Theaceae, Guttiferae, Dipterocarpaceae, Elatinaceae, Frankeniaceae, Tamaricaceae, Fouquieraceae, Cistaceae, Bixa- ceae, Cochlospermaceae, Koeberliniaceae, Canellaceae, Viola- ceae, Flacourtiaceae, Staehyuraceae, Turneraceae, Malesherbia- eeae, Passifloraceae, Achariaceae, Caricaceae, Loasaceae, Datis- caceae, Begoniaceae, and Ancistrocladaceae, together compris- ing about 4,225 species, the large families being Guttiferae with 760 species, Flacourtiaceae with 525, Begoniaceae with 405, Violaceae with 400, Dipterocarpaceae with 320, and Pas- -inViraeeae with 315. The Parietales are prevailingly svncarpous, and have very evident connection with the Ranales through the Dilleniaceae, which were formerly included among the Ranales, and with the Rhoedales through the Flacourtiaceae and other families. The families from Dilleniaceae to Dipterocarpaceae, mainly 250 MORPHOLOGY OF ANGIOSPERMS tropical, are regarded as one line, characterized by an oily en- dosperm ; and among them such primitive characters as the spiral arrangement and indefinite number of floral members occur, and even apocarpy (Ochnaceae). Another line includes the Elatinaceae to the Frankeniaceae, chiefly a temperate group characterized by a starchy endosperm. The Fouqiiieraceae are regarded as independent of the last line on account of their sympetaly and oily endosperm. The Cistaceae and Bixaceae also form an independent line with starchy endosperm. The Cochlospermaceae and Ivoeberliniaceae are also regarded as independent and much resemble the Capparidaceae among the Phoedales. The families from Canellaceae to Achariaceae form another line, all characterized by oily endosperm, starting with completely cyclic flowers, and leading to such special develop- ments as a strong tubular development of the receptacle and even sympetaly (Achariaceae). Closely related to this line are the Caricaceae, with sympetalous corollas, but distinguished by their stamens and latex system. The last four families (Loasa- ceae to Ancistrocladaceae) are epigynous, but each one seems to be a peculiar and isolated type of development. This complex alliance is a good illustration of divergent lines of development within one general circle of affinity, and at the same time of a gradual increase in floral complexity. XXIV. Opuntiales. — This includes the single family Cactaceae, with about 1,000 species. This characteristic Amer- ican family presents a strange mixture of primitive and ad- vanced characters in the structure of the flower. The spiral arrangement and indefinite repetition of floral members are often as primitive as in the Xymphaeaccae, with which region of the Ranales the alliance may be connected. The tubular receptacle, however, enclosing the constantly syncarpous pistil relates the group to the Parietales. XXV. Myrtales.— This includes the Geissolomaceae, Pe- naeaeeae, Oliniaceae, Thymelaeaceae, Elaeagnaceae, Lythra- ceae, Sonneratiaceae ( Blattiaceae), Punicaceae, Lecythidacoae, Phizophoraceae, Combretaceae, Myrtaceae, Melastomataceac, Onagraceae, Hydrocaryaceae, Haloraghidaceae, and Cynomo- riaceae, together containing about 7, ISO species, the large fami- lies being Melastomataceae with 2,750 species, Myrtaceae with 2,5fi5, Onagraceae with 465, Thymelaeaceae with 395, and CLASSIFICATION OF ARCHICHLAMYDEAE 251 Lythraceae with -340. The high character of this alliance is in- dicated by the constantly perigynous and epigynous flowers, as well as by the constantly cyclic stamens, and the tendency to tetramerous flowers is strong. XXVI. Umbellales.* — This includes the Araliaceae, Umbelliferae, and Cornaceae, together containing about 2,660 species, about 2,100 of which belong to the Umbelliferae. The scries is clearlv the ranking one among the Archichlamvdeae on account of its epigyny, cyclic stamens, reduced number of car- pels, and mostly reduced sepals, the floral formula being the same as that of the highest Sympetalae. The three families constituting the alliance are very closely related, and the alliance as a whole stands so stiffly apart from other Archichlamydeae as to raise the question whether it does not really belong among the higher Sympetalae. It will be noted that in a large sense, and with the excep- tion of the last two alliances, the Archichlamydeae correspond to the Spiral series among Monocotyledons, in which the cyclic arrangement, although it frequently appears, is not fully estab- lished in every set of floral members. In the same sense, there- fore, the Myrtales, Umbellales, and Sympetalae, correspond to the Cyclic series among Monocotyledons. * Umbelliflorae of Engler. CHAPTER XII CLASSIFICATION OF SYMPETALAE The Sympetalae form a much better defined group than do tbe Arehiehlamvdeae, from which thev seem to have been de- rived. The sympetalous character is almost universal, and justifies the name of the group. To regard it as the crucial test, however, is to introduce the flavor of an artificial system. Among the Archichlamydeae sympetalous forms were noted, and certain families of the Sympetalae include polypetalous members. It would seem that such exceptions might apply to whole families, whose other characters would determine their affinities. For example, the Umbelliferae present the combina- tion of characters that belongs to the Sympetalae, excepting sympetaly ; and this exception does not seem to be a sufficient reason to exclude them from association among the epigynous anisocarpic Sympetalae, any more than the polypetaly of the Pirolaceae excludes them from the isocarpic Sympetalae. The general characters of Sympetalae are (1) a complete cyclic arrangement of the floral members, associated with defi- nite numbers; (2) a sympetalous corolla that usually has a common origin with the stamens: and (3) ovules with a single massive integument and a very small nucellus. The group con- tains fifty-one families, the number varying with different au- thors, and about 42,000 species, or approximately two-thirds of the number included in the Archichlamydeae. Eight alli- ance- have been recognized by Engler, coordinate with the ten alliances of Monocotyledons and the twenty-six alliances of Archichlamydeae, the contrast with the latter group in uniform- ity of floral structure beina' very striking. The natural sequence of tin1 alliances is much more evident than among the Archichlamydeae. The first three alliances are 252 CLASSIFICATION OF SYMPETALAE 253 pentacyelic and isocarpic, while the remaining five are tetra- cyclic and anisocarpic ; and of the anisocarpic alliances, the first three are hvpogynous and the last two epigynous. The three pentacyelic or isocarpic alliances are certainly most nearly allied to the Archichlamydeae, for among them poly- petaly still occurs, the two cycles of stamens are characteristic, and occasionally the ovule has two integuments. Thev may be e O if %J regarded as lines from the Archichlamydeae in which sympetaly has become prevalent. They are all hvpogynous and actino- morphic, and the floral formula is characteristically sepals 5, petals 5, stamens 5 + 5, carpels 5. These comparatively primi- tive Sympetalae are not numerous, containing only about 3,500 of the 42,000 species, and hence they are not the representative Sympetalae. I. Eric ales. — This includes the Clethraceae, Pirolaceae, Lennoaceae, Ericaceae, Epacridaceae, and Diapensiaceae, to- gether containing a little more than 1,700 species, by far the largest family being Ericaceae with about 1,360 species. The group is characteristically developed in high latitudes and alti- tudes, and its special features are well marked. The stamens are usually quite free from the petals, and this in connection with occasional polypetaly gives a strong resemblance to the Archichlamydeae ; while the peculiar dehiscence of the anthers and their frequent appendages are very characteristic. The stamens are bv no means constantly in two cycles, or distinct from the corolla or one another. A single cycle of functional stamens may be associated with staminodia, or onlv a sins-le cycle may appear, or the stamen cycle may have a common ori- gin with the corolla, or in some cases it may be monadelphous. In short, there are transition forms to the suppression of a cycle of stamens, and to a common origin of stamen cycle and corolla. A multilocular ovary with numerous ovules is also a feature of the alliance. The Epacridaceae, a well-developed Australian family of heath-like plants containing nearly 300 species, are quite ex- ceptional in having only one cycle of stamens and anthers with longitudinal dehiscence. These exceptions seem quite funda- mental, but they may be illustrations of the result of long and distant separation of allied families. Tn any event, a com- parative morphological study of Epacridaceae and Ericaceae 254 MORPHOLOGY OP ANGIOSPERMS is much needed ; and the whole series of Ericales deserves atten- tion on account of its possible genetic connections with sonic region of the Archichlamydeae. II. Piumulales. — This includes the Myrsinaceae, Priniula- ceae, and Plumbaginaceae, together containing about 850 spe- cies, approximately equally distributed among the three fami- lies. The families are closely associated in structure, but widely separated in geographical distribution, the Myrsinaceae being characteristically tropical trees and shrubs (chiefly Amer- ican), the Primulaceae north temperate and boreal herbs, and the Plumbaginaceae characteristically halophytic herbs and un- dershrubs of salt-beaches and steppes (chiefly Mediterranean and Caspian). That such dissociated families should have so much in common is a strong argument against the older idea that similarity of structure proves common origin. The two most characteristic features of the group are the single cycle of stamens opposite the petals, and the unilocular ovary with its " free central placenta " bearing numerous ovules. The single cycle of stamens and its opposition to the petals are explained by the frequent occurrence of rudiments representing an outer abortive cycle. The " free central placenta " of tax- onomists is of course a continuation of the floral axis to bear ovules, and is perhaps the most important morphological char- acter of the series. It is in this group, also, that there has been noted a peculiar origin of the petals, which are said to arise late from the primordia that have already developed the stamens. As compared with the Ericales, the Primulales may be re- garded as somewhat more advanced toward the higher Sympeta- lae, but polypetaly still occurs among them, and they give the impression of a somewhat divergent and specialized group. An investigation of the Myrsinaceae will doubtless result in a much clearer understanding of the relationships. III. Ebexales. — This includes the Sapotaceae, Ebenaceae, Styracaeeae, and Symploeneeao, together containing nearly 000 species, the large families being Sapotaceae with about 380 spe- cies, and Ebenaceae with 275. The group is chiefly developed in the tropics and the species are all shrubs or trees. The alliance is particularly puzzling in its affinities, since there is a combination of primitive and advanced characters. CLASSIFICATION OF SYMPETALAE 255 The primitive characters are the indefiniteness in the number of sepals and petals, ranging from 4 to 8, occasional polypetaly, and the often numerous stamens and carpels. Consistency would seem to demand that the Ebenales he regarded as the most primitive of the Sympetalae, even the definite cyclic num- bers not being established. At the same time, there is adherence of a single stamen cycle to a sympetalous corolla, and distinct epigyny. The stamen cycles are peculiarly fluctuating, ranging from three or four cycles, through all stages of suppression of the outer cycles, to a single opposed cycle. This latter feature is suggestive of the Primulales, but the multilocular ovary with usually large solitary ovules is suggestive neither of Primulales nor Ericales. The tropical forms certainly deserve careful mor- phological investigation, and are doubtless related to the Myr- sinaceae, and in our judgment are to be included in any discus- sion of the most primitive Sympetalae. In the five following alliances the tetracyclic character seems to be well established, and the prevailing formula is sepals 5, petals 5, stamens 5, carpels 2. In the three previous isocarpic alliances there is every transition from the pentacyclic to the tetracyclic condition, and among the more primitive anisocarpic families the carpels are often three before two becomes the established number. Of the remaining iilliances the first three are hypogynous. IV. Ge^tianales.* — This includes the Oleaceae, Salvado- raceae, Loganiaceae, Gentianaceae, Apocynaceae, and Asclepia- daceae, together containing about 4,200 species, the large fami- lies being Asclepiadaceae with about 1,720 species, Apocyna- ceae with 075, and Gentianaceae with 725. With this alliance the grouping into developmental lines becomes indefinite and perplexing, for the numerous families intergrade in every direction. There is no distinctive character that separates this alliance from the great alliance Tubiflorales. The fact that the corolla is generally twisted in aestivation seems to be the most useful character, and has suggested a name for the series, and the constantly opposite leaves is a supple- mentary character. The lower members of the alliance are the Oleaceae and * Contortae of En^ler. 250 MORPHOLOGY OF ANGIOSPERMS Salvadoraceae, in which there is sometimes distinct polypetaly, but the reduction of the stamens to two in the former family is hardly to be regarded as a primitive character. The Logania- ceae are general in their resemblances, having features in com- mon with the remaining families, and others suggestive of Tu- biflorales and Rubiales. In fact, Engler suggests that the Loganiaceae may be an older type than any of the others, and may have given rise to the Gentianales and Rubiales, in which he might have included the Tubinorales. If this family may hold any such position in reference to these great alliances it certainly deserves careful investigation. The alliance ends with the Apocynaceae and Asclepiadaceae, in which a latex-system is developed, and oilier evidences of high specialization occur: but they are also characterized by distinct carpels, a feature re- garded as primitive. The Asclepiadiaceae form a very peculiar and highly specialized offshoot, the elaboration of floral struc- tures for entomophily reaching a degree of complexity only to be compared with that of the Orchidaceae. V. Tubiflokales.'" — This includes Couvolvulaceae, Pole- moniaceae, Hydrophyllaceae, Borraginaceac, Yerbenaceae, La- biatae, Xolanaceae, Solanaceae, Scrophulariaceae, Bignoniaceae, Pedaliaceae, Martyniaceae, Orobanchaceae, Gesneraceae, Colu- melliaceae, Lentibulariaceae, Globulariaceae, Acanthaceae, Myoporaceae, and Phrymaceae, together containing over 14,000 species, the large families being Labiatae with nearly 3,000 species, Scrophulariaceae with 2,400, Acanthaceae with nearly 2,0(>0, Solanaceae with about 1.700, and Borraginaceac with about 1,550. This enormous assemblage of forms has been ordinarily con- sidered as representing at least two alliances, the Polemoniales or Tubiflorae including the first four families of the list above, and the Personales or Labiatiflorae including the remaining families. The tendencies of development are so numerous and interwoven that they are difficult to separate, but rather than merge two such alliance- together it might have been better to have broken up the Personales into five or six alliances, espe- cially if the Plantaginaceae are to be set off as a coordinate alliance Plantaginales. To distinguish them definitely would * Tubiflorae of Engler. CLASSIFICATION OF SYMPETALAE 257 probably be impossible, but an alliance at best expresses only a general evolutionary tendency more or less completely worked out. Taking the alliance as a whole, it represents the culmination of hypogynous Sympetalae, and this culmination is shown not only in the conspicuous corolla but in highly developed zygo- morphism. In fact, the Personales, with the Labiatae and Scrophulariaceae as centers of aggregation, represent the great zygomorphic group of the Sympetalae, as Leguminosae do among the Archichlamydeae, and Orchidaceae among the Mono- cotvledons. First in the alliance are the Convolvulaeeae and Polemonia- ceae on account of their actinomorphic flowers and several- ovuled carpels, in these and other features being, together with the Gentianales, the least modified of the tetracyclic families. From Gentianales they are easilv distinguished bv their lack of twisted aestivation and by their usually alternate leaves, and also by their undoubted relation to the other families of Tu- biflorales. A second natural alliance is that formed by the Hydrophyl- laceae and Borraginaceae, which leads from the preceding alli- ance through Hydrophyllaceae, with a generally unlobed ovary, to the Borraginaceae with a much modified ovary. In the latter family the two carpels are divided by a false partition, each loculus contains a single ovule, and the ovary becomes so deeply lobed as to resemble a group of four nutlets. Further modi- fications of this peculiar fruit, familiar to taxonomists, make it the most specialized and diversified structure of this large family. A third natural alliance is that formed by the Verbenaceae and Labiatae, with about 3, TOO species. It is joined to the Convolvulaeeae by the orientation of the ovule, and has fol- lowed a developmental path parallel with that of the preceding alliance in the evolution of the carpel structures. The lobing of the ovary into four nutlet-like bodies in the Labiatae, how- ever, is not accompanied by such detailed specialization as in the Borraginaceae: but the whole line is dominated bv the strong development of zygomorphy, reaching its culmination in certain groups of the Labiatae. A fourth natural alliance, the greatest of all, includes the 258 MORPHOLOGY OF ANGIOSPERMS eleven families from Xolanaceae to Globulariaceae, grouping about the Solanaceae and Scrophulariaceae. This series con- nects with the ('(hi ml vulus forms through the Xolanaceae, but does not develop its carpel-structures as do the Borrage and Labiate lines, retaining capsules with numerous ovules, but there is a strong development of zygomorphy. To summarize at this point, the primitive stock of the series seems to be the Convolvulaceae-Polemoniaceae alliance, from which three distinct lines of development have diverged : the Hydrophyllaceae-Borraginaceae line, with its modified carpel- structures ; the Verbenaeeae-Labiatae line, with its modified carpel-structures and zygomorphy; and the Xolanaceae-Globu- lariaceae line, with its zygomorphy. It should be noted in pass- ing that the zygomorphy is associated with a strong tendency to reduce the number of stamens. The three remaining families are so peculiar in certain fea- tures that Engler regards them as representing separate lines of development, although the Acanthaceae are not easily separated from certain families of the last alliance. The Myoporaceae seem to be a reduced type with no clear affinities ; and the Phrymaceae, with their achenes and orthotropous ovules, have no evident connections in this alliance, in which their strong zygomorphy has retained them. It would be our judgment, therefore, to break up this great alliance of Tubiflorales into at least four, which might be called the Polemoniales (Convolvulaceae and Polemoniaceae), Bor- raginales (Ilydrophyllaceae and Borraginaceae), Labiatales (Verbenaceae and Labiatae), and Personales (Xolanaceae, Solanaceae, Scrophulariaceae, Bignoniaeeae, Acanthaceae, Pe- daliaceae, Martyniaceae, Orobanchaceae, Gesneriaceae, Colu- melliaceae, Lentibulariaceae, and Globular iaceae), the Myo- poraceae and Phrymaceae being left undetermined or regarded as reduction forms of Personales. VL Plaxtaoixat.es. — This includes the single family Plantaginaceae with about 200 species. This family, with its peculiar habit, 4-merous flowers, membranous corolla, and char- acteristic fruit, is certainly entitled to special consideration. If such a series as Tubiflorales be maintained, however, there is no good reason why Plantaginaceae should not form one of the seven or eight sections of it. If, on the other hand, the CLASSIFICATION OF SYMPETALAE 259 series be broken up as suggested above, Plantaginales should certainly be coordinate with Polemoniales, Borraginales, Labi- a tales, and Personales. The two remaining alliances are epigynous and naturally form the culmination of the Sympetalae. In both alliances there is actinomorphy and numerous ovules, but in both there is more or less development of zygomorphy ; a tendency to reduction in numbers of members, especially of the ovules ; and a tend- encv to reduce the flowers in size and to mass them, leading to a modification of floral structures and a differentiation of the functions of individual flowers. VII. Rubiales. — This includes the Rubiaceae, Caprifolia- ceae, Adoxaceae, Valerianaceae, and Dipsaceae, together con- taining nearly 4,800 species, the large family being Rubiaceae with nearly 4,100 species. The possible relationship of this alliance to the Gentianales, especially the Loganiaceae, has been mentioned, from which it seems to be an epigynous offshoot. At the same time, rela- tions to the epigynous Umbellales among the Archichlamydeae are no less evident. It may possibly be found, as intimated in the last chapter, that the Umbellales should be associated with the Rubiales as two parallel alliances of epigynous Sympetalae. Through the Caprifoliaceae the Valerianaceae and Dipsaceae are closely connected with the alliance; while the position of the Adoxaceae is altogether uncertain. The distinguishing char- acter to separate Rubiales from the next alliance is not always clear, but in general the connivent and often united anthers of the Campanales are not present in the Rubiales ; but this char- acter is fortified by distinct developmental tendencies. VIII. Campanales. — This includes the Cucurbitaceae, Campanulaceae, Goodeniaceae, Candolleaceae, Calyceraceae, and Compositae, together containing more than 14,500 species, fully 12,500 of which are Compositae, the Campanulaceae con- taining nearly 1,100. ( Ymnivent and often united anthers, and sometimes mona- delphous stamens, prevail in the series. The peculiar tropical Cucurbitaceae occupy a special place in the alliance, and can not be related clearly to the others; while the Campanulaceae seem to represent a remnant of the ancient stock of the alliance, from which the other families have arisen. 260 MORPHOLOGY OF AXGIOSPERMS The alliance culminates in the Compositae, the greatest of all angiospermous families, not only in rank, but also in the num- ber of species, although not much exceeding the Leguminosae in this latter regard. There seems to be no question that the Compositae represent the highest expression of the various de- velopmental lines we have been tracing through the Angio- sperms. This is shown not merely in their combination of sympetaly, epigyny, and seed-like fruit, but also by such special structures as the pappus and the syngenesious anthers, by the complex organization of the head, the prevalence of diclinism, the dimorphism of corollas, etc. CHAPTER XIII GEOGRAPHIC DISTRIBUTION OF ANGIOSPERMS So vast a subject can be presented only in very brief outline in a single chapter. In a certain sense it is not pertinent to a discussion of the special morphology of a group, but the stu- dent of special morphology is aided by certain general consid- erations connected with geographic distribution, especially in any discussion of phylogeny. The distribution of a group con- taining nearly 125,000 species includes a vast mass of details, and only certain salient features can be selected for presenta- tion. Even when these are selected, the numerous exceptions to anv general statement must be disregarded. It must be un- derstood, therefore, that in the following account the statements are very general in their nature, expressing average conditions of distribution, under all of which exceptions may be cited. At the same time, it is the general tendency in the distribution of any large group that is of interest to the morphologist rather than the details of distribution of species and genera. The subject of geographic distribution presents two aspects for consideration. One involves the determination of life-zones over the surface of the earth, which is a consideration of dis- tribution from the standpoint of physiography. The other aspect disregards the life-zones, and considers distribution from the standpoint of plant-groups. What a given plant-group has been able to do in the occupation of the earth's surface is of more morphological interest than the physiographic features of the problem, and hence the following presentation will take the latter standpoint. Including only the existing vegetation gives a very inade- quate conception of the relation of any group to the earth's surface. The present distribution of a group is only the last 261 262 MORPHOLOGY OF ANGIOSPERMS stage in a long history of distribution, and a knowledge of this history is an essential factor in any explanation of the present distribution. Unfortunately, very little of this history is avail- able, and this presentation must content itself with indicating the present relation of groups to the earth's surface, without any attempt at explanation. This is particularly unfortunate, since a lack of historical evidence may vitiate many conclu- sions. If this lack of historical testimony be added to the lack of any adequate record of the geographic distribution of existing species, it becomes evident that the generalizations pro- posed must be of the most tentative character. With this ex- planation the following statements may be given their proper weight. MOXOCOTYLEDOXS It is possible to present the distribution of the ten alliances of Engler in the order of their supposed relationship, a method that may be of service in the subsequent consideration of the ancient history and phylogeny of the group. One genetic group is supposed to include the three following alliances. Paxdaxales. — The Pandanaceae (screw-pines), apparently the most primitive of Monocotyledons, belong to the general region of the Indian Ocean. Associated with them in relation- ship are the Typhaceae, found in aquatic conditions throughout the world, but most abundant in the tropics : and the Spargania- ceae, restricted to the temperate and boreal regions of the northern hemisphere and also of the Australasian region, and not represented in the tropics. The series as a whole shows wide adaptations to temperature, but not to soil conditions, with the primitive forms massed in the oriental tropic-. Pai.males. — The Palmaeeae are about equally divided be- tween the oriental and occidental tropics, with no temperate outliers, but not a species or a genus is common to the two hemispheres. The geographical association of the palms and screw-pines in the orient is in favor of their supposed relation- ship, but the palms of the Occident need explanation, especially since Phyteleplms, regarded as a genus intermediate between Pandanaceae and Palmaeeae, is an American genus. The pres- ent distribution of palms is an excellent illustration of the de- velopment of continental diversities, which in this case has GEOGRAPHIC DISTRIBUTION OF ANGIOSPERMS 2G3 resulted not only in distinct genera, but almost every tribe is either oriental or occidental. Furthermore, the much larger number of monotypic genera in the orient must be associated with its larger and more broken tropical area. Syxaxthales. — The Cyclanthaceae are as restricted to the American tropics as the Pandanaceae are to the oriental tropics. If this general " palm " type, comprising these three alli- ances, was once connected in the two hemispheres by a northern distribution, the palms alone found both hemispheres congenial in the tropics, while the Pandanaceae disappeared from the western and the Cyclanthaceae from the eastern hemisphere. Helobiales. — This primitive series is very widely dis- tributed and contains relatively few species, probably on account of its aquatic character. Three of its families (Pota- mogetonaceae, Xaiadaceae, and Hydrocharitaceae) have a world-wide distribution. The remaining five families are some- what restricted as follows : Aponogetonaceae in the Indian Ocean region, Triuridaceae in the tropics of both hemispheres, Butomaceae extending from the tropics into temperate regions, while Juncaginaceae and Alismaceae are mostly outside of the tropics in the northern and southern hemispheres. Akales. — The possible relationship of this group to the pre- ceding one has been mentioned. The aquatic Lemnaceae are universally distributed, but 92 per cent of the Araceae are within the tropics, being massed chiefly in South America, India, and the East Indies. This family, as the palms, affords a good illustration of the development of continental diversi- ties. In this case, however, the diversity has not reached so extreme a stage as in the palms, in which even the tribes of the orient and Occident are for the most part distinct. Among Aroids the tribes of the two hemispheres are by no means dis- tinct, at least two tropical genera (Cyrtosperma and Ilomalo- mena) have species in both hemispheres, and the monotypic Pistia is found in every tropical region. The species are more numerous in the American tropics, but the number of genera is nearly twice as great in the oriental tropics. The Aroids differ further from the palms in having at least six genera characteristic members of north temperate vegetation, and these for the most part are common to both hemispheres. 18 2G4 MORPHOLOGY OF ANGIOSPERMS Glumales. — The -world-wide distribution of this great alliance, from tropical to boreal conditions, has resulted in no continental tribes, comparatively few continental genera, and very numerous cosmopolitan species. So far as geographic dis- tribution is concerned, it may well represent the primitive stock from which the following alliances have branched. Farixales. — This alliance is made up of a remarkable group of isolated families, apparently being poorly adapted for cos- mopolitan distribution. Only three of the eleven families have a more extensive distribution than a hemisphere, Eriocaula- ceae, the most cosmopolitan family, being massed in the tropics, Commelinaceae occurring everywhere except in boreal condi- tions, and Pontederiaceae being represented in all warmer re- gions. Four families (Flagellariaceae, Restionaceae, Centro- lepidaceae, and Eapateaceae) belong to the southern hemi- sphere, three ( Mayacaceae, Xyridaceae, and Bromeliaceae) are restricted to the western hemisphere, and Philydraceae are Australian. Lieiai/es. — This series, in contrast to the Farinales, is made up of characteristically cosmopolitan families. Liliaceae and Iridaceae are literally cosmopolitan, Amaryllidaeeae and Tac- caceae are massed in all tropical regions, Juncaceae are best de- veloped in the cool temperates of the northern and southern hemispheres, Haemodoraceae are represented in tropical Amer- ica and Australia, Stemonaeeae are scattered in patches in Australia, Asia, and Xorth America, and Dioscoreaceae are mainly tropical. Only Velloziaceae are restricted to a single hemisphere, and the restriction is remarkable, since all of the 70 species are credited only to Brazil. Scitamixales. — The four families of this series are all tropical, two of them ( Musaceae and Zingiberaceae) being re- stricted to the oriental tropics, and two (Cannaceae and Maran- taceae) to the occidental. Oechidales. — The massing of orchids in the tropics of both hemispheres is well known, but they are by no means restricted to tropical conditions. As a rule, the numerous tropical genera are not only restricted to hemispheres, but are often very local ; while the temperate genera are represented in both hemi- spheres; and the most northern genera even contain cosmopoli- tan species. GEOGRAPHIC DISTRIBUTION OF ANGIOSPERMS 265 Upon examining such data, certain generalizations in refer- ence to the distribution of Monocotyledons become apparent. These will doubtless be modified by a fuller knowledge of the distribution of families, but they will serve to illustrate certain facts : 1. Four great terrestrial families (Gramineae, Cyperaceae, Liliaeeae, and Iridaceae) of Monocotyledons are world-wide in their distribution. This means that they have been able to Income adapted to every condition of soil and climate possible to high-grade vegetation. 2. The Monocotyledons include a remarkable number of purely hydrophytic families which also have a world-wide dis- tribution so far as fresh and brackish waters are concerned. The families are Typhaceae, Potamogetonaceae, Xaiadaceae, Ilvdrocharitaceae, Lemnaceae, and Pontederiaceae, four of Them belonging to the Helobiales. In spite of this wide dis- tribution, these families contain less than 200 species. When this fact is taken in connection with the 10,000 species belong- ing to the four cosmopolitan terrestrial families mentioned above, it becomes evident that the very diverse conditions of the land surface are far more favorable, to the production of species than the comparatively uniform aquatic conditions. 3. There is a decided massing of monocotyledonous families in the tropics. This is so marked as to suggest that Monocotyle- dons as a whole are essentially tropical. 4. As a corollarv to the last statement, the entire absence of boreal forms, excepting the few belonging to the families of universal distribution, is noteworthy. 5. The poor representation of Monocotyledons in the southern hemisphere, exclusive of the world-wide families, is remarkable. Especially is this true of Australia, a region prolific in endemic forms among Gymnosperms and Dicotyle- dons. 0. Very few families are characteristic of temperate re- gions, and these ( Sparganiaceae, Juncaginaceae, Alismaceae, and Juneaceae) are represented in both the northern and southern hemispheres, and none of them are of the higher petaloideous type. 7. The tropical representation of Monocotyledons is ap- proximately equal in the two hemispheres, not merely in num- 206 MORPHOLOGY OF ANGIOSPERMS ber of species but also of families. The tropical families repre- sented in both hemispheres are Butomaeeae, Triuridaceae, Palmaceae, Araceae, Eriocaulaceae, Commelinaceae, Amaryl- lidaceae, Taccaceae, Dioscoreaceae, Burmanniaceae, and Orchi- daceae. Those peculiar to the oriental tropics are Pandanaceae, Aponogetonaeeae, Musaceae, and Zingiberaceae. Those peculiar to the occidental tropics are Cyclanthaceae, Mayacaceae, Xyri- daceae, Bromeliaceae, Haemodoraceae, Yelloziaceae, Canna- ceae, and Marantaceae. 8. The great preponderance of epiphytic forms in the American tropics is probably associated with the culmination of the rainy forest. The two great epiphytic families are Bromeliaceae and Orchidaceae, the former being restricted to the occidental tropics, and the latter much more abundant there than in the oriental tropics. 9. The peculiar distribution of the three genera of Stemona- ceae is noteworthy and suggestive. Stemona. with four or five species, ranges from the Himalayas to southern Australia. Croomia has one of its species (C. pauciflora) in Florida, Geor- gia, and Japan; while the other (('. japonica) is restricted to Japan. The monotypic Stichneuron is restricted to the East Indies. The occurrence of a single species of this oriental family in Georgia and Florida, and that species native also to Japan, is difficult to explain. ARCIIICHLAMYDEAE It is impossible to consider the geographic distribution of the Archichlamydeae in such detail as that of the ALonocotyle- dons. The series are so numerous and indefinite that a presen- tation of their separate distribution would be confusing and not very significant. An examination of available but very insufficient data has resulted in the following extremely general statements : 1. Xo family has developed a world-wide distribution as have several families of the Monocotyledons and Sympetalae. It must be understood that this fact is related to the great diversities in the group, that have resulted in the recognition of numerous families. The family differences recognized by taxonomists are perhaps not to be pressed too far in any com- parison of the geographic distribution of the three great Angio- GEOGRAPHIC DISTRIBUTION OF ANGIOSPERMS 267 sperm groups. If they are of equal value, the Archichlamydeae respond more readily to geographic conditions than do the other groups. We suspect, however, that they are of very unequal value, and that the kind of response shown by the Archichlamy- deae to changed conditions happens to concern the structures used for determining families more than in the other groups. 2. Among the Archichlamydeae no distinctly boreal family has been developed, as among the Sympetalae. 3. The great tropical family is the Leguminosae, by far the largest Angiosperm family excepting the Compositae. If the Mimosa forms are to be regarded as the primitive ones, it is interesting to note that they are massed in tropical Africa and Australia, and that it is the highly specialized Papilip forms that have chiefly occupied the temperate regions. ■i. Certain great families are characteristic of the north temperate regions, usually being comparatively insignificant in the tropics. These are the Polygonaceae, Caryophyllaceae, Kanunculaceae, Cruciferae, Saxifragaceae, Rosaceae, Onagra- ceae, and Umbelliferae. 5. As among the Monocotyledons, aquatic forms are com- mon and cosmopolitan, but this habit does not characterize whole families so frequently as in the former group. The fact that the aquatic habit is found chiefly among the Monocotyle- dons and Archichlamydeae must be associated with the fact that in these groups the most primitive Angiosperms occur. The cosmopolitan character of such forms may be illustrated by the Ceratophyllaceae, which with only three species extends from the arctic to the antarctic regions, occurring even in Aus- tralia and the Fiji Islands. 6. There is a distinct pairing of continents especially in tropical display, as was noted among the Monocotyledons, in this case America usually being one member of the pair and Asia or Africa the other. In this pairing, what may be called the Pacific-distribution, involving Asia, the East Indies, or Australia on the one hand, and the Americas on the other, is particularly prominent. For example, the Amarantaceae are massed in South America and the East Indies, the Lardiza- balaceae in South America and southeastern Asia, the Calycan- thaceae in North America and Japan, the Lauraeeae in Amer- ica and Asia, the Malvales chiefly in America and Asia, the 268 MORPHOLOGY OP ANGIOSPERMS Myrtaceae in South America and Australia, etc. This pairing is still more evident if closelv related families are included, as the Sarraceniaceae in Xorth America and the Xepenthaceae in tropical eastern Asia and the East Indies. The pairing of Australia and Africa is less notable, as the Mimosa tribe, massed in tropical Australia and Africa, and the Thvmelaeaceae, chiefly • occurring in temperate Australia and the Cape region. The pairing of America and Africa, or the Atlantic-distribution, is quite rare. 7. The predominance of the American tropics in the devel- opment of Archichlamydeae is marked, as might be inferred from the last paragraph, almost all of the tropical groups being represented there, and two great families (Cactaceae and Melas- tomaceae) being almost exclusively American. 8. As might be expected, there is a much greater display of Archichlamydeae in the north temperate regions than in the south. Two large families, however, are characteristic of the south temperate regions — namely, the Proteaceae, chiefly Aus- tralian, some South African, and a few South American ; and the Thvmelaeaceae, characteristic of Australia and the Cape region. 9. It is of interest to note that the dominant tree-groups, so characteristic of Archichlamydeae, are of different alliances in the different regions. For example, in north temperate re- gions the Juglandales, Fagales, etc., dominate: in the tropics the Lauraceae are the characteristic tree-forms; while in south temperate regions the Proteaceae are the prominent archi- chlamydeous forest trees. 10. There is a notable diffusion of typos into all regions, so that very few families are restricted in their representation, although most of them have a fairly definite region of massing. Characteristic tropical families have representatives in the tem- perate regions, and families chiefly developed in the temperate regions have tropical representatives. SYMPETALAE The alliances of Sympetalae are comparatively so few and well defined that they may be considered separatelv. Ericalks. — This alliance is peculiar in containing distinct- ly temperate and boreal forms. It includes an arctic family GEOGRAPHIC DISTRIBUTION OP ANGIOSPERMS 269 (Diapensiaceae), an Australian family (Epacridaceae), and a great massing of heath-forms in the Cape region. Pkiml'lales. — The three families are very distinct in their geographic distribution, Myrsinaceae being tropical, especially American, Primulaceae north temperate and boreal, and Plumbaginaceae characteristically oriental in the halophytic conditions of the Mediterranean and Caspian regions. Ebexales. — The alliance is almost exclusively tropical, and in both hemispheres. Gextlaxales. — The alliance as a whole is more largelv massed in the tropics through the tropical display of its largest families, Apocynaceae and Asclepiadaceae. It contains also a great liana group (Loganiaceae) characteristic of South Amer- ica and Asia, and there is a pairing of Africa and Asia by the Salvadoraceae. The Gentianaceae have almost a world- wide distribution, but are notable in their numerous alpine species. Tubifeorales. — This great series is in the main broken up into fairly well-restricted areas, and the chief features of their distribution may be stated as follows: The Labiatae are world-wide in their distribution, being notably massed in the Mediterranean region. The Borr'aginaeeae and Scrophularia- ceae are the great north temperate families. The Solanaceae are everywhere in the tropics, extending into temperate regions especially in America. The Convolvulaeeae, Polemoniaceae. and Hydrophyllaceae are characteristically American, the first being chiefly tropical, and the other two characteristic of west- ern North America. The Gesneraceae belong to all regions of the southern hemisphere ; while the Yerbenaceae, Xolana- ceae, and Acanthaceae are notably in tropical South America. There are also two Mediterranean families, the Orobanchaceae and Globulariaeeae. The pairing of South America and Asia is shown in the display of Verbenaccac and Acanthaceae; and of tropical Asia and Africa in the display of Pedaliaceae. Plaxtagixales. — The genus Plantago is cosmopolitan. Eubiaees. — The Tiubiaceae are prominently tropical Amer- ican ; the Caprifoliaceae and Yalorianaeeae are north temper- ate throughout both hemispheres; while the Dipsaceae seem to be confined to the temperate regions of the eastern hemi- sphere. 270 MORPHOLOGY OP ANGIOSPERMS Campanales. — The Cucurbitaceae are tropical ; the Cam- panulaceae belong to the north and south temperate regions, with the lobelias as tropical representatives ; the Goodeniaceae and Candolleaceae are Australian ; the Calyceraceae are mainly tropical American ; and the Compositae are world-wide in their distribution. The main conclusions to be derived from the above facts are as follows : 1. The Sympetalae as a whole are better defined geograph- ically than the Archichlamydeae. This probably follows from the fact that they are better defined structurally. 2. There is a much more even distribution between the tropics and temperates than among the Monocotyledons and Archichlamydeae. Of course the tropical display is the larger, but it is hardly more than might be regarded as the normal ratio of increase in passing from the temperates to the tropics. 3. The Sympetalae as a whole, the youngest of the Angio- sperm groups, seem to have become prominently adapted to the relatively unoccupied temperate and boreal conditions, and to have made in them their most characteristic display. From this general point of view, the Monocotyledons and Archi- chlamydeae are characteristically tropical, and the Sympetalae as characteristically temperate. 4. There is a remarkable paucity of aquatic forms as com- pared with Monocotyledons and Archichlamydeae. This is probably associated with at least two facts — namely, the lack of primitive angiospermous types among the Sympetalae, and the previous occupation of the water conditions by the older Monocotyledons and Archichlamydeae. 5. The Sympetalae show no such notable continental pair- ing as is characteristic of the Archichlamydeae. It would seem that this may be related to the temperate and boreal develop- ment of the group, which would retain continental connections much longer than would be possible for a group of more tropical tendencies. 6. The dominance of America in the tropical display of Sympetalae is almost as notable as among the Archichlamydeae. The excessive rainfall is doubtless one factor in the explana- tion, but whether it is the chief one is uncertain. GEOGRAPHIC DISTRIBUTION OF ANGIOSPERMS 271 7. The sympetalous families of world-wide distribution are the Compositae, Labiatae, and Plantaginaceae. 8. The great north temperate families are the Borragina- ceae and Serophulariaceae. 9. The characteristic boreal group is the Ericales, a group that finds no parallel among the Monocotyledons and Archi- chlamydeae. CHAPTER XIV FOSSIL ANGIOSPERMS The importance of a knowledge of the ancient history of Angiosperms can not be overestimated. The morphological conclusions as to phylogeny that can be confirmed by historical evidence rest: upon the securest available foundation. Unfor- tunately, the paleobotanical record of Angiosperms is very frag- mentary and poorly understood. The published accounts are dominated mainly by stratigraphy rather than by plant-groups, and the named material is often so uncertain as to its affinities that the morphologist is extremely perplexed in drawing any conclusions. Even when all data are rejected excepting those that rest upon reasonably secure botanical evidence, any con- clusions must be extremely tentative, not only because much of the evidence is negative, but also because much of the re- jected material undoubtedly contains valuable testimony. In spite of this uncertainty, it may be useful to put together such testimony as we possess. Even this may modify some concep- tions as to phylogeny. MONOCOTYLEDONS When the parallel venation of leaves was taken to be a dis- tinctive character of the Monocotyledons their presence in the Carboniferous was claimed. But since it has become known that such leaves are equally characteristic of the great Paleozoic group ( 'ordaites, as well as of other Gymnosperms, and of cer- tain heterosporous Pteridophytes, this claim rests upon no sub- stantial basis. So far as we have been able tofexamine the testimony, it must be said that the existence of Paleozoic Mono- cotyledons has not been proved. There is no historical evidence that the Monocotyledons have ever been a dominant race, as the Gymnosperms have 272 FOSSIL ANGrOSPERMS 273 been, and as the Dicotyledons now are, although they do not seem to be so abundant now as they were during the Tertiary. When they do appear in undoubted forms, they are almost com- pletely differentiated and widely distributed. Their ancestral forms are obscured in the maze of unintelligible forms that pre- cede them. The only suggestion of paleobotany as to the origin of the Monocotyledons is that they are certainly a younger type than the Gymnosperms. Rejecting the claim for Carboniferous Monocotyledons, we encounter one for their existence during the Jurassic. This rests upon the occurrence of certain forms of grass-like habit, which suggest Monocotyledons, but such evidence can not be accepted as conclusive. There is certainly no clear proof of the existence of Monocotyledons in any strata earlier than the Cre- taceous.* The probability of Monocotyledons during the Jurassic rests not upon positive discovery, but upon the fact that during the Cretaceous they were abundant everywhere, and give evidence of their long presence. The earliest history of the group, therefore, is an absolute blank, and we are introduced to it in an advanced stage of development. The record can be considered under three general catego- ries— namely, (1) those families represented during the Cre- taceous, (2) those whose earliest representatives are in the Ter- tiary, and (3) those only known since the Tertiary. It must be observed that the second and third categories are based upon negative evidence — that is, representatives of these families have not been found as yet at any earlier period. It must also be remembered that many plants have a habitat and structure unfavorable to their preservation as fossils, so that failure to discover them in the geological series is no positive evidence that they did not exist. With the uncertainties understood it may be safe to present such evidence as we have. Cretaceous Families. — There seems to be sure evidence of the existence of five families during the Cretaceous, and a possibility of the occurrence of a sixth. The Pandanaceae were present and were widely distrib- * See Seward, A. C. : Notes on the Geological History of Monocotyledons. Annals of Botany 10: 205-220. pi. 14. 1896. 274 MORPHOLOGY OF ANGIOSPERMS uted. This fact seems to substantiate the claim as to the primi- tive character of this family, and to discount the theory of its origin as a reduction type. Not only did the screw-pine exist, but the family was richer in forms than at present, all the living genera containing more numerous species than now, and at least one extinct genus having been recognized. A little later in the Cretaceous the Palmaceae occurred abundantly, but in genera that are now for the most part extinct. Their distribution was very wide-spread, remains hav- ing been found in deposits from Greenland to Egypt. This early association of Pandanaceae and Palmaceae is corrobora- tive of the idea of their genetic relationship, and the later ap- pearance of the Palmaceae further confirms the morphological evidence that they may have been derived from the Panda- naceae. The Potamogetonaceae were abundant, a fact that coincides well with their morphological position as the most primitive of the Helobiales, and controverts the idea that they are a reduced type. That they were more abundantly displayed dur- ing the Cretaceous than now is evidenced by the fact that the majority of our present genera were represented, and at least three extinct genera have been detected. The above families would be expected by a morphologist to occur among the earliest Monocotyledons, but the Cretaceous record also discloses the presence of the Liliaceae. However, they are comparatively few in number, occur in the upper mem- bers of the Cretaceous series, and do not fairly display them- selves until the Tertiary, when numerous and now extinct gen- era appeared. These earlier liliaceous forms are of the Smilax type, but this negative evidence is very uncertain, as this type is peculiarly favorable for preservation. The Dioscoreaceae also appeared along with the Liliaceae, and are so confused with the Smilax forms as to be difficult to disentangle. The sixth family, whose existence during the Cretaceous is possible but far from certain, is the Araeeae, to which certain doubtful forms have been referred. Tt may have been scantily represented, and its association with the Potamogetonaceae would be confirmatory of Engler's suggestion as to their genetic connection. FOSSIL ANGIOSPERMS 275 Tertiary Families. — To the five monocotvledonous fami- lies represented during the Cretaceous the Tertiary adds at least fourteen, the older families also showing a largely increased development. It will be interesting to note how these addi- tional families fill out the ten great series of Monocotyledons. In each case the Cretaceous representative is put in paren- thesis. 1. Pandanales. — (Pandanaceae), Typhaceae, Spargania- ceae. This primitive series is thus completed as at present recognized. 2. Helobiales. — (Potamogetonaceae), Juncaginaceae, Buto- maceae, Hydrocharitaceae. This series is completed by the appearance of its highest member, and the Butomaceae are fairly representative of the xAlismaceae. 3. Glumales. — Gramineae, Cyperaceae. The occurrence of grass-like forms during the Jurassic has been referred to, but the absence of grasses from the Cretaceous record seriously militates against the claim that these Jurassic forms were grasses. It is since the Tertiary that the Gramineae have be- come most richly developed and widely spread, numerous ex- tinct genera having been described. Although it would seem impossible to determine the relationships of grasses from frag- mentary material, and doubt must be expressed as to the rela- tionships implied in such names as Poacites, Arundinites, etc., there is good evidence for the statement that the earliest grass types were related to such tropical forms as Arundo, Phrag- mites, Bambusa, etc. 4. Palmales. — (Palmaceae). The only family of the series became much more largely developed and wide-spread during the Tertiary. 5. Synanthales. — Cyclanthaceae. This family, the only member of the series, appeared during the Eocene Tertiary, and its early association with the screw-pines and palms con- firms its supposed relationship to them. 6. Amies. — (Araeeae?). The doubtful appearance of this family during the Cretaceous has boon mentioned, and this claim is not helped by the fact that they are no better known during the Tertiary. Such forms as do occur resemble Acorns and Pisfia. The so-called " Protolemnas " seem too doubtful to be included. 2.0 MORPHOLOGY OP ANGIOSPERMS 7. Farinales. — Restionaceae, Centrolepidaceae, Eriocaula- eeae. Three of the eleven families of the series are thus intro- duced, the first two now being restricted to the southern hemi- sphere, but during the Tertiary ranging through Europe. 8. Liliales. — (Liliaeeae, Dioseoreaeeae), Juneaceae, Irida- eeae. The last family is the highest member of the series, and its appearance before certain of the lower families is altogether doubtful. 9. Scitaminales. — Musaceae. The series consists of four families, and this one, now confined to the oriental tropics, is recognized as the most primitive. 10. Orchidales. — Xot represented. At the end of the Tertiary, therefore, there is reasonable evidence as to the existence of all the great series of Monocotv- ledons excepting the highest, and of nearly one-half the fam- ilies. DICOTYLEDOXS Any evidence as to the comparative antiquity of Monocoty- ledons and Dicotyledons is much to be desired, but as vet the historical evidence is not definite, for no undoubted Monocoty- ledon has been recorded from strata older than those in which typical Dicotyledons first occur, and vice versa. The great and sudden prominence of the Dicotyledons in the Upper Cretaceous and Tertiary was long a puzzle, only relieved by the solitary Popuhis primaeva of the Lower Cretaceous, Comparatively recent studies, however, of contemporaneous beds in the United States and Portugal now regarded as Lower Cretaceous have thrown much light upon the subject, and since 1888 our knowl- edge of the origin of the Dicotyledons has increased rapidly. It should be remembered that the group is largely composed of herbaceous plants, and could not have a fair representation among fossil forms. Lowee Cretaceous Dicotyledons. — The dicotyledonous flora of the Lower Cretaceous was an abundant one, and is of great interest in the history of Dicotyledons. It consists of a plexus of forms, some of which are clearly related to existing Dicotyledons, others are clearly Dicotyledons but with no living n presentatives, while others are vague in their relationship to Dicotyledons. The few forms that can be referred with any FOSSIL ANGIOSPERMS 277 definiteness to modern groups are fairly submerged by the ex- tinct and vague types. Such a plexus is consistent with any evolutionary theory of the origin of Dicotyledons, and that it has been definitely discovered in the Lower Cretaceous is of great importance. Proangiosperms. — These are the vague forms referred to above as being not definitely Dicotyledons but suggestive of them. They are recognized by stem-structure and leaf-vena- tion, and seem to be related to numerous modern families, being good illustrations of so-called " comprehensive types." It is hardly to be doubted that many of them represent primi- tive Dicotyledons. If the Lower Cretaceous be divided into five periods, the Proangiosperms not suggestive of modern groups are the only dicotyledonous forms in the first. In the other periods they also occur, but in diminishing impor- tance as compared with the increasing number of recognizable forms. These clearly antecedent and for a time associated forms are very suggestive of their significant relation to modern Dicotyledons. Forms suggestive of Modern Groups. — After the first period of the Lower Cretaceous, forms suggestive of modern groups appear. They are so clearly Dicotyledons as not to be included among the Proangiosperms, but they are just as distinctly not modern types. Their generic names suggest the modern resem- blances, but these must not be taken to indicate relationships. For example, such names as Leguininosites, Menispermites, Myrsitwpihyllum, Proteophyllum, Peucedanites, etc., tell of cer- tain superficial resemblances, but may be very far from indi- cating real relationships. Modem Genera. — As already stated, no modern genera were associated with the Proangiosperms during the first period of the Lower Cretaceous. In the second period, however, an ex- tinct species of Populus has been recognized, the most ancient living genus of Dicotyledons known. In the third period Mag- nolia and Liriodendron are recorded; in the fourth Salix, Aris- tolochia, Sassafras, Adoxa, and Aralia appeared ; and in the fifth Myrica, Laurus, Eucalyptus, and Viburnum arc recorded. In considering this record of the Lower Cretaceous the fol- lowing thing- become evident: 1. The genera, so far as they are identical with living gen- 278 MORPHOLOGY OP ANGIOSPERMS era, are practically all members of the Archichlamydeae. The case of Viburnum, and even of Aralia, is peculiar, and perhaps suggestive of a far more complete development of the Dicoty- ledons than the records have shown. 2. The early appearance of Populus confirms the general primitive character of naked flowers and the anemophilous habit. 3. Xone of the known chalazogamic forms are represented in the above list, so that chalazogamv can hardly be regarded as a primitive character, as has been claimed, unless it be as- sumed that these earlier genera were chalazogamic and later became porogamic. 4. Of the twelve modern genera represented in the list, no less than eight are recognized by morphologists as primitive in character. 5. The occurrence of one of the Sympetalae in the upper- most member of the Lower Cretaceous, and that an epigynous form, needs explanation. It leads to at least one of three con- clusions. Either the determination is a mistake, or a large representation of sympetalous genera remain to be discovered in the Lower Cretaceous, or the present view as to the relative rank and phylogeny of sympetalous families must be modified. If the determination of Viburnum is the correct one, its associa- tion with Aralia is confirmatory of a genetic connection which we have Ions- maintained. 6. That epigyny had appeared among the undoubted Archi- chlamvdeae during the Lower Cretaceous is seen bv the exist- ence of such a genus as Eucalyptus. Upper Cretaceous Dicotyledons. — Much less is known of the flora of the Tapper Cretaceous than of the Lower Creta- ceous. There must have been a large development of existing genera, such as Sali.r, Populus, and Liriodendron being well known, as well as an introduction of new ones. Tertiary Dicotyledons. — The record of the dicotyledo- nous flora of the Tertiary is naturally made up of the trees and shrubs. The forest display was evidently as extensive and va- ried as now. In addition to the genera mentioned above, all of which show increasing development, there appeared the Betula- ceae, Fagaeeae, Juglandaceae, Moraceae, Proteaceae, Berberi- daceae, Staphyleaceae, Aceraceae, etc. This means an almost FOSSIL ANGIOSPERMS 279 complete display of the more primitive Archichlamydeae. A notable introduction during the Tertiary was that of the Legu- niinosae. That these appeared first only as Mimosa forms is a strong confirmation of the primitive character of this tribe, as well as of its possible relation to the Rosaceae. The above facts in reference to the early history of the Di- cotyledons seem to warrant the following conclusions : 1. The modern Dicotyledons were derived from a plexus of vague forms developed largely in the Lower Cretaceous and known as Proangiosperms. 2. The Cretaceous and Tertiary display is almost exclu- sively made up of Archichlamydeae, the dominant types being the more primitive Archichlamydeae. 3. The Sympetalae are practically absent from the Creta- ceous and Tertiary, and represent therefore a comparatively recent type. 4. The possible appearance of Viburnum, associated with Aralia, at the close of the Lower Cretaceous suggests a connec- tion of Lanbellales with the Sympetalae not recognized by tax- onomists. 5. Xone of the highly specialized groups of the Archichla- mydeae are represented in the Cretaceous and Tertiary, such a family as the Leguminosae being represented by its most primi- tive type, and all the types being what may be called " compre- hensive." 6. The identity of genera in the eastern and western hemi- spheres indicates the absence of continental diversities, which later became so striking a feature in geographical distribution. 7. The theory that simple flowers are necessarily reduced rather than primitive structures seems to have a complete refu- tation in the testimony of history. 19 CHAPTER XV PHYLOGENY OF ANGIOSPERMS The phylogeny of any great group will probably always remain a baffling problem. At the same time, theories of phy- logenv serve to coordinate knowledge and stimulate investiga- tion. The phylogeny of Angiosperms is an unusually obscure problem. The hypotheses proposed seem to include almost every possibility, but thus far they have been more interesting than convincing. When similarity of structure was taken as a sure indication of genetic relationships, the problem promised an approximate solution. But since it has been proved that similar structures may develop independently, the difficulty of solution has apparently become insurmountable. Under such circumstances it is questionable whether a discussion of the sub- ject is profitable, but a statement of the problem may not be out of place. The first phase of the problem has to do with the common or independent origin of the Monocotyledons and Dicotyledons. It has been assumed generally that the two groups are mono- phyletic. The chief argument, and in fact the only morpholog- ical one for the monophyletic theory, lies in the great uni- formity of the peculiar development of both the male and female gametophytes. It is argued that the independent origin of such exact details of development and structure is inconceivable, and this argument has been reenforced re- cently by the discovery in both groups of the peculiar phe- nomenon called " double fertilization." The argument is cer- tainly a very strong one. and yet there are rebutting proposi- tions. Even such similarity in structure may be the natural outcome of the changes that resulted in the evolution of seeds, and these are now generally believed to have appeared in inde- 280 PHILOGENY OF ANGI08PERMS 281 pendent lines. Again, the fundamental differences in the de- velopment of the embryos of the two groups are hard to recon- cile upon the theory of monophyletic origin. Add to this the fundamental differences in the structure of the stem and in the character of its vascular bundles, and the derivation of one group from the other seems more inconceivable than the deriva- tion of the Dicotyledons from the Gymnosperms. Still another argument against the monophyletic theory is furnished by the historical testimony. The Proangiosperms of the Lower Cre- taceous, so far as known, appeared associated with undoubted Monocotyledons, and merged gradually into recognizable Di- cotyledons, without indicating any relationship to the Mono- cotyledons. The emerging of Dicotyledons from this vague group either indicates that Monocotyledons and Dicotyledons originated independently, or that the Proangiosperms were transition forms between Monocotyledons and Dicotyledons. This latter alternative is in turn inconceivable, especially since the most primitive Dicotyledons are recognized to be even more primitive than any of the Monocotyledons. Recently, however, the morphological arguments in favor of the monophyletic origin of Angiosperms have been reen- forced by anatomical investigations, which point to origin from a common proangiospermous stock, or the derivation of the Monocotyledons from the more primitive Dicotyledons. In the following chapters it will be noted that on anatomical grounds Jeffrey regards the Monocotyledons as strictly monophyletic and modern, derived from the Dicotyledons or their parent stock ; and on the same ground Queva 9 thinks that the Mono- cotyledons are derived from the lower Dicotyledons. In her study of the origin of the cotyledon in Monocotyledons, Miss Sargant 18 concludes that the Monocotyledons are a specialized branch from the Dicotyledons. In Anemarrhena, one of the Liliaceae, she finds two opposed vascular bundles in the ter- minal cotyledon. These run down into the short hypocotyl, where each divides into two, and the four phloems so formed are continuous with those of the tetrarch primary root. This suggests that two cotyledons are represented, which were sepa- rate in some dicotyledonous ancestor. The same investigator also finds in Erianthis, one of the Ranunculaceae, a possible illustration of this dicotyledonous ancestor; for the petioles 282 MORPHOLOGY OF ANGIOSPERMS of the cotyledons are united throughout their length, showing two opposed bundles, as in the cotyledon of Anemarrhena. Attention should be called to similar cotvledonarv tubes in Dicotyledons, and since nearly all of these are geophilous plants Miss Sargant 21 has inferred that the fused condition of the cotyledons in the monocotyledons has arisen in connection with the geophilous habit. We herewith reproduce Miss Sargant's list of dicotyledonous seedlings with a well-marked cotvledonarv tube. Anemone coronaria, A. alpina, A. blanda, A. narcissiflora, A. rupi- cola, Ranunculus parnassifolius, Trollius Ledebouri, Erianthis hiema- lis, Delphinium nudicaule, D. hybridum and vars., Aconitum Anthora, Leontice vesicaria, L. altaica. Podophyllum peltatum, P. Emodi, Car- damine spp., Oxalis spp., Rhizophora Mangle, R. conjugata, Megar- rhiza californica, Smyrnium perfoliatum, S. rotundifolium, S. Olusa- truni, Bunium luteum, Chaerophyllum bulbosum, Prangos ferulacea, Serratula radiata, Dodecatheon Meadia, Polygonum Bistorta, P. sphae- rostachyum, and Rheum Moorcroftianum. Holm 7 has also studied the two completely united cotyledons of Podophyllum, which suggested to him the possibility that the " pair " may be regarded as a single cotyledon. In her study of the " monocotyledonous Dicotyledons," Miss Sargant 21 claims that the so-called single cotvledon is a fusion of two cotyledons, special reference being made to the well-known ease of Ranunculus Ficaria. It may be noted also that in 1896 Delpino 2 urged the origin of the monocotyledonous phylum from Dicotyledons through Butomus. Recently Hallier,20 bas- ing his phylogeny upon sporophylls and foliage leaves (" tro- phophylls"), has urged the origin of Monocotyledons from Dicotyledons, claiming that thev have arisen from the region of the Ceratophvllaeeae and Ranunculaceae. There can be no question that among the Xymphaeaceae, Ranunculaceae, and Berberidaceae there occur anatomical struc- tures very suggestive of Monocotyledons, as Campbell 15 has recently pointed out, but that this proves the origin of Mono- cotyledons from Dicotyledons rather than the reverse is not evident. Even the evidence derived from cotyledons has been taken by Lyon 14 as indicating that the dicotyledonous condi- tion has been derived from the gradual splitting of the single cotyledon of Monocotyledons. If the view of the phylogeny of PIIYLOGENY OF ANGIOSPERMS 283 the cotyledon maintained by Lyon 17 be true (see Chapter IX), the Monocotyledons are more primitive than the Dicotyledons and have given rise to them. It is an old view, however, that the Dicotyledons are the more primitive, and that the Monocotyledons have been derived from them as a reduction series. Later the relatively primi- tive character of the Monocotyledons was maintained without serious opposition. A detailed presentation of the phylogeny of Angiosperms from this point of view may be found in Bessey's 3 " Phylogeny and Taxonomy of the Angiosperms." In our judgment the evidence is strongly in favor of the independent origin of the two groups, which have attained prac- tically the same advancement in the essential morphological structures, but are very diverse in their more superficial features. Their great distinctness now indicates either that they were always distinct or that they originated from forms that were really Proangiosperms and neither Monocotyledons nor Dicotyledons. It may be well to state in this connection that in speaking of the origin of one great group from another, the former is not supposed to have arisen as a single branch. For example, to say that Monocotyledons have been derived from Dicotyledons does not imply that a single monocotyled- onous branch arose from some definite group of the Dicotyle- dons, but that probably several monocotyledonous lines arose from one or more regions of the Dicotyledons, regions that may or may not be illustrated by living groups. The next phase of the problem raises the question whether the Angiosperms have been derived from the Gymnosperms or directly from the Pteridophytes. The general question is the same whether one believes in their monophyletic character or not. The older view is that the Angiosperms have been derived from the Gymnosperms, and Gnetum has been regarded as the nearest living representative of a transition condition between Gymnosperms and Angiosperms. The argument is based upon certain resemblances of Gnetum to the Angiosperms, chief among them being the absence of archegonia, the organization of eggs while the gametophyte consists of free cells, the presence of a perianth and true vessels, and the Dicotyledon-like leaves. This showing is strong but perhaps not conclusive. If this origin be maintained, it is evident not only from the leaf char- 284 MORPHOLOGY OF ANGIOSPERMS acters, but still more from the nature of the embryo and the structure of the stem, that the primitive Angiosperm stock would be the Dicotyledons. Strasburger recognized this neces- sity when proposing the theory, and regarded the Monocotyle- dons as a reduced branch from the Dicotyledons; which is another reenforcement of the argument derived from recent anatomical investigations. In fact, the Gymnosperm ances- try of Dicotyledons also gains a point in the entire absence of pteridophytic anatomical features in the shoots of Dico- tyledons. Lately, also, Karsten,16 in a morphological study of the Juglandaeeae, emphasizes their resemblances to Gymnosperms, and concludes that the Angiosperms have been derived from such forms as Gnetum. The historical argument against such a claim is the absence of any certain evidence of the existence of Gnetum among the numerous Angiosperms of the Cretaceous and Terti- ary. If it were related in any way to the origin of such a group as the Angiosperms, it seems probable that it would have left some evidence of its existence. Of course this is negative evidence, and remains of ancient Gnetales may be found in the tropics or in the southern hemisphere. The argument from the pres- ence of a perianth is particularly vulnerable, since the so-called perianth merely represents the bracts common among Gymno- sperms, and the most primitive Dicotyledons and Monocotyle- dons have no perianth. Further, the presence of true vessels is an argument as much in favor of the origin of the Angrio- sperms from certain heterosporous Pteridophytes as from Gne- tum. Although we regard the origin of Angiosperms from Gymnosperms as very improbable, the embryo-sac structures of Gn'etum are suggestive of the way in which the character- istic sac-structures of the Angiosperms may have arisen from a compact gametophyte. This is all the more probable since the sac-structures of certain Juglandaeeae and of Peperomia pellucida have been found to be suggestive of those of certain species of Gnetum. If the Gymnosperms are not the ancestral forms of the An- giosperms, their direct derivation from the Pteridophytes be- comes a matter of course. The Pteridophyte that has been most persistently associated with the origin of Angiosperms is Isoe- tes. Its resemblances to the Monocotyledons have suaa-ested PHYLOGENY OF ANGIOSPERMS 2S."> that it may be the nearest living representative of their ancestral forms. Isoetes is a remarkably isolated group among the Pteri- dophytes, with no clear affinities, so that its own connection with the Pteridophyte stock is not evident. The most striking re- semblance to Monocotyledons occurs in the embryo, in which the single cotyledon is terminal and the stem-tip arises later as a lateral structure. The development of the male gameto- phyte resembles Angiosperms more than it does Gymnosperms, while the female gametophyte is equally suggestive. However, these gametophyte characters are shared by Selaginella. The general habit and vegetative structure of Isoetes bear some re- semblance to those of an aquatic Monocotyledon, and the anat- omy of the stem is suggestive of such forms as Yucca and Dra- caena. There can be no question that the resemblances of Isoe- tes to the Monocotyledons are more numerous than those of any other living Pteridophyte. The most telling resemblance is the character of the embryo, but the fact that its axis is trans- verse to that of the suspensor is a serious obstacle. Campbell has called attention to the fact, however, that in the embryos of Lilaea subulata and Zannichellia the apex of the root is not directed toward the suspensor but to one side, so that the axis of the embryo is oblique to that of the suspensor. A possible explanation of these laterally directed roots, however, is sug- gested by Murbeck's recent account of Ruppia (page 196), in which a primary root is formed with the normal orientation, but soon disorganizes, while a lateral root, formed very early, is the first functional one. As between the Gnetum origin of Angiosperms and the Isoetes origin of Monocotyledons the latter view must be preferred. Such a view, of eonrse, does not imply that the present Monocotyledons have been derived from the present Isoetaceae, but that the ancestral forms of the two were probably genetically connected. If this be true, doubtless Isoetes represents a reduced branch of some old stock that gave rise to the more vigorous Monocotyledons. The only possible alternative as to the origin of Monocotyledons, in case they have arisen independently of the Dicotyledons, seems to be to regard them as the end of a heterosporous line that developed independently from the eusporangiate Filicales, whose Pteridophyte members are extinct. Such an hypothes- is only necessary in the event that those based upon known 2S6 MORPHOLOGY OF ANGIOSPERMS structures prove to be insufficient; but the problem seems to have reached this contingency now. To many, any conclusion as to the origin of the Monocoty- ledons involves that of the Dicotyledons, which they would re- gard as an ancient branch from the Monocotyledon stock. We have already cited reasons why such a view does not commend itself to us, and prefer to regard Dicotyledons as of independent origin. If the two lines have a common origin, it seems to us that the arguments in favor of the derivation of Monocotyle- dons from the more primitive Dicotyledons are the more con- vincing. Both lines to-day include very primitive forms, and the structure of the flower and character of the megasporan- giate archesporium are more primitive among existing Dicot- yledons than among Monocotyledons. Whether Dicotyledons represent an independent angiospermous line, as we prefer to believe, or the primitive Angiosperm stock, it remains to dis- cuss their possible origin. The fact that they emerged from a primitive group called Proangiosperms, which was largely developed in the first period of the Lower Cretaceous, seems to be fairly well established by paleobotany. The question thus concerns the origin of the Proangiosperms. They do not seem to warrant the belief that they represent a common stock from which both Monocotyledons and Dicotyledons have been derived, for the Monocotyledons are believed to have ex- isted in unmistakable forms before the large assemblage of Pro- angiosperms gave rise to unmistakable Dicotyledons. Still less conceivable is it that the Proangiosperms represent the transition forms from Monocotvledons to Dicotvledons, for nothing in their known structure seems to suggest such a view. That they were derived from Gnetam-\ike forms is discredited by the fact that there is no sure record of the existence of Gnetum at such an early period, and to have given rise to such an assemblage of forms it must have been a conspicuous group. If we turn to the earlier groups that were sufficiently prom- inent and at all suggestive of having given rise to the Pro- angiosperms, we encounter the Coniferales, Cycadales, Lycopo- diales, and Filicales. The Gymnosperm-origin of Dicotyledons seems to be most unlikely with the exclusion of Gnetum. At the same time, it might be claimed that Dicotyledons represent PHYLOGENY OF ANGIOSPERMS 287 an independent line from the Gymnosperm-stock, that advanced in the Mime direction and much farther than did the Gnetum- line. At the same time, all the essential morphology of the Gymnosperms is less favorable to such an origin than is that of the heterosporous Pteridophytes. The Lvcopodiales certainly deserve serious consideration in this connection. The structures of Selaginella are about as suggestive of Dicotvledons as those of Isoetes are suggestive ~o *• CO of Monocotyledons, the embryo being as distinctly dicotyledo- nous as that of Isoetes is monoeotyledonous, and the seed-like character of the megasporangium supplies a still more striking resemblance. Such a view does not imply that the present com- paratively modern genus Selaginella has given rise to the Pro- angiosperms, but that the latter may have been derived from the same ancient Lycopodium stock. The only remaining alternative hypothesis is that mentioned in connection with the origin of the Monocotyledons, namely, the derivation of the Proangiosperms as an independent heteros- porous line from the abundant ancient eusporangiate Filicales, and this view is supported by anatomical testimony. It may be that further knowledge of the Proangiosperms will help to establish such an hypothesis. It seems to us that the last two hypotheses deserve the most consideration, as likely to include the future results of investi- gation. It should be noted in connection with the origin of Dicoty- ledons that there is much evidence in favor of the view that they include two independent lines. For example, Campbell inclines to the view that one line is derived from the Arales, passing by way of the Piperales and amentaceous groups to the isoear- pous Sympetalae, while the other arises from the apocarpous Helobiales, and by way of the Ranales and later groups cul- minates in the anisocarpic Sympetalae. Although not inclined to accept the origin suggested, the existence of two such inde- pendent lines of Dicotyledons has very much in ks favor, whether derived from the Monocotyledons or not. A summary of our present views, as developed in the preced- ing pages, may be stated as follows: The Monocotyledons and Dicotyledons represent two independent lines derived directly from Pteridophyte stock, probably from the Filicales. At the 2SS MORPHOLOGY OF ANGIOSPEUMS same time, the arguments in favor of the monophyletic origin of Angiosperms are strong; and if this view be accepted, the derivation of Monocotyledons from primitive Dicotyledons seems to rest on stronger evidence than the reverse relationship. It must also be said that the Gymnosperm origin of Angio- sperms is not to be discredited so much now as formerly. The student of the phylogeny of any group of vascular plants should be acquainted with certain general theoretical views. Among them the origin of the sporophytie generation is one of the most fundamental. Two theories are under dis- cussion, known as that of homologous origin and that of anti- thetic origin, names applied by Celakovsky. According to the former theory, the sporophyte is the lineal descendant of the sexless individuals common among Thallophytes and homolo- gous with the sexual individuals ; according to the latter the- ory, the sporophyte is a new structure intercalated in the life history of plants and holding no phylogenetic relation to any preceding individuals. The theory of homologous origin is re- ferred to Pringsheiin in 1876; that of antithetic origin was formulated by Celakovsky in 1877, but was presented in detail by Bower in 18i»0. In 1896 the theory of homologous origin was again brought into prominent notice by Scott in a presi- dential address before the British Association ; and two years later Bower, upon a similar occasion, defended the theory of antithetic origin. A general presentation of the subject by Klebs,4 Lang,5 and Hartog 6 followed, including the testimony of recent investigations. Undoubtedly the strongest argument in favor of the homologous origin of the sporophyte is derived from the phenomena of apogamy and apospory ; and among Ferns these have been coming to light so rapidly and are in- duced so readily that the powers of gametophyte and sporo- phyte, at least in this group, seem to be easily interchangeable, a fact most easily explained by their homologous character. It will be noted that in -dl this discussion there is no suggestion that sporophytes may have arisen in both of these ways, a possi- bility that will be considered a little later. One of the most suggestive theories of recent years i~ Bower's 1 theory of the strobilus. Xo better statement of its main points can be made than that of the author himself in his summary. PHYLOGEXY OF ANGIOSPERMS 289 1. Spore-production was the first office of the sporophyte, and the spore-phase has constantly recurred throughout the descent of the Archegoniatae ; the spore-bearing tissues are to be regarded as primary, the vegetative tissues as secondary, in point of evolutionary history. 2. Other things being equal, increase in number of carpospores is an advantage ; a climax of numerical spore-production was attained in the homosporous Vascular Cryptogams. 3. Sterilization of potential sporogenous tissues has been a wide- spread phenomenon, appearing as a natural consequence of increased spore-production. 4. Isolated sterile cells or layers of cells (tapetum) served in many cases the direct function of nourishing the developing spores, being themselves absorbed during the process. 5. By formation of a central sterile mass (columella, etc.) the spore- production was, in more complex forms, relegated to a more superficial position. 6. In vascular plants, parts of the sterile tissue formed septa, par- titioning off the remaining sporogenous tissue into separate loculi. 7. Septation to form synangia, and subsequent separation of the sporangia, are phenomena illustrated in the upward development of vascular plants. 8. Such septation may have taken place repeatedly in the same line of descent. 9. The strobilus as a whole is the correlative of a body of the nature of a sporogonial head, and the apex of the one corresponds to the apex of the other. 10. Progression from the simpler to the more complex type de- pended upon (a) septation, and (b) eruption to form superficial appen- dicular organs (spoi'angiophores, sporophylls) upon which the sporan- gia are supported. il. By continued apical growth of the strobilus, the number of sporophylls may be indefinitely increased. 12. The sporophylls are susceptible of great increase in size and complexity of form ; in point of evolutionary history, small and simple sporophylls preceded large and complex ones. 13. In certain cases foliage-leaves were produced by sterilization of sporophylls. This theory means that the leafy sporophyte is derived from such a sporopbytic structure as is displayed by the sporogonium of Bryophytes; but, as suggested by Klebs and Lang, it may have had an entirely independent origin, and may have no phvlogenetic connection with such a structure as a sporogonium. This view, together with its possible relations to the question of antithetic versus homologous origin of the sporophyte, has 290 MORPHOLOGY OP ANGIOSPERMS been discussed by Coulter,8 the substance of whose paper may be stated in the following extracts : It has been common to regard the distinct sporophyte as having been established once for all by the Bryophytes, and the sporophytes of the higher groups to have been derived from those of the Bryo- phytes. In searching for the origin of the leafy sporophyte, therefore, attention has been focused upon the sporogonia of Bryophytes. . . . The doctrine that any plant structure, however important, can have but one phylogeny, is hardly tenable at present. ... In contrasting the sporophytes of Bryophytes and Pteridophytes, they seem to have nothing in common except that they are usually derived from the oospore and represent an asexual generation. These facts are im- portant, but so are the numerous other facts in which they differ sharply. . . . It may be well to contrast the leafless and leafy sporophytes. In the former case the structure is never independent of the gametophyte, develops no lateral members, has nothing comparable to sporangia, and its whole tendency is to render complex the spore-producing region. In the latter case the sporophyte is dependent upon the game- tophyte only in its embryonic stage, develops prominent lateral mem- bers, has distinct simple sporangia, and its whole tendency is to render complex the sterile or nutritive tissues. As one traces the evolution of the Bryophyte sporogonia they give evidence of increasing com- plexity and hence rigidity, and little promise of originating such a diverse tendency as that shown by the sporophyte of Pteridophytes. . . . The origin of leaves on the gametophore of mosses suggests that leaves may develop in response to more favorable conditions for their work, and such development may result in the great reduction of chlorophyll work done by the less favored region, and its consequent simplification. It is evident that with the exchange of an aquatic for a terrestrial habit the thai lose body would not be a favorable type for chlorophyll work, and that the development of chlorophyll tissue upon erect structures of various kinds might follow. Among Bryophytes the erect structure laid hold of is the gametophore, and not the sporo- gonium. . . . In considering whether it is possible to disregard the Bryophytes in our search for the origin of the leafy sporophyte, we are largely influenced by the fact that the Bryophyte sporophyte, throughout its whole history, is dominated by a tendency which does not appear in the Pteridophyte sporophyte. Before the establishment of alternate generations the plant body may be said to have had three functions, namely, chlorophyll work, and the production of gametes and spores. The appearance of the Bryophyte sporogonium was dominated by the separation of spore-formation from the other functions, chlorophyll work being retained by the gametophyte, along with gamete-produc- PHYLOGENY OF AXGIOSPERMS 291 tion. Attention has been focused so long upon the gametes and spores as the two dominant factors in differentiation that it is hard to con- ceive of the possibility of the domination of another factor. It is entirely conceivable, however, that another form of differentiation may have occurred, dominated by the needs of the chlorophyll work, and not by spore-production. Certainly a great need for change, when aquatic conditions were exchanged for terrestrial, was in connection with the display of chlorophyll tissue. It would seem as if the Bryo- phytes had laid emphasis upon spore-production, and therefore never became organized for the fullest use of terrestrial conditions, while the Pteridophytes laid emphasis upon chlorophyll work and became highly organized for terrestrial life. It would seem possible, therefore, with the three factors to take into account, that two distinct asexual lines may have been organized, distinct in the factor selected to domi- nate. . . . If more favorable structures can be developed in response to the needs of spores or gametes, there seems to be no good reason why more favorable structures may not be developed in response to the needs of chlorophyll work. If such a response in structure is possible, it would naturally express itself first in developing the largest display of chlo- rophyll tissue in the most favorable region of the body, which would gradually become differentiated more and more distinctly from the rest of the body. It does not seem clear why the appearance of an erect leafy axis, bearing neither gametes nor spores, is not quite as supposable as the appearance of a sporophore with neither gametes nor leaves, or a gametophore with neither spores nor leaves. . . . With such an origin of the leafy sporophyte, it wrould follow that foliage leaves are not secondary but primary structures, and that sporo- phylls have arisen from the differentiation of foliage leaves bearing sporangia, a state of things certainly suggested by the most primitive Pteridophytes known. It would further follow that the evolution of the strobilus has followed the development of foliage leaves, a view in accordance with the older morphology. Such a view would make intelligible the great "gap" recognized as existing between Bryophytes and Pteridophytes, as the two groups would not be phylogenetically connected, and would have developed along very divergent lines from the first. It would mean that at least two independent sporophyte lines have appeared, the Bryophyte line probably with an antithetic origin, and the Pteridophyte line possibly with an homologous origin. The great prominence of the latter line, with its Spermatophyte sequence, is correlated with the development of a vascular system, and it would seem as though the evolution of an elaborate vascular system must have depended upon the domination of chlorophyll work. Knowledge of the various theories as to the origin of species is so much a part of the essential training of the morphologist 292 MORPHOLOGY OP ANGIOSPERMS that no resume of the subject is necessary. Until very recently, the various theories involve the idea that a species is produced as the cumulative result of slight variations through successive generations. In sharp contrast to this De Vries has recently proposed what is called the mutation theory, a brief statement of which may be of service. The experimental work that fur- nishes a substantial basis for the theory was conducted with Angiosperms, and a special student of the group should be pre- pared to recognize any testimony for or against it. A suggest- ive feature of the work of De Vries is his attempt to break away from the speculative method and to subject the problem to experimental investigation. Whether his results indicate a general method of the origin of species in nature or an occa- sional method, or are capable of an entirely different expla- nation and hold no relation to the normal origin of species, remains for future work to determine. In any event, the theory will stimulate investigation and deserves consideration. The occasional sudden appearance of what have been called " sports " is well known, but they have not been prominently associated with the origin of species. They have been referred to as cases of " saltatorv evolution," and in 1864 Kolliker seems to have been responsible for the term " heterogenesis " as applied to this phenomenon. Quite independently and sim- ultaneouslv De Vries n and Korschinskv 12 have elaborated the same theory as to the origin of species, the former calling it the " mutation theory," the latter using Kolliker's name " hete- rogenesis." Korschinskv has brought together a mass of data from the records of gardeners and horticulturists to show that most of the culture " varieties " have arisen through heterogene- sis rather than by selection. De Vries, on the other hand, has experimented extensively with (Fjuoihera Lamarckiana, a spe- cies showing mutability in a high degree. This American spe- cies was found naturalized on an area in Holland about IS 75, and afterward spread rapidly. When observed by De Vries, in 1886, two new species were detected among the normal forms, and they have maintained themselves ever since. From 1886 until the publication of his book, De Vries made observations upon the naturalized areas and carried on cultures in the botan- ical garden at Amsterdam. As a general result, it may be stated that out of 50,000 seedlings of (E. Lamarckiana 800 were mu- PHYLOGENY OF ANGIOSPERMS 293 tants. Of these 800, about 200 were the new species named (E. lata ; that is, the same new species appeared about 200 times. Various other new species appeared, and were preserved by culture. The mutants also occurred in every direction in the same environment, showing no indication of being responses to external conditions. In the great majority of cases the mu- tants were constant from the outset, there being no development and fixation of characters through selection, and no transition between parent and offspring. Experiments with other species seem to indicate that the majority of species are at present immutable, varying within certain narrow limits, but not giving rise to mutants. Solms-Laubach 10 has shown that in all probability Cap- sella Heegeri has arisen in this way from C. B ursa- pastor is ; and Carlson 13 has suggested a similar origin for certain Swed- ish forms of Succisa pratensis; while Jordan's work with Draba verna has discovered about 200 immutable forms within the old species limits, that probably represent true species derived by mutation from a parent of great mutability. The experiments of De Vries seem to indicate that there is a definite limit to individual variability, beyond which selection can not go. Furthermore, it is claimed that selection never fixes a character, but reversion may occur after any number of gen- erations of culture. In short, natural selection can not create anything new, but can modify within definite and narrow lim- its ; while mutation brings into existence something new, which will continue as a new species if it can survive the struggle for existence. There is thus drawn a sharp contrast between muta- bility and ordinary variability, the latter being governed by environment, the former independent of it. Hence, while most species are immutable, all are more or less variable. At its present stage such a theory can not be accepted or rejected. Either alternative will demand a vast amount of care- fully sifted experimental evidence. It should be remembered that the subject lends itself readily to observations that are really inferences, and a vast amount of data will doubtless be forthcoming that can not be regarded as testimony. The stu- dent of Angiosperms, however, is in a position to encounter useful data, for the group is a very modern one and seems to contain many mutable species. It should further be remem- 294 MORPHOLOGY OP ANGIOSPERMS bered that the whole theory is based upon the present concep- tion of species, a conception so variable that it can not be defined. Furthermore, although there may be a fixed limit to ordinary variation, there must also be a fixed limit to the extraordinarv variation called mutation, and this remains to be defined. In fact, there is evidence that extreme mutation re- sults in functional derangement of organs, and the result is a monstrosity, which may be regarded as an impossible new spe- cies. Finally, even if mutation be found to explain the origin of many new species, it does not follow that other processes also may not be working to the same result. In a recent paper, Strasburger 19 takes occasion to discuss the origin of species, taking the view that the results of natural selection have been overestimated, and that new species have arisen through mutation, due to internal causes alone, and through " use and disuse," by means of which a certain amount of adaptation to environment is secured. To him the only func- tion of natural selection appears to be to remove the less valu- able forms produced through mutation and " use and disuse." It follows that the ordinary physiological operations do not result in species, but affect them after they have appeared, and that the origin of species is a morphological rather than a phys- iological problem. LITERATURE CITED 1. Bower, F. O. A Theory of the Strobilus in Archegoniate Plants. Annals of Botany 8 : 343-365. 1894. 2. Delpixo, P. Applicazione de nuovi criterii per la elassificasione delle piante. Mem. Real. Accad. Sci. Bologna V. 6: 83-116, 1896 ; see review Bot. Central bl. 67 : 370. 1896. 3. Bessey, C. E. Phylogeny and Taxonomy of the Angiosperms. Bot. Gazette 24: 145-178. 1897. 4. Klebs, G. Alternation of Generations in the Thallophytes. An- nals of Botany 12 : 570-583. 1898. 5. Lang, W. H. Alternation of Generations in the Archegoniates. Annals of Botany 12: 583-592. 1898. 6. Hartog, M. Alternation of Generations. Annals of Botany 12: 593-594. 1898. 7. Holm, Theo. Podophyllum peltatum; a Morphological Study. Bot. Gazette 27 : 419-433. figs. 10. 1898. 8. Coulter, J. M. The Origin of the Leafy Sporophyte. Bot. Gazette 28: 46-59. 1899. PHYLOGENY OF ANGIOSPERMS 295 9. Queva, C. Contributions a l'anatomie des Monocotyledonees. I. Les Uvulariees tubereuses. Lille. 1899. 10. Solms-Laubach, H. Cruciferenstudien. 1. Capsella Heegeri Solms, eine neue entstandene Form der deutschen Flora. Bot. Zeit. 58i : 167-190. pi. 7. 1900. 11. De Vries, H. Die Mutationstheorie, Versuche und Beobachtungen fiber die Entstehung von Arten im Pflanzenreich. Vol. I. Leip- zig. 1901. See reviews: Biol. Centralbl. 21: 257-269, 289-305. 1901; Bot. Centralbl. 87: 170. 1901; Bot. Gazette 33: 236. 1902. Also The Origin of Species by Mutation, Science 15 : 721-729. 1902. 12. Korschinsky, S. Heterogenesis und Evolution. Ein Beitrag zur Theorie der Entstehung der Arten. Translated from the Russian by S. Tschulok. Flora 89: 240-363. 1901; also review in Bot. Gazette 33: 396. 1902. 13. Carlson, G. W. F. Ett par afvikande former af Succisa praten- sis. Bot. Notiser 1901 : 224-226. 14. Lyon, H. L. Observations on the Embryogeny of Nelumbo. Minn. Bot. Studies 2: 643-655. 1901. 15. Campbell, D. H. On the Affinities of Certain Anomalous Dicot- yledons. Amer. Nat. 36: 7-12. 1902. 16. Karsten, G. Ueber die Entwickelung der weiblichen Bliithen bei einigen Juglandaceen. Flora 90: 316-333. pi. 12. 1902. 17. Lyon, H. L. The Phylogeny of the Cotyledon. Postelsia 1901 : 55-86. 1902. 18. Sargant, Ethel. The Origin of the Seed-leaf in Monocotyledons. New Phytologist 1 : 107-113. pi. 2. 1902. 19. Strasburger, E. Ein Beitrag zur Kenntniss von Ceratophyllum submersum und phylogenetische Erorterungen. Jahrb. Wiss. Bot. 37: 477-526. pis. 9-11. 1902. 20. Hallier, H. Beitrage zur Morphologie der Sporophylle und des Trophophylls in Beziehung zur Phylogenie der Kormophyten. Jahrb. Hamburgischen Wiss. Anstalten 19: 1-110. 1902. 21. Sargant, Ethel. A Theory of the Origin of Monocotyledons, founded on the Structure of their Seedlings. Annals of Botany 17: 1-92. pis. 1-7. 1903. 20 CHAPTER XVI COMPARATIVE ANATOMY OF THE GYMNOSPERMS AND THEIR ALLIES * The skeletal structure of vascular plants has in the past been used for phylogenetie purposes to a much smaller extent than that of the higher animals. During recent years, however, important advances in our knowledge of the anatomy of fossil plants have made it apparent that the primary fibrovascular skeleton of the vascular plants is even more conservative than their reproductive organs, and consequently of great impor- tance in arriving at the relationships of the larger groups. The most extreme ecological conditions, acting for long periods, seem to have little effect in modifying the essential features of the primary fibrovascular framework, so that, for example, the extremely xerophytic cactus and the hydrophytie water-lily have exactly the same type of skeleton from the standpoint of comparative anatomy. It sometimes happens, however, that the woody framework is extremely complex in the adult. Re- cent investigations which cover the whole field of living vascular plants make it clear that the study of the development of the sporeling or seedling provides a satisfactory key to the inter- pretation of the most intricate skeletal structures of maturity. A brief account of certain general results of recent anatom- ical and developmental research in the case of the vascular plants is accordinolv necessarv for an understanding' of those skeletal features of the Gymnosperms and their allies which are of phylogenetie importance. PTERIDOPHYTES The simplest type of stem in the Pteridophytes is that in which there is present a single pithless fibrovascular conductive * Contributed by Professor Edward C. Jeffrey, of Harvard Universit .-. 296 COMPARATIVE ANATOMY OF UYMNOSPERAIS 297 strand embedded in the parenchyma of the fundamental tissue. Part of a transverse section of such a stem is seen in Fig. 108, A. In the center is the concentric fibrovascular bundle or stele, which consists of a mass of xyleni completely surrounded by phloem. The stele or central cylinder is bounded in turn by brown sclerenchymatous fundamental tissue. This type of stem, since it is a very primitive one, may convenientlv be railed " protostelic " (Jeffrey19). Another common condition of the stem is seen in Fig. 108, B, which represents a cross-section of the rhizome of Adiantum pedatum. In this case the central cylinder is not a solid fibro- vascular strand as in the preceding example, but a hollow cyl- inder filled with fundamental tissue like that external to the stele. The plane of section is just above the point of origin of a leaf-trace, which may be distinguished as the smaller of the two concentric masses of fibrovascular tissue. At a higher level the gap in the cauline central cylinder closes, and the stele be- comes circular instead of crescentic in cross-section. Similar gaps appear above all the outgoing leaf-traces, and as a conse- quence the central cylinder is essentially a concentric fibrovas- cular tube, with gaps in its walls corresponding to the leaf- traces. The type of central cylinder which has just been de- scribed may appropriately be termed " siphonostelic " (Jef- frey 19 ) . Fig. 108, C, is from a photograph of the adult stem of Pteris aquilina, the common bracken fern. In this case there are numerous concentric fibrovascular bundles present in the fundamental tissue of the rhizome, and accordingly stems of this type have been designated by Van Tieghem " polystelic." It has been shown, however, that in such stems as are exempli- fied by P. aquilina the primitive condition of the central cyl- inder is a stelar tube with foliar lacunae (Jeffrey11'). Fig. 108, D, from the young stem of P. aquilina, sufficiently demon- strates the truth of this statement. The voung stem gradually passes into the condition represented in Fig. 108, C, first by the foliar gaps becoming so long as to overlap, and second by the derivation of the large central strands from the inner wall i f the primitive stelar tube. ( 'onsequently the stem of P. aqui- lina may be regarded on ontogenetic grounds as siphonosteli<- and essentially similar to that of Adiantum pedatum. 298 MORPHOLOGY OP ANGIOSPERMS Fig. 108, E, shows a type of central cylinder which at first sight appears very like that of the adult stem of Pteris aqui- lina ; above on the right is a gap in the tubular stele, which in this case corresponds to a branch. Laterally, on the left, a foliar trace is to be seen in the fundamental tissue. The leaf- trace is very small, and there is no gap in the central cylinder corresponding to it. As in P. aquilina, there are two medullary fibrovascular strands. It has recently been shown (Jeffrey 13, 19) that in certain great groups of plants foliar gaps are constantly present, while in other great groups they are unfailingly ab- sent. The type of tubular stele characterized by the presence of foliar gaps has been called " phyllosiphonic," and that pos- sessing only gaps for the branches or ramular lacunae " clado- siphonic." These distinctions are extremely constant, and con- sequently of great phylogenetic value. Fig. 108, F, is from a photograph of a cross-section of the central cylinder of Osmunda Claytoniana. It is of special in- terest because it is obviously of the same type as the central cylinder of the living Gymnosperins, viz., a ring of collateral bundles surrounding a medulla and separated from each other by medullary rays. Van Tieghem 4 regards this type of stele as derived by dilatation from the prostostelie condition, with the formation of pith and medullary rays from the stelar pa- renchyma. According to this view, the pith and rays are mor- phologically different from and have nothing in common with the fundamental tissue surrounding the stele. Fig. 109, G, shows the forking of the central cylinder of Osmunda cinnamomea. In this example the pith is obviously continuous with the external cortex, and a strand of the very characteristic brown sclerenchymatous tissue of the cortex is passing down into the medullary parenchyma through the gap between the divisions of the fork. It is to be noted further that the phloem passes inward around the divisions of the fork for a considerable distance, and the endodermis is as well marked on the inside as on the outside of the crescentic zones of bundles. In Fig. 109, II, there appears a not unusual con- dition of the central cylinder in 0. cinnamomea. Unlike 0. Claytoniana, there is present an internal endodermis along the inner margin of the bundles, and the medulla is often charac- terized by the presence of a mass of brown sclerenchyma similar COMPARATIVE ANATOMY OF GYMXOSPERMS 299 to that which constitutes the external portion of the funda- mental tissue of the stem. Fig. 109, /, shows a central cylinder of Osmunda clnna- momea, where not only an internal endodermis is present but also internal phloem as well. In Fig. 109, •/, a part of the wall of the same central cylinder is shown more highly mag- nified. The sieve-tubes are easily recognized as large, appar- ently empty elements. It has been suggested by Jeffrey 10 and Faull,18 as a result of the study of the anatomy of the whole order, that the type of central cylinder found in the Osmunda- ceae is the result of reduction from a siphonostelic condition with internal phloem. This view of the matter is strengthened by the fact that brown sclerenchyma is sometimes found in the pith of Osmunda regalis and Todea barhara, although in these species there is no longer any communication between pith and cortex in the region of forking. Moreover, exactly similar series of degeneration to that supplied by the Osmundaceae have been shown to exist in the case of certain poly pod iaceous ferns. Hence it may be assumed, in the present connection, that the type of central cylinder exemplified by the Osmunda- ceae has arisen by degeneracy from the siphonostelic type with internal phloem ; and that the medulla often shows signs of its origin by striking histological resemblance to the cortex, even when there is no longer any communication between the med- ullary and cortical fundamental tissues. Fig;. 109, A", shows the structure of one of the tracheary strands of Osmunda cinnamomca. The protoxylem or primi- tive wood appears as a cluster of small elements, just external to a mass of wood-parenchyma.- The protoxylem does not abut immediately on the pith, as in the seed-plants, but is separated from it by a considerable amount of wood-parenchyma and me- taxylem r most of the metaxylern, however, lies external to the protoxylem. This type of tracheary bundle is very character- istic of the ferns, and has been designated " mesarch." In the case of the Lycopodiales, the tracheary bundle is of still another type. If Fig. 108, E, be examined, it will be seen that on the left of the central cylinder, opposite the leaf-trace in the cortex, is a cluster of protoxylem. The primitive wood in this case is external and next the phloem. This feature is very characteristic of the Lycopods and their allies. Bundles 300 MORPHOLOGY OF ANGIOSPERMS of the tvpe just mentioned have been designated by Scott e " exarch." Hence it may be stated that the bundles of the Fern- like plants are characteristically mesareh ; that the Lycopods and their allies have exarch bundles ; and that the prevailing type in the Spermatophytes is the endarch bundle, the primitive wood here coming next the medulla. These anatomical distinc- tions, however, are less trustworthy than those depending on the presence and absence of foliar gaps, for many Ferns have endarch bundles, while some (Lygodium, etc.) have even exarch traeheary strands; on the other hand, Phylloglossum, a recog- nized Lycopod, has distinctly mesareh cauline bundles. There are no known examples, however, of siphonostelic Lycopods (Jeffrey10) with foliar gaps, or of siphonostelic Ferns without them. CYCADOFILICES Recently Potonie11 has established a group, the Cycado- filiees, to include a number of fossil forms which are neither true Ferns nor typical Gynmosperms, but which possess to a large degree anatomical features of both alliances. These forms can now be more advantageously discussed after the general anatomical account presented in the foregoing paragraphs. The vegetative anatomy of the Cycadofilices is of special importance, both because of our entire ignorance of their reproductive or- gans at the present time and because their anatomical structure presents such an interesting transition from the pteridophytic to the gymnospermous type. Heterangium. — Fig. 109, L, taken from Scott's admirable Studies in Fossil Botany, shows the structural features of the stem of Heterangium Grievii, a primitive representative of the . ' Cycadofilices. The central cylinder is obviously protostelic and •,, very similar to that of Gleichenia fldbellata of Fig. 108, A. A striking difference, however, is the presence, on the outside of the pithless primary wood, of a narrow zone of secondary' wood which is clearlv distinguishable by reason of the regular radial arrangement of its elements. In the cortex may be seen leaf-traces and groups of sclerotic cells. The external cortex is bounded by a very characteristic hypodermal zone, which in transverse section appears to be made up of alternating stripes of parenchymatous and sclerenchymatous cells. Viewed longi- COMPARATIVE ANATOMY OF GYMNOSPERMS 301 tudinally, the hypoderma is seen to be composed of a tangential network of sclerenchymatous fibers having the meshes filled with parenchyma. Medullosa. — Fig. 110, M, reproduces a diagrammatic trans- verse section of the stem of Medullosa anglica. On the outside of the stem can be distinguished the same curious hypoderma which is characteristic of the genus briefly described above. The central cylinder in this case, however, is obviously not pro- tostelic, but polystelic. Each of the large fibrovascular strands is characterized by the presence of a considerable zone of sec- ondary wood, which is indicated in the diagram by radiating lines. There are no sclerifications in the cortex ; but numerous mucilage duets, similar to those of the Marattiaceae and the ('reads, mav be seen in the fundamental tissue, both outside and between the large fibrovascular strands, although their occur- rence in the latter position is not shown in the diagram. Very often the arrangement of the bundles in species of Medullosa was much more complex than that appearing in Fig. 110, M. It has been shown recently that in ferns with even the most complex arrangement of the bundles in the adult, by following the development it is possible to arrive at the simple stelar tube as a starting-point (Jeffrey19). It is consequently extremely probable that the bundle system of the Medullosae is to be regarded as primitively siphonostelic, like that of Pteris aquilina. In Fig. 110, N, is represented a cross-section of a part of the stem of Medullosa Sohnsi. Here are to be seen numerous bundles, some of which are broad and plate-like and others small and rounded in outline. The broader bundles are known as " plate-rings," and the smaller ones as " star-rings." An interesting feature of the outer plate-rings is the fact that the zone of secondary wood on the external face of the bundles is often very much thicker than that on the internal face. This peculiarity is especially well marked in old stems of M. stellata. Lyginodendron. — Fig. 110, O, taken from Williamson and Scott,6 reproduces admirably the general features of structure of the stem Lyginodendron Qldhamium. On the outside is the same curious hypoderma] layer which occurs in Heteran* gium and Medullosa. There is present also a zone of periderm external to the fibrovascular tissues. In the cortex may be seen 302 MORPHOLOGY OF ANGIOSPBRMS clusters of sclerenchymatous tissue. These are also found in the foliar gaps and in the pith. In the case of Lyginodendron the primary wood is comparatively poorly developed and occurs as distinct islands along the margin of the medulla. The sec- ondarv wood is characterized bv the regular radial seriation of its elements and is abundant, but, in common with many other fossil Pteridophytes with secondary growth, shows no indica- tion of annual rings. The continuity of the woody zone is completely interrupted at intervals by the foliar gaps which subtend the outgoing leaf-traces. Fig. 110, P, is a photograph of part of the ligneous zone of L. Oldhamium. The protoxylem, distinguished by the small size of its elements, is seen to be embedded in the primary wood. Most of the primary metaxvlem lies on the medullary side of the protoxylem, and a smaller portion between it and the sec- ondary wood. Hence the primary bundle is mesarch, as is often the case in the Ferns and their allies. Another important fili- cinean feature is the presence of well-marked foliar gaps. Fig. 110, Q, taken from Williamson and Scott,6 shows an interesting departure from the usual state of affairs in L. Old- hamium: a primary wood-bundle is present, and external to it is the usual secondary wood. In this case, however, there is secondary wood and phloem on the medullary side of the bundle as well. The condition represented in the figure is quite un- usual in L. Oldhamium ; but, as has been shown by Seward, is of common occurrence in L. robustum. The facts just described are of particular interest, because Scott 6 has made a specific comparison between the central cylinders of Lyginodendron and Osmunda ; and indeed, if we imagine a secondary zone of wood present in the latter genus and the primary wood-bundles cor- respondingly reduced in size, the resemblance becomes very close. The occurrence of internal phloem and secondary wood is paralleled by the discovery of internal phloem in 0. cinna- momea. The forms described above sufficiently illustrate the variety of structure in the stem of the Cycadofilices, and it now be- comes necessary to discuss their phylogenetic significance. First ■>f all is to be noted the fact that they represent the three types of stelar structure described at the beginning of the chapter: Heterangium being protostelic like Gleichenia ; Medullosa sipho- COMPARATIVE ANATOMY OF GYMNOSPERMS 303 nostelic like Adiantum pedatum and Pteris aquilina; and Lygi- nodendron siphoncstelic, without internal phloem, as is gener- ally the case in Osmunda, but resembling this genus in the occa- sional occurrence of internal sieve-tissue. The only striking anatomical difference between the cycadofilieinean forms de- scribed above and the parallel cases from the ferns lies in the absence of secondary growth in the latter. This feature is now known to be of minor importance, although great weight was attached to it by the Brongniartian school of paleobotanists. In regard to the particular type of the Cycadofilices which gave rise to the Gymnosperms there is some difference of opin- ion. Potonie,9, X1 YVorsdell,10, 17 and Jeffrey 19 consider the ( 'vcads to be derived from Mrdullosa-like ancestors through a Lyginodcndr on-like phase, by the gradual disappearance of the internal secondary wood, and the final suppression of the cen- tripetal primary wood. Scott,6, 15 on the other hand, regards Lyginodendron as the ancestral type, and as derived directly from Heterangium by the formation of an intrastelar pith, and not from medullosan ancestors by reduction. He further con- siders the Medullosae to constitute merely a side branch of the phylogenetic tree, and expresses the opinion that " we should involve ourselves in unnecessary complications if we endeav- ored to derive the simple primary structure of the cycadean stem from the more elaborate organization of a Mcdullosa." However, examples of phylogenetic progression from the com- plex to the simple are not at all uncommon. Striking illustra- tions of this principle are afforded by the derivation of the simple hyoid bone of the mammals from the complex hyoid apparatus of the lower vertebrates, and the evolution of the monodactyl horses of the present day from their four-toed an- cestors of the Eocene. The histological structure of the medulla in Lyginodendron strikingly resembles that of the cortex in the presence of sclerotic nests, and this feature indicates strongly community of origin of the medullary and cortical tissues. Further, the occasional occurrence of internal phloem and in- ternal secondary wood in Lyginodendron can most easily be ex- rilaine'd as a vestigial relic of a siphonostelic condition, in which internal phloem was normally present — i. e., a Medullosa with a single series of bundles. In regard to the special pteridophytic ancestry of the Cyca- 304: MORPHOLOGY OP ANGIOSPERMS dofilices there now seems to be little doubt. Scott has pointed out that their fern-like foliage and usually mesarch bundles indicate strongly a filicinean as opposed to a lycopodinean ori- gin. It has further recently been shown that they are phyllo- siphonic (Jeffrey 19), and since this feature is quite exclusively characteristic of the ferns, it seems impossible to derive the Cycadofilices from the Lycopods, as has been done by Renault.2 CYCADALES The leaves and fern-like habit of the Cycads afford good external evidence of their filicinean origin, and their multicili- ate sperms point in the same direction. The strongest evidence of their having come from the ferns, however, is supplied by their fibrovascular anatomy. Fig. Ill, R, is from a photograph of a cross-section of the stem of Zamia floridana. Both pith and cortex are occupied, as in Medullosa, by numerous mucilage ducts. In the cortex several curved lines are present, which represent the curved course of the foliar traces and are known as " girdles." Al- though some years old, the fibrovascular zone is quite narrow, and shows no evidence of annual rings, a feature of resemblance to the Medullosae and Lyginodendron. In Fig. Ill, S, the central cylinder of the same species is shown more highly magnified. Its continuity is obviously broken by gaps, which subtend the outgoing leaf-traces. The mucilage ducts of the medulla join with those of the cortex through the foliar gaps. The central cylinder of Zamia, which is quite typical of the Cycads in this respect, is consequently phyllosiphonic. The mucilage ducts of the Cycads do not pene- trate into the leaf-traces or root-steles. Hence it may be as- sumed that, as in the Marattiaceae and Medullosae, they are characteristic only of the extrastelar fundamental tissue. The pith of the Cycads, which contains mucilage ducts continuous with those of the cortex, is to be compared, therefore, with the mucilaginous medulla of one of the Marattiaceae or of a Med id- losa, and is to be regarded as extrastelar. The foliar traces of the Cycads are quite unique in struc- ture and of considerable phylogenetic importance. The first complete description of them was given by Mettonius.1 As has already been pointed out, the course of cycadean leaf-traces is COMPARATIVE ANATOMY OF GYMNOSPERMS 305 peculiar ; for, instead of passing directly from the central cyl- inder into the leaf, they usually pursue a circular course, so that they reach their corresponding leaf on the opposite side of the stem from their point of origin. .In Zamia I have observed this arrangement of the traces even in the seedling; but in Cycas, according to Mettenius,1 the leaf-traces of the young plant at first pursue a direct course, although at a later stage girdles are present. During their cortical course the foliar traces often undergo more or less complex anastomoses. The structure of the strands in the cortex, and even in the base of the petiole, is often concentric. Fig. Ill, T, is from a photograph of a cortical bundle of Cycas revohda. The center of the bundle is composed almost entirely of the large tracheitis of the primary wood, which is surrounded by the radially arranged secondarv wood and phloem. Higher up, in the lower part of the petiole, the bun- dles lose most of their secondary wood and assume mesarch structure. This is well seen in Fig. Ill, U, which may be compared with Figs. 109, K, and 110, P. A striking feature of the bundle at this stage is that the primary wood is mostly centripetal, and has consequently .a markedly cryptogamic ap- pearance. Before discussing further the significance of the peculiar structure of the foliar traces of the Cycads, it will be con- venient to refer to an interesting discovery made by Scott.7 Mesarch bundles have been found bv him in the central cvlinder of the peduncle of the cone of Stangeria paradoxa and certain other Cycads. The conservatism of reproductive organs is rec- ognized by the universal use made of them in botanical classi- fication. It is Scott's opinion that in the conservative repro- ductive branches (i. e., cones) of certain living ('vends the an- cestral type of bundle is retained. Hence he believes that the cauline central cylinder of the more or less remote ancestors of the living Cycads must have had a structure similar to that of the stem of Lyginodendron. This hypothesis is borne out by the- fact that the course of the leaf-traces in the cones of Cycads is the same as in the seedling of the genus Cycas, and in the vegetative stems of the extinct group of Cycad-like Ben- nettitales; for they pass directly into the leaves (sporophylls) and do not form girdles. Jeffrey 10 has pointed out a similar 306 MORPHOLOGY OF ANGIOSPERMS conservatism in the structure and course of the bundles in the cones of Equisetum. Leaf-traces are likewise extremely conservative in structure, for where eenogenetic modifications are present in the ordinary cauline strands, the primitive type of fibrovaseular bundle is often retained in the leaf-traces, as well as in the reproductive axis and in the seedling. Ancestral features are retained more- over in the leaf-traces, especially those of the cotyledons, long; after they have disappeared elsewhere. Hence it is assumed that the mesarch structure of the foliar bundles of the Cycads supplies a further argument for their derivation from ances- tors like Lyginodendron. The fact that cycadean leaf-traces are often concentric in the lower part of their course has been used as an argument by Worsdell 16 in favor of the hypothesis that the cauline bun- dles of the ancestors of the Cycads were originally concentric. This argument seems to have the same force as the similar argu- ment in the case of the mesarch collateral bundles ; and the fact that concentric strands are comparatively rarely present in the living Cycads is probably due to the concentric condition being further in the phylogenetic background. The structure of the conservative tracheary strands of the leaves and peduncles of the Cycads would seem to point to a more immediate ancestry with the general organization of Lyginodendron, derived in the remoter past from forms like Medullosa. BENNETTITAT.ES The external vegetative features and the reproductive organs of this interesting group have already been dealt with in the companion volume treating of Gymnosperms (p. 142). Al- though their reproductive organs differ very strikingly from those of anv living Cvcads, the fibrovaseular anatomv of the Bennettitales is strikingly cycadean (Scott15). They possessed a large cycadean pith penetrated by mucilage canals and bound- ed by a thin fibrovaseular ring. The continuity of the fibro- vaseular zone was broken at intervals opposite the large leaf- Traces, which separated in the cortex into arcs of bundles pass- ing directly into the leaves. The direct course of the foliar bundles is to be compared with that present in the cones only of living Cycads. This condition is probably to be regarded COMPARATIVE ANATOMY OF GYMNOSPERMS 307 as ancestral, because it occurs also in cycadean seedlings. The foliar bundles of the Bennettitales were characterized by the same peculiarities as those of the more modern Cycads. COKDAITALES On page 135 of the companion volume treating of Gymno- sperms, the reproductive features and general morphology of this interesting alliance are sufficiently described. The central cylinder of the Cordaites enclosed a large pith, and was charac- terized by considerable secondary growth. Like the Cycads and unlike the Conifers of the present day, the secondary wood gen- erally showed no annual rings. The wood of Cordaites, in some cases at least, is to be identified with Araucarioxylon and Da- doxylon, fossil woods which occur as far down in the strata as the Devonian. Scott 14 has shown that in some species of Araucarioxylon the primary wood of the stem was mesarch. In a good many cases, however, the primary cauline bundles of Cordaites are only distinguished by exceptionally large de- velopment as compared with those of the higher living Gymno- sperms. The leaf-traces were mesarch like those of the Cycads, and Scott 15 compares the structure of a cordaitean leaf to that of a pinna of Zamia. Fig. Ill, V, shows the structure of a transverse section of part of a leaf of a species of Cordaites. The organization of the cauline and foliar bundles of the Cordaites favors the view of their derivation from a pterido- phytic ancestry quite as much as that of their reproductive organs. Their well-marked foliar gaps and their large leaves clearly indicate their filicinean affinities. The thickness of the woody cvlinder and the freelv branching habit of the Cordaites indicates a greater proximity to the Coniferales than to the Cvcadales. GINKGOAT.ES The discovery of multiciliate sperms in Ginkgo is good evi- dence for the antiquity and the affinities of the group. Still, its pteridophytic features have suffered very considerable re- duction as compared with the Cvcadales. Evidences of mesarch structure are accordingly comparatively scanty. The bundles of the stem are throughout endarch, and even the leaf-traces show slight traces of the presence of centripetal wood. Wors- 308 MORPHOLOGY OP ANGIOSPERMS dell,8 however, has found that the bundles of the cotyledons show fairly well-developed cryptogamic wood. Fig. Ill, IV, taken from Worsdell, makes the truth of this statement appar- ent. The anatomical evidence leads to the conclusion that we have in Ginkgo a comparatively modern genus as compared with the living representatives of the cyeadean stock. Distinct foliar gaps are present, which, taken together with the large leaves and the multiciliate sperms, point strongly to a filicinean an- cestry. COXIFEKALES The Coniferales are the prevailing Gymnosperms of the present day, and it is not surprising that they should present few anatomical features which can be considered ancestral. Their usually small acicular leaves offer a striking contrast to the large fern-like foliar organs of the older gymnospermous groups. On account of the peculiar appearance of their foliage it is not to be wondered at that they should have been associated by Kenault,2 Campbell,5 and Potonie 1X with lycopodineous an- cestors. Kecent work on the anatomy (Jeffrey19) of vascular plants in general appears to show that in the case of the Conife- rales the microphyllous habit has merely an ecological interest ; for, unlike all the Lycopodiales, they have well-marked foliar gaps in their cauline woody cylinder. The researches of Worsdell 8 on the foliar bundles of the Conifers have resulted in a clear demonstration of striking pteridophytic features. Fig. 112, X, represents a cross-section of the cotyledonary bundle of Cephalotaxus drupacea. On the lower side of the fibrovascular strand centrifugal wood, such as is ordinarily present in the bundles of the Conifers, can be made out. On the upper side of the bundle are large, thick- walled elements, which are to be compared with the centripetal tracheids of the cyeadean bundle in Fig. Ill, U. Fig. 112, Y, shows a longitudinal section of a cotyledonary bundle of 0. Fortunei. On the left are some pitted tracheids of the second- ary wood. In the center of the bundle is the disorganized pro- toxylem, while on the right is a single reticulated tracheid of the ancestral centripetal wood. The cotyledonary bundles of ( 'ephalotaxus are consequently mesarch like those of the ordi- nary leaves in Cycads, but show striking signs of degeneracy COMPARATIVE ANATOMY OF GYMNOSPERMS 309 in the centripetal cryptogamic wood. On the flanks of the bundle ihe centripetal wood is continuous with the short-pitted cells of the " transfusion tissue " discovered by Frank in 1864. In the bundles of the adult leaves of most of the living Con- iferales there are only the very slightest traces of centripetal wood. Worsdell has reached the interesting general conclusion that the " transfusion tissue which occurs almost universally in the leaves of gymnospermous plants as an auxiliary con- ducting system has been phylogenetically derived from the centripetally formed xylem of the vascular bundle." Fig. 112, Z, shows the topography of a cross-section of a branch of Thuja occidentaUs. The leaves in this species are extremely reduced, especially those occurring on the upper and lower sides of the flattened branches. It might naturally be expected that under these circumstances the foliar gaps would be obscure or absent, but such is not the case, for subtending the traces, which pass to the specially small leaves on the upper and lower sides of the flattened branch, are two distinct foliar lacunae. An examination of a large number of Conifers, some with a very considerable xerophytic reduction in the size of their leaves, has shown that the presence of foliar gaps is quite constant in the group (Jeffrey 19). It is now known that foliar gaps are unfailingly absent in the tubular central cylinder of living and fossil Lycopodiales and Equisetales, while on the other hand they are invariably present in the Filicales. Hence it may be assumed that the Coniferales, much as thev resemble the Lycopods in external appearance, are really derived from filiemean ancestors by adaptation to a xerophytic mode of life. Themicrophyllous habit is obviously a cenogenetic adaptation, for the structure of the fibrovascular skeleton plainly indicates that the coniferous stock is palingenetically megaphyllous, and thus allied to the Ferns. Fig. 112, A A, shows the structure of the root of Pinus Stratus., Tb.e cortex and phloem surround a considerable mass of secondary wood, in the center of which may be distinguished The exarch primary wood. This feature is more clearly seen in Fig. 112, BB, which represents the center of the section shown in Fig. 112, A A, more highly magnified. It is an in- teresting fact, to which Van Tieghem 4 has drawn attention, that the mode of growth of the primary wood is the same in all 310 MORPHOLOGY OF ANGIOSPERMS the vascular plants, viz., exarch and centripetal. The root of the Spennatophytes is consequently conservative, and retains intact ancestral pteridophytic features. It seems phylogenet- ically significant that the exarch type of wood, so typical of the Lycopods and their allies, is always present in roots, and never the mesarch type so characteristic of the Fern-alliance. This feature probably indicates that the Lycopod stock is an extremely old one, a conclusion borne out by the fact that the Lycopsid series had already culminated in the Carboniferous age. It appears also not improbable that the Pteropsida, large- leaved fern-like plants, took their origin from the microphvl- lous lycopodinean stock in remote antiquity, and still exhibit a trace of their origin in the primary structure of their roots. GXETALES This group is generally regarded as the highest of the Gym- nosperms, a view which is borne out both by a consideration of its anatomy and its reproductive organs. The latter show in the case of Tumboa and Gnetum a considerable advance toward the condition of true flowers, and this advance is paral- leled by a reduction in the amount of female prothallial tissue antecedent to fertilization. The Gnetales on the anatomical side show indubitable evidence of gymnospermous relationship, in the presence of quite typical foliar transfusion tissue. They are distinguished anatomically from all other Gynmosperms, however, living or fossil, bv the presence of rudimentary vessel.-. Fig. 113, CC, shows the structure of the wood in Gnetum Gnemon. The secondary wood in this species consists of tra- cheids and vessels, the latter being easily distinguished by their larger size. In some cases the fact that direct communication between two contiguous vessels is merely the result of the dis- appearance of the membrane of a bordered pit can be made out.* * For list of literature cited see end of Chapter XVII. CHAPTER XVII COMPARATIVE ANATOMY OF ANGIOSPERMS * The question of the relationship of the two great divisions of the Angiosperms has for many years been a matter of dis- pute. Anatomically the differences between the Dicotyledons and Monocotyledons are sufficiently well marked, but it has not been easy to decide from ordinary anatomical data which should be regarded as haying- the more primitive and antece- dent organization. There can be little doubt that the two groups are closely related, for in addition to the striking general re- semblance of their sporophytic tissues there is almost an identi- cal organization of the male and female gametophytes. The Monocotyledons have bv some been regarded as primitive on account of the absence of a cambium in their ordinarily closed bundles. This 'view has been strengthened bv statements as to their appearing earlier in the geological strata than the Dicot- yledons. It is now known bevond doubt, however, that manv of the earlier cryptogamous groups had well-marked secondary Growth, so that the absence of cambial activity is bv no means necessarily a primitive feature. Further, a more careful study of plant fossris has made it clear that many of the remains for- merly considered to be Monocotyledons are in reality Pterido- phytes or Gymnosperms. Discussion of these interesting prob- lems will be more profitable after the salient features of the anatomy and development of the Angiosperms have been de- scribed. DICOTYLEDONS It has been shown by Jeffrey 13 that the primitive condition of the central cylinder in the Angiosperms is siphonostelic. The tubular central cylinder of the seedling of Ranunculus, * Contributed by Professor Edward C. Jeffrey, of Harvard University. 21 311 312 MORPHOLOGY OP ANGIOSPERMS for example, is characterized by foliar gaps such as are found in the Filicales and Gymnosperms. Often in the seedling of this genus there is present an internal limiting layer of the stelar tissue which degenerates in the adult. Hence it may be assumed, in the absence of negative evidence, that the pith of Ranunculus belongs to the same morphological category as the cortex. Marie, from a coni])arative study of the anatomy of all the Ranunculaceae, has reached the conclusion that the genus Ranunculus is the starting-point from which all the other gen- era of the order have been derived. It follows apparently that the central cylinder of the Ranunculaceae in general is suscep- tible of the same interpretation as that of Ranunculus. If the central cylinder of the Ranunculaceae be siphonostelic with foliar gaps, i. e., phyllosiphonic, it may fairly be assumed that the central cylinder of Dicotvledons in general is to be similarly interpreted, especially as foliar gaps are universally present, even in such extreme cases of xerophytic reduction as Casuarlna and the Cactaceae. There are some instances of the occurrence of concentric bundles in the Dicotyledons, but they appear to be of ceno- genetic origin, and consequently of no phylogenetic importance ; for in the cases which have been investigated, the concentric condition is ordinarily absent in the seedling, the leaf-traces, and the reproductive axes. This feature is illustrated by Pri- mula farinosa, in which the bundles of the seedling, the repro- ductive axis, and the leaves are always collateral ; whereas those of the older vegetative stem are usually concentric. Similar phenomena have been observed in the jSTymphaeaceae, Halo- raghidaceae, etc. In the older subterranean stem of Ranunculus acfis the fibrovascular tube becomes broken up into a series of segments or bundles by the overlapping of the foliar gaps ; quite often in the stouter subterranean axis of Ranunculus acris (Jef- frey 13) the foliar bundles tend to run in the pith before passing out to the leaves, thus offering a striking feature of resemblance to the normal course of the leaf-traces in the Monocotyledons. In the aerial stem, however, this feature is not present, as may lie seen in Fig. 113, DD, in which the arrangement of the bundles shown is the typical one for the Dicotyledons. There are a good many exceptions to the rule, however, e. g., Podo- COMPARATIVE ANATOMY OF ANGIOSPERMS 313 phyllum, Gunnera, the Nyinphaeaceae, etc. In the last-men- tioned eases, the study of seedlings shows that the circular dis- position of the fibrovascular strands is primitive. In Podo- phyllum the scattering arrangement of the bundles is present only in the aerial stem, and is absent in the rhizome, as well as in the seedling. Fig. 113, EE, is from a photograph of one of the bundles of Ranunculus acris. The bundle is surrounded by a seleren- chvmatous sheath, which is thickest externally. The xylem and phloem are separated from one another by a narrow zone of cells arranged in radial rows, indicating that a slight but unmistakable cambial activity is present. The bundle is con- sequently an open one. The protoxylem is obviously the inner- most part of the primary xylem, so the bundle is endarch. En- darch fibrovascular strands with secondary growth bv means of a cambium are characteristic of the Dicotyledons. In aquatic Dicotyledons (e. g., the Xymphaeaceae), however, secondary growth is frequently absent. The trachea ry tissue of Dicotyledons with considerable sec- ondary growth shows a further division of labor over the highest Gymnosperms. In the oak, for example, there are thinner and thicker-walled tracheids as well as vessels. The latter have practically lost their water-conducting function and have very few extremely small pits in their walls. They have thus been differentiated for the purpose of support. In the beech this division of labor among the tracheids does not take place, for all the tracheids are of the same type and have well-developed 1 'ordered pits in their walls. Strasburger 3 is of the opinion that the woopl-fibers of the Cupuliferae throughout are modified tracheids, and hence merit the name of fiber-tracheids. Such fibers are present in a number of the dicotyledonous orders. In other cases, according to Strasburger, the wood-fibers are to be regarded as derived from wood-parenchyma and not from tracheids. In these instances they may properly be called libri- form filters. It is not clear, however, that a sharp distinction can always be drawn between the two sorts of wood-fibers. The sieve-tissue of the Dicotyledons is also more highly specialized than that of the Gymnosperms, for the sieve-tubes have special accessory cells. These accessory cells are derived from the same mother-cells as the sieve-tubes, and are known 314 ' MORPHOLOGY OF ANGIOSPERMS as companion cells. Companion cells are quite absent in the Gymnosperms, but Strasburger has pointed out that here the # H marginal cells of the medullary rays perform the physiological function of companion cells. \ The Dicotyledons as a group are distinguished anatomically from the Gymnosperms by the entire absence of palingenetic pteridophytic features of any sort in the fibrovascular tissues of their stems and leaves. The bundles are throughout endarch collateral, except in the root, where they are exarch, as in all other vascular plants. The concentric bundles which occasion- ally occur in the Dicotyledons are obviously cenogenetic, and have no phylogenetic significance. Both the xylem and phloem of the Dicotyledons show a marked advance in differentiation over all the Gymnosperms. The central cylinder of the stem in the Dicotyledons is characterized by the presence of foliar gaps, and accordingly, if the Dicotyledons are to be regarded as derived ultimately from pteridophytic ancestors, as appear- to be the case, their descent is apparently from the Filica]c>, either directly or through some living or extinct phylum of the Gymnosperms. The argument for descent from a gymnosper- mous ancestry seems to gain great force from the entire absence of pteridophytic features in the shoot or leaves of the dico- tyledonous Angiosperms. MONOCOTYLEDONS The arrangement of the bundles in the adult stem of the Monocotvledons is very characteristic. Instead of being dis- posed in a circle, as in the Dicotyledons, they are scattered throughout the central cylinder. Fig. 113, FF, illustrates this peculiarity. Not infrequently, however, e. g., in the Lilia- ceae, the bundles are obviously segments of a fibrovascular tube, just as is typically the ease in the Dicotyledons. Fig. 113, GG, shows this feature in the rhizome of Clintonia borealis. Sub- tending gaps between the bundles are to be seen smaller fibro- vascular strands, which are leaf-traces. In this example we have obviously to do with a fibrovascular tube with foliar gaps1. Interestingly enough, the tubular arrangement of the fibrovas- cular elements is frequently present in monocotyledonous seed- lings, although characteristically absent in the adult. Hence it may be inferred that the tubular central cylinder with foliar COMPARATIVE ANATOMY OF ANGIOSPERMS 315 gaps is the ancestral condition in the Monocotyledons. In some cases, e. g., Symplocarpus foetidus, the pith and cortex are continuous in the seedling through the foliar gaps, although they no longer appear to be so in the adult. An internal endo- dermis or stelar boundary is also sometimes present in the young plant, but has usually quite disappeared in the adult. The typical bundle of the Monocotyledons is amphivasal concentric. Such a bundle is shown in Fig. 113, HH. In this type of bundle the tracheary tissue surrounds the phloem, and not the phloem the tracheary tissue, as is generally the case in the Pteridophyta. The amphivasal concentric bundle is char- acteristic of the Monocotyledons from the grasses (Zizania, etc.) to the orchids (Ilabenaria, Cypripedium, etc.), and is quite as constant a feature as the scattering disposition of the tibrovascular strands. This type of bundle resembles the am- phicribral concentric bundles of the Pteridophytes in showing- no evidence of secondary growth. Amphivasal strands are ab- sent in the leaves and reproductive axes of the Monocotyledons, and generally in the seedlings as well. Unlike the concentric strands of the Gymnosperms, they are accordingly a cenogenetic and not an ancestral feature, but on account of their widespread occurrence in the group have an important phylogenetic signifi- cance. Secondary growth has been supposed to be entirely lacking in the collateral strands of the Monocotyledons, but Queva 12 has recently shown that characteristic secondary growth is present in the bundles of the tuberous base of the stem of the liliaceous genus Gloriosa. The activity of the cambium becomes apparent during the season after the formation of the tuber, when it is passing its reserve products into the aerial stem. From the occurrence of a cambium in Gloriosa, etc., Queva has drawn the conclusion that the Monocotyledons are derived from the lower Dicotyledons. The most salient anatomical features of the Monocotyledons are the scattering disposition of their closed tibrovascular strands, and the presence of amphivasal concentric bundles. These features, although practically universal, are not primi- tive; for a study of the leaves, reproductive axes, and seedlings shows often a dicotyledonous disposition of the generally col- lateral strands. Hence we may infer that the Monocotyledons 316 MORPHOLOGY OF ANGIOSPERMS are a strictly monophyletic and modern group, since they possess in common a very characteristic mode of arrangement of bun- dles of a unique type, and since neither the structure of the bundles nor their mode of disposition is palingenetic. Further, the evidence of secondary growth in Gloriosa, etc., would seem to indicate that the Monocotyledons have come off somewhere from the Dicotyledons, which they resemble so closely in their essential reproductive organs. This view of the matter seems strengthened by the greater reduction of the sporogenous tissue in the megasporangium of the Monocotyledons as compared with the lower Dicotvledons, and bv the entire absence of the probably primitive phenomenon of chalazogamy, which is so characteristic of the lower Dicotyledons. In the present state of our knowledge we are apparently justified in considering the Monocotyledons to be a modern, strictly monophyletic and specialized group, derived from the Dicotyledons or their parent stock, possibly by adaptation in the first instance to an amphibi- ous mode of life.* LITERATURE CITED 1. Mettentus, G. H. Beitrage zur Anatomie der Cycadeen. 1857. 2. Renault, B. Cours de Botanique Fossile. Paris. 1880-1884. 3. Strasburger. E. Histologische Beitrage. III. 1891. 4. Tieghem, P. VAX. Traite de Botanique. Pains. 1891. 5. Campbell. D. H. Mosses and Ferns. New York. 1895. 6. Williamson and Scott. Further Observations on the Organiza- tion of the Fossil Plants of the Coal-measures. Part. 3. Lygi- nodendron and Heterangium. Phil. Trans. Roy. Soc. London B. 186: 1896. 7. Scott. D. H. The Anatomical Characters presented by the Ped- uncle of the Cycadaceae. Annals of Botany 11: 399-419. pis. 20- 21. 1897. 8. Worsdell. W. C. On Transfusion Tissue ; its Origin and Func- tion in the Leaves of Gymnospermous Plants. Trans. Linn. Soc. London Bot. II. 5: 301-319. pis. 23-26. 1897. 9. Potonie. H. Metamorphose der Pflanzen im Lichte Palaeontolo- gischer Thatsacben. Berlin. 1S9S. 10. Jeffrey. E. C. The Development. Structure, and Affinities of the Genus Equisetum. Mem. Boston Soc. Nat. Hist. 5 : 155-190. pis. 26-30. 1899. * It should be noted that the manuscript of Chapters XVI and XVII was completed April 1, 19i>2. c Kt<^C-Tj Fig. 108.— .4, part of stem of Gleichenia flnhellata; £, stem of Adiantum palatum ; C, rhizome ol Ptn-is at/ nil inn : I>. _\ oung stem of same ; /;", stem of Selaginella laevigata ; I\ central cylinder of Osmunda Claytoniaua. .. -1F2»e$rs Fiq. 109. — \ adventitious root: oc, outer cortex; pet, petiole. Pte 0 Fig. 110. — M, diagrammatic transverse section of stem of Medullosa unglica. after Scott: st, concentric strands; pd, periderm; It, leaf trace; .Y. diagram of pari of transverse section of stem of Medullosa Solum, after Weber and Stebzel, from Potonie: pla, pU, larger concentric strands; *2, smaller concentric strands; 0, transverse >> of stem of Lyginodendron Oldhamium, after Williamson and Scott; /'. part of woody zone of same; \>. same, showing internal secondary wood and internal phloem. Fki. 111. — R. stem of Zamin florid 'ana : S, central cylinder of same; 7'. cortical foliar 1 mill lie of Offcas n valuta : J', petiolar bundle of same : I". section of part of leaf of a species of Cordaites; II', cotyledonary buudle of Ginkgo biloba: px, protoxylem : x1, centripetal wood ; x», centrifugal wood. " I .4.1 BB Fig. 112. — JT, cotyledonary bundle of Cephalotaxus drupacea: px, protoxyleni ; c1, c tripetal wood ; x*, centrifugal wood ; tf, transfusion tissue; F, longitudinal section of cotyledonary bundle of Cephalotaxus Fortutu i : ph. phloem ; other lettering as in .Y: '/.. -~ i j i i*. 1 1 branch of Th dentalis\ .t.t. Pinus Strobns; Jill, part <>f .nvanthes Candida, 170. 213. Gonystylaceae, 240. Goodeniaceae, 259, 270. Goodyera. 193, 104. Gramineae. 57. 63. 77, 98, 104. 109, 112. 113. 157. 174, 205, 230, 265, 275. Gray, A.. 8. Grebel, Dr., 213. Grubbiaceae, 243. Guignard, L., 30, 33, 38, 39, 59, 60, 61, 62, 63. 64. 65. 71, 77, 80, 81, 82, 84, 85. m. 87. 89, 90. '.»4. 95, 96, 97, 98, 99. 101. 104. 105. 122. 133, 136, 147, 151, 153, 154, 155. 156, 157, 158, 159. 165, 169, 170, 172. 178, 180, 202, 203, 204, 216, 217. 221. Gunnera. 89. 90, 166, 313. Guttiferae, 249. Gymnadenia. 77. 92, 94. 95: conop- *sea, 64. 82, 148, 19',. 217, 221. Gymnosperms. comparative anato- my of, 296; contrasted with An- giosperms. 1 : embryogeny of, 2 : gametophyte of, 3: sporophyte of. 2. Gynoecium. 24. Gynostemium. 238. Habenaria. 315: blephariglottis, 195; tridentata. 195. Haemodoraceae, 236, 264, 266. Hall, J. G.. 63. 77. 92. 95. 146, 167, 171. 175. 192. 215. 216. 222. Hallier. H., 2^2. Haloraghidaceae, 250. 312. Haisted, B. D„ 136. Hamamelidaceae. 246. Hamamelis. 30. 41 : virginiana, 147. Hanausek. T. F.. 221. Hanstein. J.. 188, 196. 198. Hartig. Theodore. 145. Hartog, M., 288. Haustoria. 104. 109. 202. Hautsehicht, '.15. Hebenstreitia, 177. Heckeria. 79, 90, 101. 167, 170. 178, 170. 201. Hedysannn coronarium. 203. Hegelmaier, F.. 102. 178, 192, 206, 207, 217. 218, 221. Heleocharis palustris, 128. Heliantbemum, 61. 122. Helianthus annuus, t55, 156. Helieonia, 171. Helleborus, 64, 84; cupreus. 60; foet- idus. 82. 156. Helobiales, 171. 220. 231. 234. 263. 275. 287. Helosis. 70. 95, 103, l»it;: guayanen- ~i~. 91. 92, 218. 340 MORPHOLOGY OP ANGIOSPERMS Hemerocallis, 04, 76, 77, 104, 135; fulva, 33, 74, 125, 126, ti9. Hepatica, 30, 38, 53, 94, 99, 100. Hernandiaeeae, 245. Hesperis, 136. Heterangium, 300, 301, 302; Grievii, 300, Fig. 109. Hibiscus, 156. Hicoria, 148. Hill. T. G., 99, 192. Himantoglossura, 156; hircinum, 82. Hippeastrum aulicum, 148. Hippocastanaceae, 248. Hippocrateaceae, 248. Hippuris, 55, 64. Hofmeister, W., 18, 32, 47, 48, 49, 51. 53, 71, 94, 101, 106, 125, 132, 143, 146, 147, 148, 176, 178, 181, 206, 221, 222. Holferty, G. M., 63, 76, 77, 78, 96, 176, 192. Holm, Theodore, 282. Homalomena, 263. Hooker, J. D., 227. Houstonia, 55, 202. D'Hubert, E., 79, 108, 147, 151. Humiriaceae, 247. Humphrey, J. E., 64, 77, 104, 154, 171. 173, 192. Hyaeinthus orientalis, 74, 75. Hydnoraceae, 244. Hydrocaryaeeae, 250. Hydroeharitaeeae, 157, 171, 229, 230, 263, 265, 275. Hydrophyllaceae, 176, 256, 269. Hydrostachyaeeae, 246. Hypericum, 24; calycinum, 18. Hypophysis, 188, 198. Icacinaceae, 248. Ikeda, T., 77, 96, 99, 104, 111, 112, 153, 157, 158, 174. Impatiens, 131, 205. Integument, 53. Intine, 131. Iridaceae, 64, 236, 264, 265, 276. Iris, 77, 99, 155; sibirica, 217, 221; squalens, 81; stylosa, 64. Irmisch. T.. 206. Isobilaterality, 16. Isoetaceae, 285. Isoetes, 196, 284, 285, 287. Jasminum, 95. Jeffersonia, 64, 76, 84; diphylla, 101. Jeffrey, E. C, 214, 215, 222, 281, 296, 297, 298, 300, 301, 303, 304, 305, 308, 309, 311, 312. Johnson, D. S., 79, 89, 90, 101, 104. 105, 136, 137, 153. 166, 167, 168,. 170, 176, 178, 179, 200, 201, 242. Johnson, T., 47, 55. Jonsson, B., 221. Jordan, K. F., 293. Juel, H. O., 73, 74, 76, 80, 82, 92, 101, 124, 126, 128, 129, 147, 166r 170, 201, 211. Juglandaceae, 46, 157, 243, 278, 284. Juglandales, 243, 268. Juglans, 91, 146, 147, 148, 150, 156r 157. 158; cinerea, 149; cordifor- mis, 60, 79, 84, 87; nigra, 92, 96; regia, 90, 149. Juncaceae, 236, 264, 265, 276. Juneaginaceae, 196, 229, 230, 263, 265, 275. Juncagineae, 171. Juncus, 121. Jurassic Monocotyledons, 273. Justicia, 131. Kamienski. F., 206. Karsten, G., 60, 79, 84, 87, 91, 92r 96, 157, 158, 284. Kauffmann, N., 28. Kerner. A., 42. Klebs, G., 288, 289. Koch, L., 80, 206. Koeberliniaceae, 249. Kolliker, A., 292. Kornicke. F., 180. Koernicke. M., 63, 81, 137. Korschinsky, S., 292. Labiatae. 16, 24, 104, 106, 176, 256, 269, 271. Labiatales, 258. Lacistemaceae, 242. Lactoridaceae, 245. Land, W. J. G.. 29. 82, 151, 155, 156, 160, 169. Figs. 35 and 36. Lang, F. X., 107, 108. Lang. W. H., 288, 289. Lappa. 122. Lardizabalaceae. 245, 267. Larix europaea, 15%. INDEX 341 Lathyrus, 136; heterophyllus, 204; odoratus, 204. Lauraeeae, 245, 207. Laurus, 277. Lawson, A. A., 129. Leaves. Monocotyledons and Dicot- yledons contrasted, 5, 6. Leavitt. R. G., 193, 194. Lecythidaeeae, 250. Leeuwenhoek, A., 213. Leguminosae, 15, 16, 20, 55, 65, 97, 174, 202, 246, 267, 279; embryos of, 202. Leguminosites, 277. Leitneriaceae, 242. Leitneriales, 242. Lemna. 10. 30, 39, 63, 77, 92, 95, 103.136.167,193; reduced flowers, 10; minor, 37, 40. Lemnaceae. 233, 234, 263, 265. Le Monnier, G., 181, 182. Lennoaceae. 253. Lentibulariaceae, 256. Lepidium, 157. Leptosiphon, 103. Leucojum vernum, 81. Lilaea, 28, 46, 99; subulata, 27, 47, 196, 285. Liliaceae, 64. 76, 82, 97, 103, 109, 157, 174, 193, 209, 236, 264, 265, 274, 276. Liliales, 236, 264, 276. Liliiflorae, 236. Lilium, 25, 41, 58, 64, 73, 77, 80, 81, 84, 95, 97, 104, 123. 134, 135, 136, 137, 146, 151, 157, 159, 161, 169, 193, 195; type of embryo, 193: auratum, 134, 138; candi- dum, 61, 81, 86, 130, 131, 153, 154; croceum, 81; Martagon. 81. 122, 130. 154, 156, 158, 221; philadel- phicum, 29, 54, 61, 81, Figs. 35 and 36. 88. 135, 157, 160, 193: pyrenaicum. 156; tigrinum, 16, 81, 134, 135, 157. Limnanthaceae. 248. Limnocharis, 63. 77, 92, 95. 167, 171, 175. 176. 192; emarginata, 146, 215. 216. 222. Liliaceae, 247. Linum. 103. Liriodendron, 277, 27S. Listera, 194; ovata, 82, 193, 194. Lloyd, F. E., 55, 58, 61, 80, 82, 85, 86, 97, 101, 102, 104, 108, 202. Loasaceae, 176, 249. Lobelia, 80, 103, 111. Lobeliaceae, 24, 30, 48, 58, 106, 110. Loganiaceae, 255, 269. Longo, B., 150. Lonicera, 80; coerulea, 125. Loranthaceae, 55, 65, 97, 104, 110, 176, 243. Loranthus, 50, 61, 85, 86, 91, 92, 97, 177; europaeus, 221, pentan- drus, 49, 200; sphaerocarpus, 48, 50, 199, 200. Lotsy, J. P.. 28, 34. 48, 49, 50, 51, 79, 92, 136, 166, 218. Luerssen, C, 131. Lupinus, 202, 204; luteus, 205; mu- tabilis, 205: pilosus, 205; poly- phyllus, 205: subcarnosus, 205; truncatus, 205. Lychnis, 21. Lycium, 80. Lycopodiales, 286. Lyginodendron, 301, 302, 303, 305, "306; Oldhamium, 301, 302, Fig. 110; robustum, 302. Lygodium, 300. Lyon, F. M., 49, 74, 94, 100. 126, 129, 136, 151. Lyon. H. L., 169, 201, 207, 208, 282. Lysichiton, 63, 192; kamtschatcense, 98, 192. Lythraceae, 104, 110, 250. Magnolia, 277. Magnoliaeeae, 245. Magnus, P.. 28. Mahonia indica, 64. Maize, xenia, 180. Male cells, 136: not concerned in fertilization, 161. Male gametophyte, 121. Male nucleus, *136, 152, 157, 166; change in size and form. 152; fusion. 153: its part in fertiliza- tion, 160: movements of, 157; ver- miform. 161. Ma lesherbiaceae, 249. Malpighi, M., 143. Malpighiaceae, 247. Malva. 38. Malvaceae, 33, 131, 157, 249. 342 MORPHOLOGY OF AXGIOSPERMS Malvales. 249. 267. Mangifera indica, 221. Marantaceae, 171, 237. 264, 266. Marattiaceae, 301. Marcgraviaceae, 249. Marie, M., 312. Martyniaceae. 256. Massula, 39. Mayacaceae, 235. 264, 266. Medicago, 104; falcata, 204. Medinilla, 1,2. Medullosa, 301, 302: anglica, 301, Fig. 110; Solmsi, 301, Fig. 110; stellata, 301. Megasporangium, 46; archesporium of, 57: cauline, 46; mother-cell, 66; parietal cells, 62; time of de- velopment, 52. Megaspore. 71: germination of, 87; number of, 76; the functional, 84. Melastomataceae. 250, 268. Meliaceae. 247. Melianthaceae, 248. Melissa officinalis. J(2. Mellink, J. F. A.. 71. 84. Menispermaceae, 245. Menispermites, 277. Mentha, 38: aquatica, 32. 33. Menyanthes. 103: trifoliata. 32. Merrell, W. D.. 34. 35, 82, 101, 103, 136. 137. 151. 158. 199. Mertensia. 136. Mesembrianthemum. 63. Metamorphosis, 8, 10. 22. Mettenius, G. H., 304, 305. Microspermae, 238. Microsporangium. 27; archespori- um, 32: cauline, 28: development of, 32: mother-cells. 38; number of, 29: parietal layers, 34: tape- tum, 36; time of formation. 30. Microspores. 121; germination of, 132; number of, 125: wall of, 131. Mimosa, 203, 247. 267. 268, 279; Denhartii. 216. 221. Mimoseae. 30, 33, 132. Mirbel. C. F., 56. Mohl, H. von. 145. Mnnimiaceae. 245. Monocotyledons, 4, 11; anatomy of, 314: classification of. 227; cyclic number of, 5; embryo of, 4. 7, 188; fossil. 272; in Carboniferous, 273; in Cretaceous, 273; in Juras- sic, 273; in Tertiary, 275; geo- graphic distribution of, 262 ; leaves of, 5, 6; phylogeny of. 281; pro- phyllum of, 7 ; roots of, 7 ; seed germination, 6; vascular bundles of, 4. Monotropa, 148, 158, 206; Hypopi- tys, 11,5, 156; uniflora, 96* 102, 147, 153, 157, 159, 167. Monotropaceae, 176. Moraceae, 243, 278. Moringaceae, 246. Morus albus, 221. Mother-cell, of megasporangia, 38; of microsporangia. 38. Mottier. D. M., 52, 60, 61, 62, 76, 77. 78, 82, 84, 87, 94. 99, 101, 103, 124, 129, 130, 134, 136, 146, 153, 154. 169, 199. Murbeck. S., 55, 58, 59, 79. 82, 87, . 93, 96, 104, 150. 175, 196, 211, 212, 218. 219, 221, 285. Musaceae. 171. 237, 264, 266, 276. Muscari neglectum, 81. Mutation theory. 292. Myoporaeeae. 256. Myoporum. 103. 200; serratum, 201. Myosurus, 64. 99. Myrica. 277. Myricaceae. 242. Myricales, 242. Myristica, 53. Myristicaceae, 245. Myrothamnaceae. 246. Myrsinaceae. 254. 269. Myrsinophyllum, 277. Myrtaceae, 201, 250, 268. Myrtales, 250. Myzodendraeeae, 243. Myzodendron. 105, 110; punctula- tum, 47, 55. Nageli. C, 32. Xaiadaceae, 97, 157, 171, 229, 230, 263, 265. Naias. 28, 41. 46. 133. 171. 192; flexilis, 21: major, 81, 157, 165, 170, 216, 221. Narcissus. 77. 99. 156. Xawaschin, S.. 90, 146, 148, 149, 150, 155. 156, 180. INDEX 343 Xelumbo, 1G9. 201, 207, 208. Nemec, B., 74, 75. Nemophila, 130, 176. Xeottia, 131, 133; nidus-avis, 82, 122; ovata, 88, 39. Xepenthaceae, 246, 268. Xicotiana, 80. 96, 97. 157, 158; Taba- cum. 136. 147, 151, 158. Xigella, 99, 151, 158; daniascena, 157, 159; sativa, 156. Xolanaceae, 256, 269. Xothoseordon fragrans, 213, 221. Nuphar, 50, 176; lutea, 208. X\ ctaginaceae. 96, 97, 244. Xyctandra. J2. Nymphaea, 9, 22. 23, 50, 53, 176, 201, 207: alba. 82. Nymphaeaceae, 103, 110, 176, 245, 282, 312, 313. Obolaria. 50. Oehnaceae, 249. Oenothera. 104; Lamarckiana, 292; lata. 293. Olaeaceae. 243. Oleaeeae, 97, 255. Oliniaceae, 250. Oliver, F. W., 55, 80, 85, 95. 106, 109. Ill, 134, 148, 169, 177. 178, 199. Onagraeeae. 30, 97, 131, 250, 267. Onobrychis petraea, 204. Ononis, alopecuroides, 204; fruti- cosa, 204. Opiliaceae, 243. Opuntia. tortispina, 214: vulgaris, ■214. 221. Opuntiales, 250. Orange. 213. Orchid, 195: type of embryo, 194. Orchidaceae, 15, 30, 58, 64, 97. 103, 113. 132. 136, 147, 157. 171, 194, 2i hi. 234, 238. 266. Orchidales, 238, 264. 276. Orchis, 51, 77. 145. 156: latifolia, 1',',. 145; maculata, 33. 88, 39. I',',: mascula, 82. 121. 126; Morio, l','i. 145. 221: pallens, 64. Organogeny of flower. 16. Ornithogalum, 64, 97. 99 : nutans, 91 : pyrenaicum, 61. Orobanchaceae, 176, 206, 256, 269. Orobanche, 80. 23 Orobus angustifolius, 65. 20 ) : au- reus, 21)4. Osmunda, 302, 303: einnamomea, 298, 299, Fig. 109; ( laytoniana, 298, Fig. 108; regalis, 299. Osterwalder, A., 99, 100, 111, 221. Osyris. 105. Ovary, 24. 26. Overton, E., 71. 221. Overton. J. B., 63, 64, 81, 82. 94. 100, 170. 199, 212. Ovulary, 24. Ovules, foliar. 50: morphological nature of, 51; development of, 53; forms of, 56. Oxalidaceae, 247. Paeonia speetabilis, 82. Palet, 231. Palmaeeae, 231, 262, 266, 274, 275. Palmales, 231, 262, 275. Pandanaceae, 228, 262, 266, 273, 275. Pandanales, 228, 231. 262. 275. • Papaver, 136: orientale, 65. Papaveraeeae. 65. 246. Papilio. 247. 267. Parietales. 249. Paris quadrifolia. 159, 160. Parthenogenesis, 210. Passiflora, 131. Passifloraceae, 249. Payer, J. B., 16, 20. Pechoutre, F.. 59, 199. Pedaliaceae, 97, 106, 110, 176, 177, 256, 269. Pedicularis, 106. Tenaeaceae. 250. Pentaphyllaceae. 248. Peperom'ia. 79. 88, 90, 136, 137, 153, 178. 179. 200; pellucida, 89, 160, 168, 178, 200. 242. 2S4. Pepo macrocarpus, 143. Perigyny. 13. 14. Perisperm, 103: function of, 179. Peristylis grandis, 1!'4. Personalea, 15. 24. 258. Petasites, 101. Petit-Thouars, 213. Peucedanites, 277. Pfeffer, W.. in. Pfitzer. E., 1!>4. Phajus, 156. Phalaenopsis grandiflora, 194. 344 MORPHOLOGY OF ANGIOSPERMS Phaseolus, 179, 208; rnultiflorus, 204. Philydraceae, 235, 204. Phlox Drummondii, 113. 1'hyllocactus, 108. Phragmites, 275. Phrymaeeae, 256. Phylloglossum, 300. Phyllosiphonic, 298. Phylogeny of Angiosperms, 280. Phytelephas, 178, 231, 262. Phytolacca, 179. Phytolaccaceae, 103, 179, 244. Pinguicula vulgaris, )2. Pinus, 160; Strobus, 112, 309. Piper, 79, 90, 167, 178; medium, 16S. Piperaeeae, 46, 56, 79, 103, 178, 179, 201, 242. Piperales, 242, 287. Pirolaeeae, 253. Pirotta, R., 150. Pirus Malus, 15, 221. Pistia, 178, 192, 195, 201, 263, 275; type of embryo, 192. Pistil, 25. Pi sum sativum, 204. Pittosporaeeae, 246. Placenta, 25. Plantaginaceae, 102, 106, 176, 258, 271. Plantaginales, 258, 269. Plantago, 269; lanceolata, 107. Platanaeeae, 246. Plumbaginaceae, 254. Poaeites, 275. Podophyllum. 53, 282, 313; pelta- tum. 31, 82. 12',. Podostemonaceae, 246. Polar nuclei, 92: fusion of, 95. Polemoniaceae. 103, 256, 269. Polemoniales, 258. Pollen mother-cell, division of, 126 Pollen-tube, 143; branching of, 148; development of. 146: discharge of, 152: entrance into sac, 151: in cleistogamous flowers, 146; Prop- fen. 148: time between pollination and fertilization, 146. Pollination, relation to endosperm, 169. Pollinium. 132. Polyembryony, 213. Polygalaceae," 104, 110, 247. Polygonaceae, 46, 56, 179, 244, 267. Polygonales, 244. Polygonum, U4; divaricatum, 94. Polypompholyx, 108. Polystelic, 2i)7. Pontederia, 104, 146, 151; cordata, 81. Pontederiaceae, 34, 37, 63, 77, 78, 97, 235, 264, 265. Populus, 52, 133, 277, 278; monilif- era, 30, 31; primaeva, 276; trem- uloides, 60. Portulaca, 143. Portulacaceae, 244. Potamogeton, 63, 76, 77, 96, 97, 104, 136, 176, 192; natans, 78; folio- sus, 33, 62, 78. Potamogetonaceae, 229, 230, 234, 263, 265, 274, 275. Potentilla, 18. Pothos longifolia, 148. Potonie, H., 300, 303, 308, Fig. 110. Prantl, K., 8, 56. Primula farinosa, 312. Primulaceae, 19, 103, 254, 269. Primulales. 254, 269. Principes, 231. Pringsheim. X., 288. Proangiosperms, 277, 281, 283, 286. Proembryo, 188. Propfen,*148. Prophyllum. Monocotyledons and Di- cotyledons contrasted. 7. Prote'aceae, 131, 243, 268, 278. Proteales. 243. Proteophyllum, 277. Protocorm. 209. Protolemna, 275. Protostelie, 297. Prunus Cerasus, 125. Pseudo-monocotyledons. 206. pscudo-polyembryony, 221. Psilotum triquetrum, 15^. Pteridophytes. anatomy of, 296. Pteris aquilina, 297, 298, 301, 303, Fig. 108. Punicacoae. 250. Purkinje. J. K., 34. Pyrethrum, 85, 87; balsaminatum, "(11. Pyrola rotundifolia, Jf2; secunda, 206; uniflora, Jf2. Pyndaceae, 176. INDEX 345 Quercus. 34, 66, 79, 94, 147, 148, 208; Rolmr, 147; velutina, 31, 58, 60, 147. Queva, C. 281, 315. Quiinaceae, 249. Rafflesiaceae, 244. Ramondia pyrenaica, -}2. Ranales. 245, 287. Ranunculaceae. 21, 60, 64, 78, 84, 99, 102, 111, 153, 157, 158, 169, 245, 267, 282, 312. Ranunculus. 11, 16, 36. 37, 51, 55, 64, 78, 87, 100, 131, 136, 151, 158, 170, 199. 311, 312: flowers of, 11, 16; abortivus. 00: acris, 312, 313, Fig. 11.1; Cymbalaria. 157; Ficaria, 206. 207. 282; Flammula. 150: niultifidus, 65, 88; septentrionalis, 61. Rapateaeeae, 235, 264. Ray. John, 227. Reichenbach, H. G., 132. Renault, B., 308. Reseda. 20. 156; odorata, 173. Resedaeeae, 57. 157, 246. Restiaeeae, 56. Restionaceaej 235, 264, 276. Reversion, 22. Rhamnaeeae, 249. Rhamnales, 249. Rhinanthus. 106. Rhizophoraceae, 250. Rhododendron. 132. Rhoedales, 246. Rhopaloenemis, 79. 92, 136; phal- loides, 28, 34, 49, 51. Ricinus. 24, 179. Riddle, Lumina C, 63, 65, 199. Rohinia, 147. Rohrbaeh, P., 28. Romulea, 94, 99. 104. Root, Monocotyledons and Dicoty- ledons contrasted. 7. Rosa. 18, 84, 87, 221, 247: livida. 58, 221. Rosaceae, 59, 60, 62, 63, 87. 199, 246, 267, 279. Rosales, 240. Rosanoff, S., 33, 132. Rose, J. X., 1 :'..->. 130. Rosenberg, O., 36, 37, 74. 77, 81, 122, 124. Rubiaceae, IS, 58, 61, 80. 97, 102, 111. 113. 2D2. 259, 269. Rubiales, 259, 209. Rubus. IS, 59. Rudbeckia speciosa, 156. Rumex, 21: Patientia, 125. Ruppia, 196, 285: rostellata, 175. Ruta, 97: graveolens. 62. Rutaceae, 20, 247. Sabiaceae, 248. Sachs, J., 15. Sagittaria, 90, 104, 135, 136, 137, 169, 175, 176; variabilis, 152. 154, 1"<5, 188, 189, 101. Salicaceae, 97, 242. Salicales, 242. Salix. 52. 60. 87. 94. 136. 151. 199, 277. 278: glaucophylla. 30, 58, 79, 95; petiolaris, 28, 57. 95. Salvadoraceae. 255. 209. Salvia, 95. 97 : pratensis, 85. Sambucus, 130. Sanguisorba, 58. Santalaceae, 55. 105. 110. 176. 243. Santalales. 243. Santalum, 91. 94. 105: album. 221. Sapindaceae, 20. 248. Sapindales, 248. Sarcodes, 25. 80. 97. 134. 169, 178, 199: sanguinea. 148. Sargant, Ethel. 73. 81, 82, 157, 182, 207. 209. 281, 282. Sarraceniaceae. 240, 268. Sarraceniales. 240. Sassafras, 41, 277. Saururaceae, 97. 104. 109. 176, 242. Saururus. 79. 104, 110, 170. 179: eemuus, 104, 105. Saxifraga caespitosa. 125. Saxifragaeeae, 59. 97. 240, 2G7. Scaevola, 103. Schacht, H.. 55. 94. 131. 143. 144. 145. 221. Sclia finer. J. H.. 28, 38, 53. 03. 74. 77. SI. SS. 90. 121. 120. 133. 135. 136, 137. 13S. 140. 151. 152. 153. 154. 109. 175. 188, ISO. 191. 193, 215. Schleiden, M. J.. 9. 52, :>5. 144. 145. Schlotterbeck, M., loo. Schmid, I'... 206. Schnegg, II.. S9. 90, LC6, 340 MORPHOLOGY OF ANGIOSPERMS Schniewind-Thies, J., 77, 81, 84. Schrankia uncinata, 221. Schwere, S., 102, 199, 210, 221. Scilla, 04, 84, 150; non-scripta, 81; sibirica, 81. Scitaminales, 237, 204, 2/0. Seitamineae, 57, 04, 77, 97, 103, 104, 109, 171, 192, 237. Scrophularia nodosa, 10, 125. Serophulariaceae, 90, 97, 103, 106, 110, 170. 250, 209, 271. Scleranthus animus, 125. Scott, D. H., 288, 300, 301, 302, 303, 304, 305, 306, 307, Fig. 110. Scytopetalaceae, 249. Sedum, 51. Seed, Monocotyledons and Dicotyle- dons contrasted, 6. Selaginaceae, 176, 177. Selaginella, 285, 287; laegivata, Fig. 108. Senecio, 87, 101, 169, 199. Seward. A. C, 273, 302; aureus, 61. Sherardia arvensis, 101. Shibata, K., 96, 102, 147, 148, 153, 157. 159, 167. Sboemaker, D. N., 30, 147. Sibbaldia procumbens, J/2. Silene, 94. Silphium, 34, 101, 103, 136, 137, 151, 156, 158, 160, 199; integrifolium, 35, 82; laciniatum, 82, 155. Simarubaceae, 247. Sinningia Lindleyana, 221. Siphonostelic, 297. Sium, 65, 94, 90, 103, 199; cicutae- foliuni, 79. Sisyrinchram, 77; 'ridifolium, 64. Sniilax, 274; herbacea, Fig. 113. Smith, Amelia C, 80, 170, 206. Smith, Anna, 30. Smith, R. W., 34, 37, 63, 73. 77, 78, 80, 81, 94, 95, 135, 136, 146, 151, 170. Snow, Laetitia M., 160. Solanaceae, 136, 157, 176, 256, 269. Solanum, 41 ; Lycopersicum, J/2. Solms-Laubach, H., 196, 206, 293. Sonneratiaeeae, 250. Sparganiaceae. 98, 228, 262, 265, 275. Sparganium. 112. 133, 192, 228, 229, 233; simplex, 47, 98, 135. Spartium junceum, 203. Spatlie, 232. Spatliirlorae, 233. Spergularia rubra, 46. S]ierniacoceae, 202. Spermatozoids, 136, 160. Sperms. 136, 160. Spiral series, 11, 228. Spiranthes, 193. Sporangia, foliar and cauline, 27,46; in winter, 30; periblem origin of, 27, 46. Sporophyte, 41; Angiosperms and Gymnosperms contrasted, 2. Stachyuraceae, 249. Stackhousiaceae, 248. Stamen, 23; morphology of, 22. Staminodia, 24. Stangeria paradoxa, 305. Staphylea, 136, 145; pinnata, 172. Staphyleaceae, 248, 278. Stellaria. glauca, 125; Holostea, 84. Stemona, 266. Stemonaceae, 236, 264, 266. Sterculiaceae, 249. Sterzel, J. T., Fig. 110. Stevens. W. C., 124. Stichneuron, 266. Stigma, 25. Strasburger, E., 38, 43, 53, 58, 62, 63r 64, 71, 73. 74, 76, 77, 81, 82, 83, 87,. 92, 94, 99, 104, 121, 122, 123, 124, 126, 128, 132, 133, 136, 138, 145, 146, 148, 154. 157, 158, 159, 171, 172, 173, 177, 181. 182, 183, 201,. 202, 208, 213, 214, 215, 217, 221,. 284, 294, 313. Strelitzia, 171. Strobilus, theory of, 288. Succisa pratensis, 293. Suspensor, 113, 190, 192, 193, 194,. 202. Stylidaceae, 103, 106, 108, 110, 170. Stylidium squamellosum, 107, 113. Styracaceae, 254. Sympetalae, 97 ; classification of,. 252; geographic distribution of,. 268. Sympetaly, 13. Symphytum officinale, 125. Symplocaceae. 254. Symplocarpus. 31, 37, 136; foetidus,. 315. Synanthae, 232. INDEX 347 Synanthales, 232, 263, 275. Synapsis, 126. Synearpy, 13. Synergids, 91, 94; as an haustorium, 111; disorganization of, 151. Syringa, persica, 125; vulgaris, 125. Taecaceae, 236, 264, 266. Tamaricaceae, 249. Tangl, E., 125. Tapetum, 36. Taraxacum, 101, 102, 157, 199; offi- cinale, 216, 221. Tertiary. Dicotyledons, 278; Mono- cotyledons, 275. Tetrads. 71. 121, 126. Tetragonolobus pnrpureus, 203. Thalia dealbata, 171. Thalictrum, 63, 64, 78, 94, 199; dio- icum, 100; Fendleri, 212; purpu- rascens, 100, 170, 212. Theaceae. 249. Theobronra Cacao, ^2. Thesium, 61, 105. Thomas, Ethel M., 137, 152, 156, 157, 158. Thuja occidentalis, 309, Fig. 112. Thunbergia. 131. Thymelaeaceae, 250, 268. Tiliaceae, 249. Tischler, G., 172. Todea barbara, 299. Torenia, 111, 136; asiatica, 104, 106. Tovariaceae, 246. Tozzia alpina. '/.?. Tracheid-like cells in nucellus, 100, 109. Tradescantia. 81, 135, 136; virginica, 63. Trapa, 171, 205; natans, 206. Trapella, 55, 80, 85, 95, 106, 110, 111, 177. 199; sinensis, 85. Tremandiaceae, 247. Tretjakow, S.. 217, 218, 221. Trenb. M., 49. 50. 59, 61, 64, 66, 71, 79. 80, 84, 85. 87. 91, 92, 149. 166, 167. 170. 193. 194, 199, 200, 201, 212. 213, 218. 221. Treviranus. 213. Tricyrtis, 64. 77. 96. 99. 104, 158, 174: hirta. 77. 111. 111. 153. 157. Trifolium, pratense, 221; resupina- tum, 203. Triglochin, 63, 192; maritima, 99. Trigoniaeeae, 247. Trillium, 30, 64, 77, 86, 89; grandi- florum, 81, 90, 159; recurvatum, 52, 7?, 81. Triple fusion, 158, 160, 166; nature of, 182. Triplochitonaceae, 249. Triticum, 63, 136, 137; vulgare, 81. Tritonia, 77. Triuridaceae, 229, 263, 266. Triuridales, 229. Tiwhodendraceae, 245. Tropaeolaceae, 247. Tropaeolum, 39. 171, 207. Trophophylls, 282. Tsehirch, A., 106. Tschistiakoff, I., 125. Tube nucleus, 133; fragmentation of, 135. Tubiflorae, 256. Tubiflorales, 256, 269. Tulasne, L. R., 106. Tulipa, 77, 89, 156, 193; Celsiana, 156; Gesneriana, 81, 215, 219, 222; sylvestris, 90, 156. Tumboa, 310. Turneraceae, 249. Tussilago, 101. Typha, 28, 38, 63. 74, 77, 104, 121, 131, 133, 229, 233; latifolia, 28, 126. Typhaceae, 97, 228, 262, 265, 275. Ulmaceae, 243. Ulmus, 147, 148, 150, 151; montana, 150; pedunculata, 150. Umbellales, 251. Umbelliferae, 15, 16, 55, 65, 79, 97, 251, 267. Umbelliflorae. 251. Unger, D. F., 55. Urticaceae, 56, 243. Urticales, 243. Utricularia. 206. Utriculariaceae, 106. Uvularia, 84. Vacciniaceae, 176, 177. Vaccinium. 80; Oxycoccus, J/2; ulig- inosum, yi. Vaillantia, 104, 111; hispida, 102, 202. 348 MORPHOLOGY OF ANGIOSPERMS Valerianaceae, 18, 259, 269. Van Tieghem, Ph., 30, 49, 52, 92, 297, 298, 309. Vascular bundles, Monocotyledons and Dicotyledons contrasted, 4. Vegetative, apogamy, 210; fertiliza- tion, 182; nucleus, 132. Velloziaceae, 236, 264, 266. Verbenaceae, 80, 176, 177, 256, 269. Verticillatae, 242. Vesque, J., 63, 65, 71, 77, 80, 84, 85, 86. Viburnum, 277, 27S. Yicia narbonnensis, 204. Yinca, 136. Yincetoxicum. medium, 221; nig- rum, 217, 221. Viola, 25. Yiolaceae, 249. Viscum, 61, 97, 176, 206; album, 221; articulatum, 79, 87. Yitaceae, 249. Viticella, 156. Yochysiaceae, 247. Ward, H. Marshall, 58, 61, 63, 71, 76, 77, 80, 87, 92, 94, 148, 194. Warming, E., 28, 32, 33, 51, 52, 53, 71. Webb, J. E., 19, 37, 39, 58, 59, 87, 108. Webber, H. J., 180, 181. Weber, M., Fig. Hi). Westermaier, M., 98, 99, 111. Wiegand, K. M., 33, 62, 63, 64, 7, 77. 78. 81. 133, 136, 192. Wille, N., 121, 123, 124, 125, If 129, 196. Williamson. W.. 301, 302, Fig. 1H Wimmel, Th., 125. Wolffia. 234. Worsdell, W. C, 303. 306, 308, 301 Wylie, R. B., 157, 170. Xenia, 179. Xyridaceae, 56, 235, 264, 266. Yucca, 63, 64, 77, 285; gloric 84. Zamia, 305, 307; floridana, 304, j| 111. Zannichellia, 27, 28, 46, 51, 192, 285; palustris, l'jr,. Zea, 94, 98, 153, 157, 158, 172, 17j Zingibenu-eae, 171, 237, 264, 2V Zinger, X., 56, 148, 150. Zizania, 315. Zostera, 37, 74, 77, 122; mari 81, 124. Zygomorphy, 15, 16. Zygophyllac-eae, 20, 247. THE END PLEASE DO NOT REMOVE CARDS OR SLIPS FROM THIS POCKET UNIVERSITY OF TORONTO LIBRARY