lit " ^ ■ 1 ■ 5 < ^^^1 ^ 1 ^orilj Caroliwa ^tate College This book luas presented Baltik^pe mo GRteNSBOROK.- »ASHINGTo4':oc j This BOOK may be kept out TWO WEEKS ONLY, and is subject to a fine of ¥fff^ CENTS a day thereafter. It is due on the day indicated below: BURSA BURSA-PASTORIS AND BURSA HEEGERI BIOTYPES AND HYBRIDS BY GEORGE HARRISON SHULL WASHINGTON, D. C. Published by the Carnegie Institution of Washington. 1909. Carnegie Institution of Washington Publication No. 112 Papers of the Station for Experimental Evolution No. 12 THE CORNMAN PRINTING COMPANY CARLISLE, PENNSYLVANIA BURSA BURSA- PASTORIS AND BURSA HEEGERI BIOTYPES AND HYBRIDS. Bv George Hakrison Shull. INTRODUCTION. Darwin recog'nized in the facts of variation a key to the riddle presented by the miiltiformity and many obvious interrelations of all living" things. Since the appearance of the Orig-in of Species the observation and discus- sion of variations have assumed a dominant place in biolog"y, and a serious conflict has recently developed reg-arding" the interpretation of the observed fact^. It is now g-enerally recognized that this conflict can be brought to a termination only through the application of experimental methods. Inspection alone can not decide the question as to how an observed varia- tion originated and what bearing it may have on the future of the race in which it occurs. Studies in the museum and in the field only discover the fact, and to some extent the rang-e, of variation occurring under a more or less limited and inadequately known rang-e of conditions, and can not cer- tainly determine its cause or causes ; neither can these means supply more than a suggestion based upon insecure inference as to the hereditary nature of any variation. The causes of variation can be determined only by subjecting- eqiiivalent material to different controlled conditions, and their hereditary relations can be learned only through the conduct of pedigree- cultures. We already know, as a result of experimental work in these directions, that variations are of fundamentally different types, having" different causes and obeying different laws of development and heredity. A knowledg-e of these facts impresses an important principle, namely, that the rang"e of applicability of any conclusion reached by the investigation of one class of material or one characteristic can be determined only by similar experiments with other material and other characteristics. This has been one of the guiding principles in the institution of the ex- perimental work at the Station for Experimental Evolution of the Carnegie Institution of Washington. Numerous species of plants and animals of quite wide relationships have been brought under observation, and it may be expected that each will in time g"ive material assistance in determining to what extent principles and hy]5otheses now available have general validit}', and will also lead to the discovery of such new ]irincii^les, or such modifi- 3 Library H. C, State CoUesr© g ^ )54*? 4 BURSA BURSA-PASTORIS AND BURSA HEEGERI : cations of the old ones, as may be necessary for the interpretation of the observed facts of evolution. Among the species now being studied , one that has very early yielded results of interest in this connection is the common shepherd' s-purse of our dooryards, Bursa bursa-pastoris (L.) Britton. Bursa {Capsella) bursa-pastoris is known to the taxonomist as an exceed- ingly variable species, which seems to have been brought to this country from Europe, but which is now naturalized and almost universally distributed throughout the North Temperate zone. So strikingly different, qualita- tively, were the characters of different individuals observed growing side by side in nature that it was difficult to believe that they all belonged to a single series of fluctviating' variations, and when the opportunity offered to make pedigree-cultures this species at once suggested itself as favorable material. This thought was based not alone upon its apparent polymorph- ism, but also upon its hardiness, ease of culture, and the impossibility of its having been subjected to any of the artificial conditions of isolation, crossing, etc., which are usually thought to render plants of economic value unfit to give information regarding the behavior of plants in nature. No one has ever attempted to "improve" the shepherd's-purse, and although its rapid extension over its present great range is undoubtedly dependent upon the agency of man, both in supplying suitable habitats and more directly in the transportation of the seeds, yet in performing these operations his work has been wholly unintentional, and he is therefore to be classed with the other accidental agents of nature, thus leaving to Bursa the solution of its problems of maintenance, extension of rang'e, and evo- lutionary progress, under conditions which are fundamentally like those that mvist be met by any other species in a state of nature. These cultures were begim in April, 1905, and were continued until the spring of 1907, when they were temporarily abandoned because other in- vestigations necessitated my absence from the Station for Experimental Evolution during rather extended periods, and thus made it impossible to continue the work advantageously on a species like Bursa, which shows no dependence upon the seasons, but blooms and fruits whenever external conditions are such as to make its development possible. It is hoped that these studies on Bursa may be continued in the not distant future, but as some of the conclusions arrived at have been already presented before sev- eral scientific bodies,* it seems desirable to publish a more comprehensive account of the work than has been done up to this time, even though the evidence is in many places more or less fragmentary. When undertaking such cultures with any new class of material much that is necessary for the most satisfactory and economic conduct of exper- *Sections F and G, A. A. A. S., New York, December, 1906; Seventh International Zoological Congress, Boston, August, 1907; Botanical Society of America, Chicago, January, 1908. BIOTYPES AND HYBRIDS. 5 iments must be learned by experience, for, as said before, mere inspection of an individual or of a single g-eneration can not distingiiish between im- portant and unimportant variations until the pedig-rees themselves furnish solutions to the question. For this reason there must be in the beginning a much more minute analysis of the material than is later found necessary or desirable. Most of the cultures which I have made thus far have been directed toward determining- what variations are of the fluctuating kind and what are ftilly transmissible to the offspring. In one group I have now had large numbers of plants of the fourth pedigreed generation and a few of the fifth under observation, and in a number of other cases the third generation has been extensively grown. I find that while certain variations which were selected disappear in the first or second generation, others remain constant, easily recognized differen- tiating marks which, except in one form, show no transgression of the char- acteristic features of any other form studied. These forms are, therefore, distinct elementary species, or biotypes, each characterized by certain con- stant features and each with its own normal range of fluctuating variability. The systematist has not yet decided what treatment to give to elementary species, and any nomenclatorial scheme must be regarded, therefore, as purely tentative ; but utility can not wait for concerted action on the part of taxonomists in devising a suitable systematic designation of elementary species, and I have therefore for the sake of convenience assigned to these elementary forms of Bursa simple names which can be attached to the accepted specific name to form a trinomial. I was at first inclined to use binomial names which would leave the Linnean specific name, bursa- pastoris, as the valid name of the aggregation of elementary forms having the same general habit and the triangular or obcordate capsules. It might, then, be looked upon as a superspecific name which would remain just as useful in the everyday conversation and experience of the botanist as when the aggregation for which it stands was believed to be a unit. The fact that corresponding series of elementary species or biotypes may occur in different related species, as will be shown later in the discussion of the hccgeri hybrids, makes the trinomial much to be preferred, for cor- responding forms may then be given the same name without confusion. Thus Bursa bursa-pastoris lieteris and Bursa heegeri heteris may be used to denote two forms which are alike in rosette -characters but different in the capsule-character, the latter character being accepted as of specific value. That Bursa bursa-pastoris is a composite species was first given public recognition by Lotsy (1906) about a year after my cultures were begun, and his statement did not come to my notice until after I had presented my first account before the American Association for the Advancement of Science at its New York meeting, December, 1906. His statement is very brief, and the chief interest of his account consists in the photographs 6 BURSA BURSA-PASTORIS AND BURSA HEEGERI '. which show samples of three families which bred triie to rather strikinoflj' distinct characters in his cultures. To these several forms he assi.arned trinomial names, thoug-h no adequate description is .sriven. No statement is made as to the extent of the cultures upon which their standin.sr as perma- nent biotypes is based, nor is there any statement as to the extent and nature of the fluctuating- variability of each of the forms figured. Until attention is given to such matters by the student of pedigTee-cultures it will be im- possible to properly estimate the security of conclusions drawn, or to rec- og-nize with any certainty the identity of biotypes discovered by different investigators. To obviate these difficulties, I wrote to Dr. Lotsy for pedigreed seeds of his several types, in order that they might be grown beside my own for the purpose of testing their identity, but received the reply that none were available. The questions as to whether Lotsy 's names are therefore to be considered nomina Jiuda and whether I am justified in assigning new names to the biotypes to be discussed below, which may or may not be identical with one or more of Lotsy' s named elementary species, I leave to the tax- onomist to decide. Indeed, in adopting the names used in this paper I have been governed entirely by the demands of present utility and not by any thought that these names will be accejDted by taxonomists as having proper standing in the nomenclatorial system now recognized by them. Still more recently, Almquist (1907) has published over 70 named " ele- mentary species" of B. bursa-pastoris, but these are described in rather general terms, and the cultures were conducted in the open garden un- guarded, usually for but 2 to 4 generations. Almquist states (p. 5) that after this length of time the cultures had either died out or were no longer pure. He assumes that this " loss of purity " is due to vicinism with the other elementary species growing on neighboring plots, and some of it doubtless was. In my own cultures I have observed that some of the dif- ferentiating characters used by Almquist disappear upon continued breed- ing, even when vicinism is carefully excluded. It seems that members of the same biotype coming from different habitats may retain certain fluctu- ating differences, such as differences in texture of leaves, degree of redness of veins, degree of epinastic growth in the petioles, etc., for several gener- ations of uniform treatment, but finally merge into identical forms, even when crossing with other biotypes is entirely precluded by careful guarding. While I can not say with certainty that any of Almquist's "elementary species" are not distinct and permanent types, my experience indicates emphatically that he has not demonstrated that they are. I think it very probable that a number of the elementary species figured and described by him will not stand the test of longer culture under more carefully controlled conditions. However this may be, it seems evident that the number of elementary species of Bursa may be considerable. At the time of the first BIOTYPES AND HYBRIDS. 7 presentation of my studies on Bursa before the American Association for the Advancement of Science (December, 1906), I was prett}^ sure that I had demonstrated the existence of 11 biotypes, but the disappearance of one or two of these since, led me to reduce the number to 4, as imbHshed in the extract from that pai:)er (Shull, 1907). These four forms, whose distinctness and permanence I have demon- strated beyond a possible ([uestion, liave been studied in their hybrid com- binations, and it is the cultures oi these and their hybrids with which the present paper will mainly deal, though some features of other cultures will be discussed. It is doubtful whether any bit of technique that has been recently added to the tools with which the biolog-ist may operate in unlocking- the mysteries of protoplasmic organization is of so far-reaching importance as the process of hybridization. I say has been recently added because, although the art of producing hybrids is very old among the breeders of animals, and nearly 200 years old among gardeners, little of scientific value could be secured by means of hybridization until some of the fundamental laws involved in the process were recognized, and it is only within the last eight years that biologists have gained sufficient insight into the behavior of hybrids to give interpretations of the results of hybridization any value as indications of protoplasmic structure and behavior. Two general types of hybrids are readily recognizable, namely, the con- stant and the splitting. The former may or may not give indication of the characteristics of their parents, being usually, but not always, intermediate between the parents in most of their characters. The splitting hybrid always indicates by its offspring what were the characteristics of its im- mediate ancestors. It is the latter type of hybrid which is of the greatest usefulness in giving insight into the stinicture of the germ-plasm. Hybrid- ization in such cases does not only serve to unite in the same individual all of the characteristics of both parents, but as successive generations are followed, it results in a complete analysis of all the points of difference existing in the two biotypes between which the cross was made. The conduct of such analysis by hybridization is particularly simple in the case of plants which, Hke Bursa, reaclily self-fertilize, because once the first-generation hybrid is secured the process of analysis goes on genera- tion after generation, until all the allelomorphic dilferences of the parents are made manifest by being separated in a pure state in different individuals. In some respects even more satisfactory evidence of the elementary char- acter of two forms is to be derived from their behavior when crossed than from their conduct in straight pedigrees. If, for instance, two forms sup- posed to be elementary to each other should be, instead, merely extreme fluctuants of a single biotype, their cross-bred progeny would show a fluc- tuating series, including perhaps the two parental extremes, but could hardly 8 BURSA BURSA-PASTORIS AND BURSA HEEGERI : be expected to produce a constant intermediate progeny, nor one which would show the two parental conditions in an expected Mendelian ratio. Occasionally in the course of pedigree-cultures some slight difference in treatment will produce a modification in some progeny, which may lead to doubt as to its classification, and perhaps even after several generations this doubt will not be wholly removed. In such cases hybridization is almost certain to give decisive results, as there often appears to be a more characteristic development of morphological features in the hybrids than in the pure-bred strains. This may be due to the fact that the allelomorphs are brought into different relations with other allelomorphs in the recom- binations which take place in the second and later generations, thus elimi- nating any inhibiting or modifying influence which may have resulted from the constant presence of some particular allelomon^h or combination of allelomorphs, or of some permanent condition of the cytoplasm, in the pure strain. This elimination of modifying factors would allow a more accurate definition of the unit-characters involved. Or, on the other hand, the better development of characters in the hybrids may be due to the fact, long recognized, that heterozygosis produces a stiinulation which increases vegetative vigor, and this should have the same effect as good cultural conditions in bringing all characters to their full tjq^ical development. MATERIAL AND METHODS. For the beginning of the cultures, seeds were secured from a number of more or less strikingly different individuals growing in the vicinity of Cold Spring Harbor, Long Island; Chicago, Illinois; New Carlisle, Ohio; Man- hattan, Kansas; and Tucson, Arizona. Leaves of each of the plants chosen as parents for the cultures were carefully preserved, in order that the con- ditions of the offspring might be compared with those of the parent and, in each pedigreed family raised since, care has been taken to preserve in the herbarium samples of every variation of sufficient magnitude to strike the eye. From 20 lots of seeds of Bursa biirsa-pastoris brought in from nature and 1 lot of B. heegcri received hx Dr. MacDougal from Professor Solms- Laubach, I have now had under observation about 200 pedigreed families, including something over 26,900 individuals of Bursa bursa-pastoris, 2 small families of Bursa heegeri, and 5 families representing reciprocal crosses between these two species and involving over 2,500 individuals. All of my cultures have been carried on imder glass, the first 7 months in the sk>--lighted room of the laboratory at the Station for Experimental Evolution, and since that time, /. e., after December 7, 1905, in the glass propagating-house at the same place. It was found that the light was too dim in the former room to bring out the characteristic features of the plants and to keep them in a state of vigorous health. The leaves became more BIOTYPES AND HYBRIDS. 9 elong'ated than is normal, and the lobing' more shallow and therefore less characteristic, so that it was less easy to estimate the uniformity and dis- tinctness of the several pedigrees than when they were grown under more favorable conditions. Even specimens belonging to the most deeply and distinctly lobed families, when grown in the darker portions of the room, retained their unlobed, juvenile tjq^e of leaf throughout life, sending up a weak flower-stem from the juvenile rosette, and in such cases the relation- ships could not be recognized, since the early leaves of all the forms studied are very similar. The same difficulty was also experienced in the propa- g'ating-house when the inembers of certain families were allowed to remain too long crowded in the seed-pans . The offspring of these juvenile plants have not been extensively studied , but from several families evidence has accumulated which indicates that when again given favorable conditions the offspring of these characterless plants return to the characters of the family from which they sprang. In all cases the seeds have been sown in soil sterilized in an autoclave in the manner usually adopted by students of pedigree-cultures, and the efficiency of the method is inferable from the fact that in all these cultures only 3 seedlings occurred whose origin was unknown. These unexpected seedlings were an Oxa/is, a Molhigo, and an Erechtites, and it does not seem likely that any of these withstood the long^-continued high temperature of the autoclave, but rather that they were blown through the ventilators of the propagating'-house in a heavy wind-storm. It was soon demonstrated that Bursa has many features which make it advantageous material from a technical point of view for pedigree-culture work. The habit of the plant, consisting as it does of a moderately lax rosette and nearly naked, erect flower-stem, allows the preservation of the rosettes as herbarium specimens whose characters are almost as easily studied as are those of the living plants, and the inflorescence may be cov- ered with paraffin-paper bags to prevent chance crosses with other speci- mens without appreciably interfering with the photosynthetic work upon which the healthy development of the plant depends. The small size of the plants makes it possible to raise them to maturity in 3-inch pots in many of the forms, though it is usually found advanta- geous to repot to 4-inch pots those which it is desired to keep for seed. This quality allows a large number of specimens to be raised in a small compass. In most instances the life-cycle is short, requiring only 3 to 4 months between the sowing of the seed and the gathering of the rij^e seed of the earliest matured individuals. In several of the forms of B. bursa-pastoris, and in B. heegeri, however, 8 to 9 months were needed. The seeds ger- minate in 5 to 8 days without a period of rest, thus making it possible to accumulate data from a number of successive generations in a short time. 10 BURSA BURSA-PASTORIS AND BURSA HEEGERI The flowers are adapted to both cross- and self-fertilization, all the evi- dence now at hand indicating- that the latter method is normally by far the more efficient. Leaving out of account 2 of the 21 original cultures which showed by their constitution that their parents were hybrids, less than 1 per cent of the plants raised from seed collected in nature showed evidence of being the result of cross-fertilization between different biotypes . Crosses between different flowers of the same plant and between different indi- viduals of the same biotype may take place somewhat more frequently than crosses between the flowers of plants belonging to different biotypes, though nothing in the resultant offspring gives any clue to the frequency of such crosses. Fig. 1. — Enlarged sections of buds and flower of Bu?-sa bursa pastoris, show- ing three stages in anthesis. A. Exposure of the stigmatic surface for the reception of foreian pollen. B. Anthers opening in contact with the stigma, thus insuring self-pollination. C. The flower fully open, allowing the access of the visiting insects to the pollen. All magnified 20 diameters. The conditions which favor cross-fertilization are : (a) Slight prote- rogyn}', which allows the stigina to receive foreign pollen some hours before the anthers of the same flowers dehisce. The distal portion of the globular or disk-shaped stigma is exposed between the tips of the unopened sepals and petals (fig. 1 , a) . Qy) Although the fully developed inflorescence is a typical raceme, the flowers and buds are arranged in a nearly flat-topped corymb having the flowers at the circumference, giving this part of the inflores- cence a condition quite analog'ous to that of the head of the Compositae, in which the whole inflorescence appears to act the part of a single flower in the attraction of insects. Small insects, particularh- flies and small bees, visit the flowers freely. These rest upon the top of the inflorescence as a whole while visiting the several individual open flowers about the circum- ference. In these fully open flowers forming the exterior circle of the inflo- rescence the anthers have opened, while in the second circle the summits of the stigmas are exposed on the same general level as the rest of the corymb. As the insect walks about over the top of the inflorescence, the BIOTYPES AND HYBRID.S. 11 chances appear favorable for the lodg-ment of i)ollen from the outer circle of flowers, or from the flowers of other plants, upon the exposed stigma-tops of the second circle. Adaptation to self-fertilization is found in the facts: («) that the stigma is also receptive on its under (proximal) surface as well as on the upper (distal) surface, and (/;) that the anthers dehisce while they are still held in contact with this under surface of the stigma by the erect segments of the perianth (fig. 1, b). vSoon after the anthers open the petals begin to spread, thus exposing- the pollen to be carried away to other flowers by visiting insects (fig. 1, c). At the beginning of these experiments paraffin-paper bags of suitable size were not available, and as insects were not abundant in the room used for the cultures, the only precaution taken to guard against cross-pollination was to set the flowering specimens intended as seed-plants somewhat apart from each other and from other flowering specimens. As will appear later, a few individuals, of unexpected character, are doubtless to be attrib- uted to this unguarded condition of the earlier cultures, but it will also be seen that the percentage of such chance crosses is extremely small. As soon as possible, suitable paper bags were secured, and since then the cul- tures, with few exceptions, have been carefully guarded. Although crossing among the unguarded cultures in the greenhouse has been of rare occurrence, it was evidently more frequent in the material secured from nature. Of the 21 original cultures 2 proved to be of hybrid origin, while a fraction of 1 per cent of the rest indicated by their atypic condition that they were probably the result of cross-pollination. Only when the pollen comes from some form which is dominant to the pistil- parent is the fact that a cross has taken place obvious in the first genera- tion. It seems fair to assume that on the average as many crosses take place with a recessive pollen-parent as with a dominant, and this assump- tion would require that hybridizations occur with twice the frequency with which they become obvious in the F, offspring. On this basis the fre- quency of cross-pollination between different biotypes of Bursa in nature, as indicated by these cultures, is about 1 to 65 as compared with the fre- quency of self-pollination and crosses between flowers of the same plant or between plants of the same biotypc. Of course this ratio is based upon a very limited number of specimens and can be expected to vary greatly in different lots of material of the same magnitude, but at least the great preponderance of self-fertilization may be safely inferred in this species when in a state of nature. These characteristics of Bursa which make the production of pure self- fertilized lines easy are opposed to the ease with which cross-fertilization may be controlled. The fact that self-fertilization takes place before the petals spread makes it necessary to carefully remove the stamens about a 12 BURSA BURSA-PASTORIS AND BURSA HEEGERI : day before the buds open. As the buds are at that time quite small, the technique of cross-pollination is somewhat delicate. However, with a needle-pointed forceps the calyx, corolla, and stamens may be readily cut away from about the young" pistil, and the eye and hand soon become so skilled that the work can be done swiftly and with as much accuracy as may be attained in the castration of a lily. PURE CULTURES OF BURSA BURSA-PASTORIS. The four biotypes of Bursa bursa-pastoris with which this paper mainly deals are distinguished from each other by certain characteristic lobings of the leaves. For convenience I have assigrned to them the names Bursa bursa-pastoris heteris, B. bp. tenuis, B. bp. rhomboidca, and B. bp. simplex. It has been impossible to determine which of Almquist's "elementary species" agree with these, but it is almost certain that he has assigned names to several fluctuations of the same biotype. If this is true my four types are more inclusive than his and under my names will need to be grouped a number of forms which he has considered distinct elementary species. Bursa bursa-pastoris heteris n. sp. element. Plants belonging to this biotype have the leaves divided to the midrib, the terminal lobe being usually separated from the nearest lateral lobes by deep, clean-cut incisions. The lateral lobes consist essentially of two fea- tures — an elongated, attenuate portion which I call the ' 'primary lobe, ' ' and a more or less rounded or angular portion which forms a ' ' secondary lobe ' ' in the distal axil of the primary lobe (fig. 2). As in the characterizations of all the following forms, this description refers to the climax-leaves of the rosettes in properly grown specimens, since the juvenile leaves of all the elementary species of Bursa are entirely unlobed, and starved or crowded specimens of all the forms may reach maturity with only juvenile leaves, as already stated. Several of Almquist's recently described forms would obviously range themselves under this description, though it is always pos- sible, of course, that forms which possess the described characteristics may have still other permanent characteristics which would render them distinct elementary species. The first of the forms recognized by Almquist which can certainly be placed here is his Capsella bp. rubella (see his figs. 5 and 6). His C. bp. angustiloba (his figs. 56 and 57) represents the most pronounced develop- ment of this type, and others which probably belong here are reiiteri (fig. 11), grandiflora (figs. 12 and 13), hiemalis (fig. 20), grossa (fig. 22), au- tiimnalis (fig. 24), segetum (fig. 28), wittrockii (fig. 44), rhombea (fig. 48), rhombella (fig. 53), ellipsoidea (fig. 55), 2iXvdifucorum (fig. 60). My first pure-bred family of B. bp. heteris (pedigree-number 040.3) was grown from seeds collected by J. Marion Shull at Edgewood, New Carlisle, BIOTYPES AND HYBRIDS. 13 Fig. 2. — Bursa bursa-pastoris heteris. A typical specimen of my first pure cul- ture of this biotype. Fig. 3. — Bursa bursa-pastoris heteris. Climax leaves of the rosette of a speci- men growing in nature in Jackson Park, Chicago. 14 BURSA BURSA-PASTORIvS AND BURSA HEEGERI : Ohio, May 28 and June 2, 1905. The cHmax leaves of the parent were of the most pronounced heteris form. From seeds sown June 26, 1905, 39 specimens were grown, all of which agreed with the parent and with each other in the character of lobes as described, though there was some fluctu- ation in general aspect due to difference in elongation of the rachis by means of which the lobes were more or less crowded, some variation in the attenuation of the primary lobes, in the degree of lobation of the upper rosette leaves, in the angle which the rosette leaves made with the hori- zontal, and in the prominence of the rounded secondary lobe. This sec- ondary lobe was characteristic, however, and in not a single individual was it absent in the climax leaves of the rosette. The significance of several of these variations was tested by using their best representatives as parents of a second generation. From 3 of these (053.30, 053.31, 053.32) have been raised 839 specimens, all but the 54 offspring of one individual being strictly like the grandparental and parental form. The one aberrant family (053.30), the parent of which had the upper leaves of the rosette less deeply lobed than usual, showed com- plete suppression of lobes in 5 individuals, and in the remaining Z^ which were noted the primary lobes were broader than usual and the secondary lobes less prominent. It is desirable to test the possibility that these char- acteristic secondary lobes may be rendered completely latent by selection, and, if this is possible, to discover by what means they may be again made manifest. At least two other pure-bred cultures coming in from nature belong to B. bp. heteris, but were considered for two generations to represent two very distinct elementar}^ forms. It has been the striking modifications ob- served in these two pedigrees which have led me to suspend judgment on all my cultures except those whose behavior in hybrid combinations has left no possible question as to their distinctness and permanence. The basis for the assumption that these two pedigrees represented dis- tinct elementary species was the facts that the aspect of each of these fam- ilies was very different from that of any other culture and that there was great uniformity among the individuals belonging to either single family, /. e. , while there was a strong break between the families there was almost no variation within the family. This was presumably the basis of Alm- (juist's estimates as to the distinctness of his forms, and the following ac- count of these two families supports my attitude of doubt as to the soundness of his results. 040.15: This specimen was found growing in Jackson Park, Chicago, by Charles A. ShuU, who collected the seeds in the summer of 1905. Climax leaves taken from the rosette show only a slight development of the heteris characters as described above (fig. 3 ) . Though the sinuses reached the mid- rib, the primary lobes were not sharply attenuate, and the secondary lobes, BIOTYPES AND HYBRIDS. 15 w y. ^\ Fig. 4. Fig. 5. Fig. \.— Bursa bursapastoris heteris. Typical specimen from offspring of plant shown in fig. 3. Fig. 5. — Bursa bursa-pastoris heteris. Offspring of sib of plant shown in fig. 4. 16 BURSA BURSA-PASTORIS AND BURSA HEEGERI : thoiig"h present, were not strongly marked. The seeds were sown in the propag-ating'-house December 27, 1905, and produced 106 offspring, all of which had a peculiar g"rayish aspect, owing to the fact that the leaves were freely dotted with small specks of red. The rosettes were very lax, the relatively few leaves made a rather wide angie with the horizontal, and the leaves, while possessing both the primary and secondary lobes, had the latter small and angular and little more marked than serrations borne by the primary lobe, which all experience up to the present time indicates are of only fluctuating value. (See fig. 4.) All had simple slender stems which came to bloom in 3 to 4 months from the time the seeds were sown as compared with 8 months required by the above-described family of typical B. bp. heteris. Self-fertilized seeds of two of these plants (0515.95 and 0515.96) were sown in May, 1906, and g"ave uniform progfenies, but having; an aspect quite different from that of the parent g"eneration just described. Although the rosettes in this second generation were strongly ascending and were few- leaved, as in the preceding' generation, the texture of the leaves was much less firm, the grayish aspect was wholly lost, owing to the absence of the red specks, and secondary lobing was also much reduced, giving- a condition resembling the g^randparent as it grew in nature (cf . figs . 3 and 5 ) . The con- ditions in the propagating-house during- the development of these families were unsatisfactory, and their g-reat change from the characteristics of the preceding- generation did not shake my faith in their distinctness froin my other types, since within the family there was still great uniformity, seem- ing thus to demonstrate that whatever differences were observed were due to the differences in environment during the time of development of these two generations. An average specimen from one of these two families was chosen as the parent of a third generation. The pollination of this plant (0695.158) was unguarded, but it was grown well separated from all other Bursa cultures. The seeds were sown on November 1, 1906, and produced 213 plants, all of which possessed well-marked the characteristics of B. bp. heteris, having completely lost all the peculiarities which had led me to believe that this pedigree belonged to a distinct biotype. (See fig. 6.) Other proofs that 040.15 was a specimen of B. bp. heteris were derived from hybrid families formed by crossing its offspring with plants belonging to other biotypes. Except in one case only the Fj hybrids from these crosses have been studied and they will not be considered in detail at this time, but these first- generation hybrids showed the characteristics which they should have pos- sessed if typical B.bp. heteris had been used . The one F^ family (0693 . 203 ) which has been reared from these hybrids will be considered later. (See p. 42.) BIOTYPES AND HYBRIDS. 17 050.80: Another orig-inal culture which now seems to belong: to B . dp. heteris was the offspring' of a very robust plant collected in a dooryard near Cold Spring- Harbor, Long- Island, April, 1906. This plant was taken up and potted in the greenhouse, where its pollination was g;uarded. The earlier leaves of this plant were recog'nized as resembling- B. bp. heteris y ex- cept in the less sharp attenuation of the primary lobes, but later leaves— the climax-leaves — were larg-e, and, in addition to the secondary lobes characteristic of B. bp. /leteris, they had somewhat quadrang-ular secondary Fig. ^.— Bursa bursa-pastoris heteris. Offspring of sib of plant shown in fig. 5. lobes in the proximal axils of the primary lobes, and these square lobes were sometimes almost cut off from the primary lobes, giving" the leaf the peculiar form usually described as interruptedly pinnate (fig-. 7). The seeds of this plant were sown May 23, 1906, and produced 352 offspring:. These were badly damagfed by the thrips, 81 being: killed and 57 so stunted as to make an estimation of their characters uncertain. The remaining- 214 formed a consistent group unlike its parent and also unlike any other 18 BURSA BURSA-PASTORIS AND BURSA HEEGERI of my cultures. Few .showed even a sugfgestion of the squarish lobes which were so conspicuous in the parent, thus indicating- that the strong' development of that characteristic in the parent was in all probability merely a fluctuation. These plants differed from those considered typical of B. dp. heteris in having the primary lobes of the climax-leaves oblong and blunt, not attenuate. However, in some specimens the later leaves showed the attenuate lobes of typical B. hp. heteris, and this fact leaves little doubt that another generation would have completely demonstrated that this family belongs to B. bp. heteris. Whether a culture (0645) produced from seeds sent by Dr. D. T. Mac- Dougal, from Tucson, Arizona, is likewise identical with the B. bp. heteris grown from seeds collected in Illinois, Ohio, and Long Island, has not been sufficiently tested. While the Tucson plants had in a most strongly marked way the primary and secondary lobes described above, there were striking' differences in the texture and color of the leaves, the form of the lobes, and the form, size, and texture of the stem-leaves or bracts (fig'. 8). As only one generation of plants has been grown from these seeds, no proper grounds exist for attempting to decide as to the permanence of the differences exhibited, but it might be expected that plants from the hot, dry, intensely lighted plains of Arizona would display considerable changes of a purely transitory nature on being transferred to the moist atmosphere and relatively dim light ^ of a more northern propagating-house . Fig. 1.— Bursa bursa -pastorts _. . ^ , . . t , ^ . heteris. Climax leaves of a rield observation indicates that the type or specimen growing near Cold rosette possessed by Bursa at Tucson, of which Spring Harbor, Long Island. , . ^ . . ^ . . this culture was a fair example, is the common type if not the only type of rosette displayed by Bursa in that locality and westward. Many facts now at hand sugg^est that the heteris type is the prim- itive type of rosette from which the forms to be described below have been derived, and the relatively wide geographical distribution of this type is in strong support of this view. Bursa bursa-pastoris tenuis n. sp. element. This differs from the preceding type in several important features. The sinuses are relatively shallow, rarely extending nearh- to the midrib in very strongly developed individuals. The terminal lobe is not separated from the nearest lateral lobes by deep, clean-cut sinuses, but these more distal sinuses are relatively shallow, so that one can with but scant pro- priety speak of the terminal lobe as a definite morphologfical structure; BIOTYPES AND HYBRIDS. 19 there is no incision on either margin of the lateral lobes and hence no rounded lobe corresponding with what I have called the "secondary lobe " in B. bp. heteris, though there may be a slight expansion of leaf-tissue in that reg-ion, especially in strongly developed specimens, which no doubt corresponds to the secondary lobe (fig. 9). All the lateral lobes tend to be more or less slender, elongrated, and acute. In the more robust speci- mens there is apt to be some secondary lobation, but in all cases these secondary lobes are also attenuate. A secondary spur directed outward Fig. S. — Bursa biirsa-pastoris heteris (?) grown from seed received from Tucson, Arizona. and proximad from near the base of the lateral lobes is often noted in well-developed plants belonging' to this biotype (fig. 10). In many in- stances the long, slender lobes are somewhat recurved at the tips, but in other cases all lobes are practically straight. Under this type may belong- Almquist's C. dp. pedemontana (his figf. 17), leontodon (fig\ 3l), dentata (fig. 35), lacerata (fig". 37), bergiana (fig-. 38), laxa (fig*. 39), querceti (fig-. 40), ramselensis (fig-. 41), and linearis (fig-. 58). 20 BURSA BURSA-PASTORIS AND BURSA HEEGERI Fig. 11. Fig. 9. Fig. 10. Fig. 9. — Bursa bursa-pastoiis tenuis from my first pure culture of this biotype. Fig. 10. — Bursa bursa-pastoris tenuis from ;i hybrid progeny. Fig. 11. — Bursa bursa-pastoris tenuis. A stunted sib of plant shown in fig. 9. BIOTYPES AND HYBRIDS. Fig. 13. 21 Fig. 12. Fig. 12. — Bursa bursa-pastoris rhoniboidea. Grown in best light-relations pos. sible in sky-lighted room. Characteristic incisions not well developed. Fig. 13. — Bursa Bursa-pastoris rJiomboidea modified by growing in dim light. 22 BURSA BURSA-PASTORTS AND BURSA HEEGERI : The two specimens (040.19 and 040.26) which form the basis of all my pure-bred cultures of this biotype, gferminated in vessels of earth over which were unpacked Bursa seed which had been sent by Prof. H. F. Rob- erts, of Manhattan, Kansas, and as the soil of one of these vessels had been sterilized it seems almost certain that these plants came from seed sent from Kansas. Seeds of these two plants were sown September 22, 1905, and January 31, 1906, and tog-ether gave a progreny of 361, all but 4 of which were referable without question to the type of the parents. These 4 specimens in family 0519 which were not quite in agreement with the rest had the lobes comparatively broad and short, though still strongly acute, and the margins were slightly crenulate or denticulate (fig-. 11). As these specimens were somewhat stunted, it is thought that their differences may have been due to the causes producing- the stunting, but this point has not yet been tested. Besides these two pure-bred families of B. bp. tenuis coming presumably from Manhattan, Kansas, this fonn appeared as a component of a hybrid family received from Edgewood, New Carlisle, Ohio, and also as a mem- ber of a hybrid family from Chicag-o, Illinois. These hybrid families wull be discussed later. Bursa bursa-pastoris rhomboidea n. sp. element. This, like B. bp. heteris, has the leaves divided to the midrib and pos- sesses a similar, more or less rhombic terminal lobe, set off by deep sinuses from the nearest lateral lobes. Each lateral lobe of the climax-leaves usually shows a prominent incision on its distal marg-in, by which a lobe is formed next to the rachis, corresponding to the secondary lobe of B. bp. heteris, and in well-developed specimens there are usually one or two sim- ilar incisions on the proximal marg-in. All lobes formed by these incisions are usually obtuse or broadly ang-ular. The terminal portion of the lateral lobes has in the best-marked examples a nearly rhombic form, which sug-- gested the name (see figs. 18 and 20, and plates 2 and 4). When grown under unfavorable conditions the characteristic incisions may be lost, the prominent incisions setting- off the rounded secondary lobe being- the most persistent. Almquist's C. bp. siibalpina (his fig. 46), densa (fig. 51), polyedra (fig-. 52), and perhaps two or three others may belong- here. Three pure cultures demonstrated to belong to B. bp. rhomboidea have been reared from specimens or seeds brought in from nature, and the same elementary species has been included as hybrids in two other orig-inal families. Complete proof of its elementary character will appear below in the section on hybrids. Cultures derived from two plants g-rowing- side by side near the Brooklyn Institute's Marine Biologfical Laboratory at Cold Spring- Harbor, Long Island, were thought for a time to belong to a biotype distinct from the BIOTYPES AND HYBRIDS. 23 more characteristic specimens of B. bp. rJwniboidea secured in hybrid fam- ihes. These two individuals were growing' in a situation which received only the morning" sun. The leaves were divided to the midrib, but the oblongf, obtuse lateral lobes frequently had no incisions, thoug'h in some cases the middle lobes of the leaves had the more characteristic incision on the distal margin and less frequently a similar incision occurred on the proximal margin. The lobes were rather distant from each other, gfivingf a very unique appearance to these plants. Seeds of these two supposed sibs were sown on June 6 and 14, 1905. The first of these (040.1) pro- duced 65 offspring-, which were studied in their relation to different condi- tions of environment in order to gfet a clue to the suscepti- bility of Bursa to immediate modification by variations in the chemical and physical con- ditions of the soil, differences in soil-moisture, atmospheric humidity, intensity of light, etc. Specimens which were grown under as favorable conditions as the sky-lig-hted room pro- vided were essentially identi- cal with the parent (fig. 12). Aside from the complete sup- pression of lobes and long dela}' of the flowering- period in dim lioht (fif>"S. 13 and 14) ^^G- 1-i- — Bursa bursa-pastoris rhomboidea. A sib the most'marked effect of en- ^^>'^"ts shown in figs. 1l> and i:;, and of the same age, showing the complete suppression of lobes, vironment was noted m plants ^^^^ p,^„j ^^^^ ^^^^^^ ^^^ ^^,^ ,^^^^^1^^ i^ ^ p^^^j^ kept in a nearly saturated illuminated corner of the sky-lighted room, atmosphere attained by cov- ering with g-lass jars. The leaves became long, of very thin membranous texture, crinkled and otherwise distorted, and with very shallow sinuses (fig. 15). Plants which had been grown in the dim light of the sky-lighted room of the laboratory and which showed a consequent reduction of the lobes returned to the fulh- lobed condition upon being removed to the better- lighted propagating-house. Seeds of one of these (051.19) were sown March 6, 1906, and produced a large progeny, only 100 of which were potted up and studied. These were uniform throughout, but as they were stunted by unfavorable conditions in the propagating-house it was impossible to determine with certainty their relation to other cultures which had been grown under better conditions. 24 BURSA BURSA-PASTORIS AND BURSA HEEGERI : The second of two original plants of B. bp. rJiornboidea (040.8) produced 330 offspring-, all but 5 of which possessed characters recog-nized as prop- erly belonging: to the parental type, namely, distinct obtuse lateral lobes, usually with an incision on the distal marg-in and sometimes with a similar incision on the proximal marg-in (fig. 16). The 5 individuals which were aberrant from the type of the other 325 had the essential characters of B. bp. heteris. As will be seen later, heteris is dominant over rhomboidea, and the simplest explanation of the presence of these 5 specimens of heteris in FiG. ]5. — Bursa bitrsa-pastoris 7-homboidea. Grown in nearly saturated air and moderate illumination. Leaves very thin and translucent. a family of rJiomboidea is the supposition that they represent chance crosses between these two biotypes. Another culture which was not at first thought to be referable to B. bp. rhomboidea was raised from seeds of a plant (050.82) growing in a door- yard near Cold Spring Harbor, Long Island. The parent had rather small leaves with crowded lobes and coriaceous texture. The lobes had a strongly marked distal incision and rarely a slight proximal incision (fig. 17). This plant was removed to the greenhouse April 18, 1906, and the flowers were BIOTYPES AND HYBRIDS. 25 g-iiarded ag'ainst cross-pollination. The 150 offspring: (06190) were some- what diseased and stunted for a time, but subsequently developed un- doubted characters of B. bp. rhomboidea (fig". 18). Bursa bursa-pastoris simplex n. sp. element. This biotype is like /?. bp. ton/is in that the sinuses never reach the mid- rib, but it differs in having- mostly simple rounded or triangfular acutish lobes, not attenuate. No incisions are seen in the lobes and there is no secondary lobing", even in the most vig^orous specimens, except some slig-ht Fig. 10,.— Bursa bursa-pastoris rhomboidea from the second original family. denticulation (fig-. 19). I am in doubt as to whether any of Almquist's named forms can be referred to this biotype, thoug-h it is possible that his C. bp. haiiniensis (his fig's. 33 and 34) is identical with my B. bp. simplex. His C. bp. gallica (fig". 62) and several others in whicli lobes are almost or quite absent might be placed here, but in most cases his fig'ures show evi- dence that the plants had lost their characteristic lobing: through some un- favorable cultural conditions, and if this was the cause of the reduction of the lobes, such specimens might belong- to any other biotype. 26 BURSA BURSA-PASTORIS AND BURSA HEEGERI: Two original cultures of B. bp. simplex were grown from seeds col- lected at Edg-ewood, New Carlisle, Ohio, by J. Marion Shull, June 2 to 17, 1905. In the parents of both these cultures (040.5 and 040.6) the lobes were undivided and not elong-ated, thougfh they were somewhat triang-ular- acutish. One of these (040.6) had a more tapering- apex than the other and its lateral lobes were more divaricate and more acute, but their progf- enies were g-enerally indistingaiishable from each other. Only one speci- men among- the offspring- of 040.6 had the long-, tapering- apex of the parent, this fact apparently showing- the difference between the two parents in this regfard to be a simple fluctuation. These two lots of seed were sown June 26, 1905. The first (040.5) produced 170 plants which, except for some slight fluctuations, were evi- dently of a single type. A second generation of 44 plants raised from the seeds of an average specimen (055.24) kept strictly to the same type. The other original culture (056) consisted of 425 specimens, all but 5 of which were typical B. bp. simplex. The 5 aberrant specimens belonged to B. bp. rhom- boidea (fig. 20), whose presence in this family was assumed to be due to chance crosses in nature. This assumption was tested by rear- ing a family from seeds of one of these (056.130). Its hybrid character was fully demonstrated, and the results are given in Fig. \1.- Bursa bursa-pastoris detail in the section devoted to hvbrids (p. 42). rhomboidea. Climax-leaves of _ ^ . , " ^ , a plant growing in a dooryard ^ome fluctuation was observed among the near Cold Spring Harbor, Long specimens of B. bp. simplex. A very few Island. somewhat stunted specimens had the leaves smaller than normal, somewhat shining, and with the lobes more crowded. One of these (055.103) was tested. Fully guarded seeds were sown July 18, 1906, and produced 48 specimens, all but one of which were typical B. bp. sim- plex, the one slightly aberrant specimen having a more coriaceous texture and slightly more distant lobes than the others. It is probable that this also represents a mere fluctuation. Other variations among the members of the original families have not been fully tested as yet. Thus in one the leaves were broader than in the usual form, in another the sinuses were deep and the lobes rather long, strongly divaricate, and acutish. There can be little doubt that these are fluctuations which would be but slightly if at all apparent in their offspring. 050.82 BIOTYPES AND HYBRIDS. A VARIABLE SERIES OF DOUBTFUL SIGNIFICANCE. 27 Besides the four forms named and described above, which have been shown to breed true to type with only sHg-ht fluctuations, most of my at- tention has been given to a series of cultures whose behavior has been up to the present time quite baffling". Some time I hope to understand this g-roup better, and I shall then have more to say about it, but its behavior is in such striking contrast to that of the biotypes already described that it seems only fair to give a short epitome of my results as they now appear. Fig. 18. — Bursa bur sa-pastorisrhomboidea grown from guarded seed of the plant shown in fig. 17. Two specimens (040.2 and 040.7) were taken into the sky-lighted room from different habitats near Cold Spring Harbor, Long Island, in the very beginning of these cultures, April 15 to 20, 1905, and allowed to ripen seed. The aspect of these two specimens was very diverse. One (040.2) was robust and had rather firm, thickish leaves with 4 or 5 pairs of oblong, ob- tuse, wavy lobes, while the other (040.7) had a small rosette with thin, flat leaves and few triangular lobes. Notwithstanding these differences, the offspring of the two plants, as followed in numerous cultures through several successive g-enerations, were indistingtiishable from each other. 28 BURSA BURSA-PASTORIS AND BURSA HEEGERI : The seed of the first of these (040.2) was sown June 6, 1905, and gave a prog'eny of 262 individuals not quite equal to each other and not then well understood, because these were my first cultures of Bursa. These plants were nearly uniform in their membranous texture, smooth surface, and rather light-g'reen color, but in the character of the lobation considerable diiTerences were noted. The most common form, and that considered in consequence to be typical, had only obtuse lobes resembling; B. bp. simplex. Besides these obtuse-lobed specimens there were about 4 which had the lobes decidedly acute or elong'ated. Fig. 10. — Bursa bursa-pastoris simplex grown from seeds received from Edgewood, New Carlisle, Ohio. Seeds of the other original specimen (040.7) were sown May 8, 1905, and produced 70 offspring, about 40 of which were obtuse-lobed. The rest varied through acute lobes not elongated, somewhat elongated acute lobes with slight serration on both margins to forms with well-marked attenua- tion of the lobes similar to those of B. bp. tenuis described above. Most of my cultures of Bursa have been made to determine the status of the varia- tions in the offspring of these two plants and their succeeding generations. Before taking up the discussion of particular cultures, it may be stated that these variations from obtuse lobes and shallow sinuses to more or less BIOTYPES AND HYBRIDS. 29 attenuate lobes, and many other variations, have behaved in many respects as if they were the normal fluctuations of a single biotype. While in cer- tain cases there appeared to be a marked capacity of one or the other ex- treme to transmit its character to its offspring-, the usual result of breeding- any one of these variations was a prog-eny g-iving- ag-ain the whole rang-e of fluctuation, or a considerable portion of it. Time and ag:ain forms were picked out so different from their sibs that they were thoug'ht to represent distinct elementary species, but breeding-tests showed that their offspring- return completely to the usual condition of the other related families. Fig. 20. — Btirsa bursa-pastoris rhomboidea from a family of B. bp. simplex. A half sib of the plant shown in fig. li). Over 100 pedig-reed families, including" more than 15,500 individuals derived from the two plants (040.2 and 040.7), have now been studied, and of these families considerably more than half ranged between forms with very obtuse lobes and others having- greater or less attenuation of the lobes, the extreme developments in the latter direction being- scarcely distinguish- able from pure-bred B. bp. tenuis; 17 of the remaining families, in which no attenuate-lobed element was noted, had been injured by too long crowding 30 BURSA BURSA-PASTORIS AND BURSA-HEEGERI : in the seed-pans. This cause, with the ravages of thrips, in many cases so affected the development of the plants during- several months of the summer of 1906 that it was often very doubtful just how much depend- ence was to be placed upon the observed results. However, enough of the families developed healthily to render it not improbable that all degrees of variation between completely unlobed individuals and those with highly developed attenuate lobes may be normal fluctuations of a single unit-form. Of the healthy cultures several may be taken as examples of the pecu- liar behavior of these plants . vSeveral noteworthy instances were presented in which the conditions of the parent were transmitted to the offspring with only a narrow range of fluctuation. Thus, in the second of the two original families from which all of these variable cultures sprang, there were among 70 specimens 13 which had attenuate lobes. Seeds of 1 of these attenuate-lobed specimens (057.20) were sown October 9, 1905, and produced 681 plants, all but 1 of which showed almost uniformly strong development of attenuate lobes. One specimen was entirely free from lobation of any kind, but it produced no good seeds and therefore it was impossible to test the significance of this unlobed condition. One of the attenuate-lobed plants of this family (0520.196) produced 35 more or less attenuate-lobed offspring, and some which were unlobed, but the latter were so crowded in the seed-pan that stunting might be considered the cause of the suppression of attenuate lobes. Progenies of 4 of these stunted individuals (05196.134, 05196.136, 05196.137, and 05196.147) have since been examined. Two of these 4 families (06136 and 06137) had the wide range of variation usuallj^ found in the related families, and the other 2 (06134 and 06147) contained unlobed or slightly obtuse-lobed specimens. Since these last 2 families were badly crowded, little de- pendence is to be placed upon this behavior. Leaving out of account all the families which may have been injured by crowding or otherwise, in 3 other instances the offspring were uniform and in fair agreement with the parent. All 3 of these belonged to the first of the 2 original families. Seeds of a well-developed obtuse-lobed individual (052.24) sown Januarj^ 16, 1906, produced 100 specimens of uniform aspect, with only well-developed obtuse lobes. An obtuse- lobed sib of the last (052.179) was the parent of 130 plants of the same uniform character, having well-marked sinuses, but acutish, not elongated lobes. One of these (05179.170) produced about 30 obtuse-lobed offspring, but thrips injured them so much that their characterization was unsafe. Another obtuse-lobed plant (052.193), a sib of 052.24 and 052.179 just described, gave a progeny of 125 plants of veryimiform aspect throughout and always with well-developed sinuses and obtuse lobes. In all these exceptional cases there seems to be a consistent behavior in that the parental character dominates the entire progeny, but in each BIOTYPES AND HYBRIDS. 31 such case there were sibs having- the same external characteristics, which behaved in an altog^ether different way when bred. Thus, a sib (057.25) of the attenuate-lobed specimen (057.20) noted above as producing- off- spring- uniformly of the parental type, was, like that plant, attenuate-lobed, but its offspring-, instead of agreeing with the parent, consisted of 1 plant entirely without lobation, 168 with obtuse lobes, 186 with some of the lobes slightly elong-ated, and 415 with attenuate lobes. This result looks very much like a case of Mendelian inheritance if we assume that the parent was a DR, but it is not at all in accord with such an assumption that in like inanner obtuse-lobed sibs of the plants whose entire progenies were characteristically obtuse-lobed as described above have produced offspring- ranging- from the totally unlobed condition to the well-marked attenuate- lobed extreme. For example, one of these obtuse-lobed plants (052.182) whose seeds were sown February 12, 1906, produced a family of 471, of which 337 were observed to have the following composition : 11 were wholly unlobed, 43 were unlobed in the distal half of the leaf, but had small trian- gular lobes in the proximal half, 89 were obtuse-lobed throughout, 102 had some lobes slig-htly elong-ated, and 192 had some lobes strongly attenuate. Besides a few families that were left crowded too long- in the seed-pans to allow of a satisfactory estimation of the foliar characteristics, 15 addi- tional families were reared from parents which had attenuate lobes and all g-ave uniformly the same result, namely, progenies showing the complete rang-e of variation from a wholly unlobed condition to the attenuate condi- tion of the parent; 17 families from obtuse-lobed parents had the same composition, as did also 7 families from unlobed parents. Besides these variations, which can be easily arranged in a simple linear series, there were noted several variations of so definite a character as to lead to the attempt to seg-reg-ate them as distinct forms. Thus one form which appeared in several families {e. g., 0527, 05182, etc.) was charac- terized by a distal unlobed half of the leaf and a proximal half with tri- angular lobes. Another g-roup of specimens (0623) of strikingly uniform appearance had very robust rosettes with leaves broad, obtuse, and entirely unlobed except for the presence, occasionally, of a few shallow triangailar lobes at the very base. If further breeding should confirm the conclusion that these families belong to a single biotype, including within its normal rang-e of fluctuation individuals with obtuse lobes or no lobes at all and others with strongly marked attenuate lobes, this form would be indistingaiishable in some of its phases from B. bp. simplex, and in other phases it would closely re- semble B. bp. tenuis. Only breeding-tests could safely distinguish these several types, and so long- as the practical taxonomist's work consists en- tirely in the classification of individuals as they occur in nature, he would be justified no doubt in refusing- to recognize such elementary species as 32 BURSA BURSA-PASTORIS AND BURSA HEEGERI : of working value, but in the taxonomy of the future they must be reckoned with, because they are the real natural entities with which all students of biology must deal. The old name and the old delimitation of Bursa biirsa- pastoris may remain as the only thing practicable for the amateur collector of plants, but the morphologist, the physiologist, the ecologist, and the evolu- tionist must be more discriminating. It appears to me that systematic botany stands at the parting of the ways. Either it is to be left stranded as a caterer to the amateur or it must adopt the cultural method in lieu of the herbarium method, which has until recently held almost supreme sway among systematists. I am not fully convinced, however, that the variable families here de- scribed belong to a single biotype. The fact that both extremes have in certain cases bred true leads to the question whether there is not some way of accounting for the anomalous behavior of the other families on the ground of hybridization. I believe that such an explanation can be foiind on the principle of latent characters which I have recently discussed else- where (Shull, 1908). At certain times the cultures have become very unhealthy on account of the attacks of thrips and other causes, such as overheating during the summer, crowding in the seed-pans, etc., and the specimens weakened in this way may very well have failed to show their distinctive characters, owing to what I have called "latency due to fluc- tuation," thus destroying completely the ratios by which hybridization phenomena would have been rendered evident. Whether these plants belong to a single biotype of wide variability or to a hybrid group whose nature has been obscured by latency will have to await further cultures. Lotsy (1906) also found that while some forms bred true with very slight fluctuation, the variability of others was consid- erable. If hybridization with latency accounts for the behavior of my variable families it may also perhaps account for a similar condition in Lotsy 's material. Whatever the situation may be in these highly variable cultures, it is perfectly certain that the true-breeding forms are distinct and elementary, as will be more fully demonstrated in the next section. HYBRIDS BETWEEN BIOTYPES OF BURSA BURSA-PASTORIS. Although artificially produced hybrids between the several biotypes above described have been studied as yet only in the first generation, except in the case of one F2 family to be described hereafter (see page 42), I have had under observation 31 hybrid families belonging to a second or later generations. My good fortune in being able to report on these second- generation hybrids is due to the facts that two of the original families were the offspring of hybrid individuals growing in nature and several other individuals in the original cultures were obviously produced by pollination from another biotype. The first of these natural hybrids (040.4) was col- BIOTYPES AND HYBRIDS. 33 lected by J. Marion Shull at Edg-ewood, New Carlisle, Ohio, May 26-28, 1905. This specimen was robust and the lobes of the leaf were lanceolate, with prominent dentations on both proximal and distal margins. The seeds were sown June 26, 1905, and produced a progeny of 284 individuals, among' which there was such a variety of form and aspect as to make them at that time entirely inexplicable, because none of the unit- differences between the forms was then known and there was no possibility of discriminating between minor fluctuations and the distinguishing- char- acteristics of the several elementary species. Efforts were made to arrange these plants into groups that would be strictly homogeneous, and in this way no fewer than 25 categories were established. Even these were not sufficient, and finally a considerable number of individuals that could not be classed with any of these were preserved in order that they might be studied at any subsequent time at which it should be desired to work out their variations and their relationship to each other and to the rest of the family. Fortunately the distinguishing marks of each of these several groups had been noted with sufficient thoroughness that later, when the differen- tiating characteristics of the several elementary species were better under- stood, it was possible to go back and reclassify the several categories in such a manner as to determine approximately the proportions in which the different elementary species were present. The maze of different forms had proved so baffling at every attempt at a satisfactory classification that the result of the redistribution of the several forms in the light of knowl- edge subsequently gained occasioned much surprise. There were among the 284 individuals composing the family, 114 B. bp. heteris, 47 B. dp. tenuis, 53 B. bp. rhomboidea, 16 B. bp. simplex, and 54 which, because of certain deficiencies in the original notes, could not be reclassified with certainty, but nearly all of which were certainly fluctuations of B. bp. heteris. It need only be assumed that of these 54 doubtful individuals 46 were B. bp. heteris, 6 B. bp. tenuis, and 2 B. bp. simplex to make the ratios of these several elementary forms agree exactly with a frequently observed Mendelian ratio, 9:3:3:1. Fortunately most of these doubtful specimens were preserved in the herbarium and were thus available for study. Of 51 thus preserved, 37 were B. bp. heteris, 9 B. bp. tenuis, 1 B. bp. rhomboidea, and 4 of doubtful affinity, these latter probably belonging to the heteris group also, but representing cases of incomplete dominance. The ratio 9:3:3:1 is the normal one for the second generation of typical Mendelian di-hybrids, /. -arding- it. The seeds were sown June 20, 1906, and produced 444 oflfspring- belong-in.o" to the dominant and recessive forms in the ratio 317 : 127 or 2.5 : 1. The dominant group appeared to belong- to two distinct types, with the pri- mary lobes broader and less sharp in the one than in the other in the ratio 110 : 32 or 3.44 : 1. The sig-nificance of the ratio thus formed by this family, 9 : 3 : 4, is well known, as it represents the simplest modification of the ratio for the second g-eneration of a di-hybrid, but without further data in support of this composition for the family, it appears advisable not to consider its sig-nificance at this time. 0514.161 : This was a typical specimen of the recessive form, B. bp. tenuis, of which the pollination was g-uarded. The seeds were sown July 4, 1906, and produced 104 specimens, all of which were like the parent. 0514.162 : This was also an extracted recessive. Guarded seeds were sown July 4, 1906, and g'ave a prog-eny of 15 specimens, all like the parent. 0514.164: A member of the dominant gfroup, B. bp. heteris. Purely fertilized seeds sown July 4, 1906, produced 134 plants. The family was badly damagred by aphis, so that 23 were killed and about 15 more so stunted as to make classification uncertain. The remaining- 93 belonged to B. bp. heteris and B. bp. temiis in the ratio 66 : 27 or 2.44 : 1. In this family the dominant group was not scrutinized with sufficient care to determine whether it was heterog-eneous like 0514.146, but it was considered of a sing-le type. 0514.165 : This was a typical specimen of the dominant type, B. bp. heteris. Purely fertilized seeds were sown Jul}' 4, 1906, and produced 42 specimens of B. bp. heteris and 13 of B. bp. tenuis, the ratio being 3.23 : 1. 0514.182 : Although the notes are defective regarding the character of this plant, the result of the breeding test leaves no doubt that it was a re- cessive. Guarded and purely fertilized seeds were probably sown in July, 1906, though no record can be found to that effect. On August 18, 1906, the entire progeny, consisting^ of 158 specimens, were potted, and without exception these belonged to B. bp. teuids. These 8 families belong-ing to the third or later hybrid g-cneration of B. bp. heteris X tenuis represent the extracted recessive in 5 cases and the heterozygous form in 3. The extracted recessives have bred true to the 42 BURSA BURSA-PASTORIS AND BURSA HEEGERI : characters of B. bp. tenuis without exception in 330 individuals. In the offspring: of 3 heterozyg-ous plants that have been tested, the ratios 2.44 : 1, 2.5 : 1, 3.23 : 1, and 3.44 : 1 have appeared, and these are no doubt as near 3 : 1 as the smallness of the families should have led us to expect. 056.130: Besides these 2 orig-inal hybrid families, it will be recalled that certain unexpected individuals oi B. bp. rhoviboidea in one of the original families of B. bp. simplex (056) were supposed to be due to chance crosses (see p. 26). Seeds of one of these (056.130) were sown Jiine 6, 1906, and produced 217 B. bp. rhomboidea and 72 B. bp. simplex, that is, in the ratio of 3.01 : 1, thus showing the assumption that they were of hybrid origin to be correct. (See plate 2). In addition to the hybrids brought in from nature and their self -fertilized offspring- as represented by the 31 families which are described above, 7 first-generation hybrid families between different biotypes of Bursa bursa- pastoris have been produced by artificial crossing. Only one F2 family has been reared from these to the present time. 0515.93 : The mother of this cross was a specimen belonging to the second original family of B. bp. hcteris (0515) described above (see p. 14). This plant was castrated and pollinated with pollen from a specimen of the first original culture of B. bp. tennis (0519, see p. 22). The seeds were sown May 2, 1906, and produced 222 plants, all resembling the mother and possessing the essential features of B. bp. heteris. 0693.203 : Self- fertilized seeds of 3 guarded specimens belonging to the last-described family were sown together October 12, 1906, and gave a progeny of 111 specimens, 84 of which were typical B. bp. heteris and 27 B. bp. tenuis, or in the ratio 3.1 : 1. (See plate 1.) As no other artificially-produced hybrids between biot5q:)es of B. bursa- pastoris have been studied beyond the first generation and only facts bear- ing upon the question of dominance can be derived from these cultures as yet, it seems best to postpone their discussion until the second generation has been studied. This is the more important, since the specimens chosen for crossing in a number of instances belonged to families whose relations to the 4 biotypes which have been involved in the above-described natural hybrids are still in doubt. The occurrence of supposed hybrids among the biotypes of Bursa bursa- pastoris has been noted by Almquist (1907, pp. 22 and 88-89), who also refers to descriptions of similar cases by Von Borbas and by Grenier. The actual hybrid character of the plants mentioned is very doubtful, however. The assumption of their hybridity is based wholly upon the facts of inter- mediacy and sterility, together with the vegetative vigor and longevity which are certainly correlated with sterility. I have occasionally observ^ed such sterile or nearly sterile plants in progenies produced from seeds fully BIOTYPES AND HYBRIDS. 43 guarded, so that hybridity was excluded with such care as is possible. Intermediacy in the size of flowers is not at all strang-e in the case described by Almquist, for the culture in which the supposed hybrids occurred was that of a very large-flowered form. Intermediacy in such a case simply means a reduction in the size of the flowers. In all of my hybrids there has been no apparent decrease in the number or viability of seeds produced, and it will be recalled that there is only one clear case of intermediacy, namely, in the hybrids between B. bp. heteris and B. bp. rhojjtboidea, in which it is due to the incomplete dominance of the former. Perhaps in other cases also dominance is not quite complete, but it is so nearly so that it is impossible to distinguish certainly between the heterozygote and one of its parents. HISTORY OF BURSA HEEGERI. Eleven years ago Professor Heeger found some specimens of a crucifer- ous plant growing in the market-place at Landau, Germany, which he could not identify. In general habit these plants resembled the almost cosmo- politan species Bursa bursa-pastoris . They differed from the latter species, however, in having the seed-capsules elliptical in longitudinal section and circular in cross-section instead of flat and triangular or obcordate, as is characteristic of Bursa bursa-pastoris . The specimens were submitted to Solms-Laubach, who was inclined at first to refer them to the genus Camelina, which is characterized by nearly spherical capsules, and Professor Ascherson, to whom he showed them, was of the same opinion. Cultures made by Solms-Laubach from seeds secured from Professor Heeger soon indicated, however, the near relationship of the new form to Bursa bursa-pastoris, when in 1898 several apparent rever- sions to the capsule-form of B. bursa-pastoris were noted. Solms-Laubach (1900) published an account of the new form, assigning to it the name Capsella kee^eri, which becomes, according to the rule of priority, Bursa hee^eri (Solms-Laubach). This very distinct species of Bursa has attracted considerable attention, for the reason that its occurrence as a component of the flora of a region so well known systematically has left little doubt of its very recent origin from B. bursa-pastoris by mutation, and it is mentioned by De Vries (1901, pp. 477-478; 1905, pp. 582-584) as an instance of mutation in nature. Shortly after the publication of the original account. Bursa heeger i disap- peared from the type locality at Landau, owing to the destruction of its habitat by covering the market-place with gravel, and it has been reported from nature only once since that time, though it has been widely grown in botanical gardens. The second report of the discovery of Bursa heegeri in nature was made by Laubert (1905), who found it along the Dahlem turnpike in 1905, but 44 BURSA BURSA-PASTORIS AND BUR.SA HEEGERI : here the Hkehhood of a new origfin is certainly exceeded by the probabihty that a seed was carried to this spot by some agency from a nearby culture, for it had been grown for several years at Dahlem in an unprotected bed several hundred meters from the place in which Laubert discovered it. Hus (1908) takes the alternative view, however, and considers this a case of repeated mutation.* Laubert points out that in addition to the capsule character noted by Solms-Laubach, there are other characters of the stem and inflorescence which serve to distinguish B. heegeri from its supposed parent and which would suggest a more distant relationship with that form than had been supposed; but, on the other hand, he found that in both species there occur frequent instances of abnormal pistillate but sterile flowers in the lower portion of the flower-stem, and he takes this fact, together with the occurrence of capsules occasionally simulating those of B. bursa-pastoris, formed when B. heegeri is attacked by Albugo and Pero- nospora, as additional proofs that B. heegeri is a derivative from B. bursa- pastoris. Of the manner of its origination from B. bursa-pastoris nothing is known, of course, but Potonie (1906) suggests that it is a reversion induced by some pathological condition. More recently Noll (1907) has investigated some plants resembling B. heegeri, which had already been found by Melsheimer in 1882 in himdreds in a field of Dattenberger Flur and again in 1884 on a height at Linz. Mels- heimert considered these plants hybrids, but could not suggest the prob- able parents, while Komicke and Wirtgent stated that they are doubtless identical with Bursa heegeri. Specimens of Melsheimer' s plants were placed in Petry's herbarium bearing the label "Capsella bursa-pastoris forma caps, ovatis. ' ' Noll received this material from Petry, together with living specimens collected by the latter at Didenhofen, Metz, Hagendin- gen, and Kreuznach. A careful comparison of the anatomical features of these plants with those of Bursa heegeri and B. bursa-pastoris led Noll to the conclusion that the Melsheimer plants are not Bursa heegeri, but a sterile form of B. bursa-pastoris, to which he gives the name Capsella pseudo-heegeri. The finding of these plants in considerable numbers tends to weaken the argument that the discovery of B. heegeri in a region so well known proves it to be a recent mutation. Perhaps Bursa heegeri will yet be discovered in some abundance in some locality where it has hitherto escaped notice. *I have called the attention of Dr. Hus to this matter and he concedes in a letter that my explanation of the occurrence of B. heegeri in nature at Dahlem is probably the correct one. t Mentioned by Noll, but not verified by me. BIOTYPES AND HYBRIDS. 45 HYBRIDS BETWEEN BURSA BURSA-PASTORIS AND BURSA HEEGERI. The prominent part which mutation may have taken in the production of new species makes it of great interest to know, in each case, just what will be the result when the supposed or the demonstrated mvitant is self- fertilized and when it is crossed with the parental form; for its behavior in these two cases is the first important factor in determining the power of the mutant to maintain itself at the time of its origin and its capacity to give rise to a successful series of genetically related individuals belonging to an independent type. The result of self -fertilization was investigated by Sol ms-Laubach before the publication of his original account, and he showed that the characters that differentiate B. hcegeri from B. biirsa-pastoris are fully heritable in a self -fertilized line. As we have seen, the Bursas normally self -fertilize to a predominant extent, and this habit, coupled with a vigorous constitution, would seem to constitute all the factors necessary to successful mainte- nance. I have now determined what will be the result of intercrossing B. heegcri and B. biirsa-pastoris. The aspects of Bursa heegeri and B . bursa-pastoris as they appeared in my cultures were so different that at the first I was skeptical concerning the near relationship which has been assumed to exist between them. Bursa heegeri was much more vigorous than B. bursa-pastoris, and the dif- ferences observed in the inflorescence by Laubert (1905) were strikingly apparent, the pedicels of the capsules being shorter, more crowded on the rachis, and diverging from the latter at a wider angle (fig. 21). While the leaves of the rosette of Bursa heegeri are of the heteris type, having the primary and the rounded secondary lobes readily distinguishable, the sinus which sets off the latter is comparatively shallow, and in consequence the secondary lobe appears low and less well-marked. The primary lobe is usually broader and less strongly attenuate than in B. bursa-pastoris heteris. Despite these considerable differences, however, it was found that B. heegeri may be crossed with B. bursa-pastoris with perfect ease in either direction and without any apparent decrease in fertility, though I crossed it with B. bp. simplex, the biotype of the latter species which is most unlike B. heegeri. The families of pure-bred B. heegeri and its hybrids with B. bp. simplex may be briefly considered under the pedigree-numbers used during their culture : 040.9 : Seeds of Bursa heegeri received through Dr. D. T. ]\IacDougal from Professor Solms-Laubach were sown July 31, 1905, and produced 26 plants, all of which agreed with the above description, the fluctuating variations being extremely slight. The unguarded seeds of one of these 46 BURSA BURSA-PASTORI.S AND BUR.SA HEEGERI Fig. 21. — Bursa heegeri (Solms-Laubach). From photographs taken at three different stages of development. BIOTYPES AND HYBRIDS. 47 (059.56) were sown November 1, 1906, and produced 24 plants, all agree- ing- perfectly with the original type as described (fig. 22). These became diseased later, however, and produced no seed. 056.88 : This plant was considered a typical specimen of B. bp. simplex, though a little more vigorous and broader-leafed than usual. It was care- fully castrated and pollinated with pollen from a i)lant belonging to my first culture of B. heegeri (059). The seeds were sown April 25, 1906, and Fig. 22. — Bursa heegeri. Second controlled generation. produced 108 offspring, all resembling B. heegeri more closely than B. hp. simplex, but they differed markedly from the former because of the imper- fect-dominance of the heteris characteristics. On this account these plants had some of the characteristics of B. bp. rhomboidea and could be properly described as intermediate between B. bp. rhomboidea and B. bp. heteris (fig. 23). A few of these died without seeding, but all that came to maturity had the triangular capsules typical of Bursa bursa-pastoris . One family was raised from unguarded seeds of one of these plants (0688.212), as described below. 48 BURSA BURSA-PASTORIS AND BURSA HEEGERI : 059.89 : This typical specimen of Bursa heegeri^zs, castrated and polli- nated with pollen from a specimen of B. bp. simplex which differed from the usual condition of that biotype in having the apex of the leaves taper- ing- and acutish. This condition has been shown to be merely a fluctua- tion (see p. 26), so that the plant used in this cross is to be considered a Fig. 23. — Bursa bursa-pastoris simplex X heegeri, Fp Incomplete domi- nance of the heteris characteristics produces a form resembling B. bp. rhomboidea, but mostly with longer, sharper primary lobes. typical specimen of B. bp. simplex. The pollen-parent of this family was a sib of the pistil-parent of the family last described, and the two speci- mens of B. heegeri involved in these two families were likewise sibs, so that these two crosses were essentially reciprocal. The seeds were sown April 25, 1906, and produced 23 offspring, which were intermediate be- tween B. bp. heteris and B. bp. rhomboidea as to the rosette characters, and in which they were indistinguishable from the reciprocal hybrid family, BIOTYPES AND HYBRIDS. 49 0688. Most of these plants produced fruit, and in all cases this was flat and triang-ular to obcordate, like that of B. biirsa-pastoris . Ung-uarded seeds of 3 of these and guarded seeds of 1 were used for the production of second-g-eneration families, 1 as 0689.196 and 3 collected tog-ether as 0689.197. The descriptions of the second-generation families follow: 0889. 198 : Seeds of this plant, which had been carefully gfuarded against cross-pollination, were sown October 12, 1906, and 217 plants were raised, of which 188 lived to produce seed. These had the following composition: 98 were B. bnrsa-pastoris keteris, 36 B. bp. tenuis, 2>2 B. hp. rhomboidea, 13 B. bp. simplex, 5 B. heegeri heteris, 1 B. h. tenuis, 2 B. h. rhomboidea, and \ B. h. simplex, giving, so far as the rosette-characters are concerned, a very close agreement with the ratio 9:3:3:1, but in the form of capsule showing a very great preponderance of the bursa-pastoris type. 0689.197: vSeeds of 3 unguarded sibs of the parent of the last family were sown October 15, 1906, under this number. Of 2,014 offspring, 1,815 came to maturity, and were recorded as having the following composition : 1,032 were B. bursa-pastoris Jieteris, 331 B. bp. tenuis, 2>02B. bp. rhomboidea , 78 B. bp. simplex, 45 B. heegeri heteris, 13 B. h. tenuis, 13 B. h. rhomboidea, and 1 B. h. simplex (plates 3 and 4). Again there is a close agreement in the leaf -characters with the typical dihybrid ratio, 9:3:3:1, and a notable deficiency in the occurrence of the heegeri type of capsule. 0688.212 : Seeds of this unguarded plant, which was a reciprocal of the parents of the last two families described, were sown October 12, 1906, and produced a large progeny, of which 621 unselected plants were potted for study and the rest discarded. Of these 621 plants, 537 reached maturity and were classified thus : 317 B. bursa-pastoris heteris, 102 B. bp. tenuis, 61 B.bp. rhomboidea, 21 B. bp. simplex, 19 B. heegeri heteris, 7 B. h. tenuis, and 4 /?. //. rhomboidea, no B. h. simplexheing observed. The same general relations of the rosette-characters and capsule -characters are obvious here as appeared in the other two Fj families described, but there is not quite as close agreement with the ratio 9 : 3 : 3 : 1 as in the other families, prob- ably because this family became somewhat diseased and the distinguishing of the several types of rosette became consequently more difficult. Reviewing the results of crossing Biirsa bursa-pastoris simplex and B. heegeri, it is seen that the Fi hybrids are essentially uniform, no matter in which direction the cross is made, and that the rosette in either case is of the heteris-rhomboidea type, owing to the incomplete dominance of heteris, while the capsule is always of the bursa-pastoris type. In Fa there appear the 4 types of rosette already described, in combination with each tyj^e of capsule. The rosette presented many instances of the best-developed examples of the 4 described forms, particularly interesting being the fact 50 BURSA BURSA-PASTORIS AND BURSA HEEGERI that much better-developed hetcris rosettes occur in the Fj than are seen in pure-bred B. heegeri, though these heteris characteristics must have come directly from the heegeri side of the cross. The numerical results of these crosses may be tabulated thus : Btirsa bp. simplex {abC) X Bursa h. heteris (ABc) (056) (059) Bursa bp. heieris-rhomboidea {ABCabc) (0688 and 0689) Bursa-pastoris series (C): heteris tenuis rhomboidea simplex Heegeri stv'xts (c): heteris .'.... tenuis rhomboidea simplex 06196 98 36 32 13 5 I 2 I 19.9: 06197 1032 331 302 78 45 13 13 I 24.2 06212 Total. 317 102 67 21 19 7 4 o 1447 469 401 112 69 21 2 Expected. 1368 456 456 152 63 21 21 7 16.9: I 21.9: I 3-0 I In this table the pedigfree-numbers 06196 and 06197 represent families in which B. heegeri was the mother, while 06212 is the reciprocal cross. It will be seen that there is no essential difference between these two crosses. In the rosette-characters these numbers are very close to the Mendelian ratios, amongf both those having bursa-pastoris capsules and those having heegeri capsules, showing b}^ these facts that the heegejd rosette has the same allelomorphic composition as the rosette of B. bursa- pastoris heteris, and that these rosette-characters are not coupled in any but a chance way with the form of the capsule. This independence of characters is thus a matter of great importance in the production of new elementary forms ; for before this cross was made there existed, so far as is known, but one elementary species of B. heegeri, while otit of the cross came 4 distinct elementary forms of this species. While there is a perfect agreement with Mendelian ratios in rosette- characters, the capsules give a surprising departure. The capsule-form is perfectly alternative; there has never appeared any intermediate condition in the plants in my cultures, and the "reversions" observed by Solms- Laubach and Laubert were pathological. As the bursa-pastoris type is dominant, simple Mendelian expectation would require the appearance of 1 B. heegeri in every 4 Fj individuals. Out of 2,540 plants of the F2 gen- eration observed, only 111 were B. heegeri, or approximately 1 in 23. That this result should be consistently given in 3 different pedigrees represent- ing reciprocal crosses adds greatly to the weight that is to be attached to BIOTYPES AND HYBRIDS. 51 it. The greatest frequency in any pedigree was 1 in nearly 18, and the least frequency was 1 in 25.* I take the fact that the reciprocal crosses give similar ratios to indicate that the heegeri type of capsule is dependent upon something carried by the germ-cells, and can not be a pathological condition transmissible from mother to offspring somatically, a possibility which might account for apparent heritability of characters in a self -fertilized line, but could not well account for equal results in reciprocal crosses. Normal Mendelian phenomena are believed to rest pretty securely on the method of formation of the chromosomes during- the reduction division, but no scheme of behavior occurs to ine which would result in the i:)roduction of a heegeri homozygote in only 1 individual in 23. It is conceivable that the union of heegeri germ-cells in the hybrids forms a less successful combination than that into which the B. biirsa-pastoris determiner enters, and that therefore fewer successful zygotes are formed by such unions, and it is also conceivable that only a small percentage of the B. /?(?£'^^r/ succeeded in reaching sexual maturity, since in each of these families a considerable number of individuals failed to fruit, but the assump- tion that every individual that failed to seed was a B. heegeri would not nearly bring that form up to one-fourth of the entire progeny. There was no evidence in my cultures that B. heegeri is in any way inferior to B. bursa-pastoris or that it is any less likely to mature; neither have I observed any indication of the material lessening of fertility which would obtain if almost all of the heegeri homozygotes should fail in the initial stages of development. *Dr. Correns tells me that he also made the cross between Bursa bitrsa-pastoris and B. heegeri several years ago and likewise found a deficiency in the number of specimens having the heegeri type of capsule, but as his numbers were small he considered the deficiency due alone to the inadequate numbers. 52 BUR.SA BURSA-PASTORIS AND BURSA HEEGERI : DISCUSSION OF RESULTS. The occurrence of elementarj^ species within the limits of recognized systematic or Linnean species is undoubtedly very general, as De Vries has maintained, and much of the ordinary conception as to the variability of certain species is attributable to this fact. Thus in the specific case with which we are dealing- here. Bursa is recog'nized by all taxonomists as exceedingly variable, but each single biotype of Bursa is much less vari- able than the Linnean species taken as a whole, for when grown under favorable conditions there are certain characteristics which are found in every individual of the given biotype which are not present in the members of other biotypes. All the apparent evidence for the permanent change of species through selection finds a ready explanation on the assumption that the selection has merely eliminated certain biotypes from the original mixture with which the selection started, thus leaving the theory that fluctuations are inherited or are capable of leading either directly or indirectly to the modi- fication of any biotype wholly tinsupported. This is not to say, of course, that such modification is impossible or that it does not take place, but merely that such a proposition must rest ^^pon experience still to be gained. There is some variation among the members of the single biotype. This variation is of the fluctuating kind by defiiiition. The most usual varia- tions of this kind are those which result from crowding, shading, poor soil, drought, the attacks of insects, or other conditions which decrease the vigor of the plants, the effect being to arrest differentiation in more or less juve- nile stages of development. This feature has been very troublesome at times in my cultures, since it is impossible under such circumstances to determine by inspection to which biotype a given specimen belongs. The breeding-test is the only method by which such a determination can be made and when many specimens have their distinguishing characteristics rendered latent in this way the labor, time, and patience required for com- plete classification become unduly increased. This suppression of characteristics through fluctuation also stands in the way of the classification of specimens observed in nature in various habi- tats where they have gTown under different conditions which are in a large degree unknown, and on this accotint there can be no question as to the advisability of retaining the Linnean names in practical taxonomy for the designation of such complex groups of biotypes. Workers in all other botanical fields inust bear in mind, however, that conclusions reached with one biotype inay not hold in some other biotype of the same Linnean species. The demonstration of the elementarj- character of these biotypes of Bursa is made complete by the fact that they Mendelize on being crossed, for such BIOTYPES AND HYBRIDS. 53 behavior can only rest upon the presence of an internal factor, of whatever nature, capable of independent movement at the time of the reduction- division in the formation of the gferm-cells. It may be assumed, perhaps, that this factor or determiner is incapable of division at the time of the reduction-division ; but however it is explained, the result is the pro- duction of certain individuals (homozyg'otes) which produce g-crm-cclls all of one kind with respect to a particular character and which must there- fore breed true with respect to that character, and other individuals (het- erozyg"otes) which produce germ-cells of two kinds with respect to the same character and which consequently can not breed true. Individuals which possess the same characteristics and which arc homozyg"ous with respect to all of these characteristics, collectively form a biotype ; hence the state- ment that the Mendelian inheritance of the several forms of Bursa demon- strate beyond a possible question that they are distinct biotypes. I will not be understood, however, to imply that biotypes must Mendelize on being- crossed, for many are known which do not, as, for example, most of the various biotypes of Oenothera. The importance of Mendel's discoveries for our conception of the signifi- cance of different kinds of variations in the evolution of plants and animals is now generally recognized. It is seen that in the presence of Mendelian inheritance the ' ' swamping " of a new character by crossing with the parent form does not take place, and that the very kind of variation which was supposed to be swamped by crossing is just the kind which is most certain to be preserved. The new form may completely disappear when crossed, but it is only hidden momentarily, not destroyed. Thus when Bursa /leegeri Qvoss&s with its parent Bursa bursa-pastoris, all of the offspring are indistingniishable from the latter, but in the next generation a portion of the offspring are just as typical B. heegeri as the original specimen, and no transitional forms occur. If the dominance of the triangular capsules were incomplete there would be some transitional stages, but this would not modify the situation in any essential manner, as there would still be just as many typical homozygotes as before which would breed true when- ever they chanced to be fertilized by their own kind, and if fertilized again by the parent form they would form heterozygotes which would produce as large proportion of typical B. heegeri as was produced by the heterozy- gotes of the previous generations. I have shown elsewhere (Shull, 1907«) that this capacity of a mutant to disappear upon crossing with its parent may be a great advantage in the struggle for existence under certain conditions, and if we grant that evolu- tion is in the main retrogressive (Shull, 1907<^), the new forms will be generally recessive to the parent and will thus be in position to profit by whatever advantage recessiveness gives. From the data I have as to the geographic distribution of the several biotypes of B. bursa-pastoris it 54 BURSA BURSA-PASTORIS AND BURSA HEEGERI : appears that B. bp. heteris is much more grenerally distributed than any of the others , and this favors the view that this is the original type from which B. bp. rJiomboidea^ tenuis, and simplex ^nqxq, derived by retrog-ressive muta- tion. Recessiveness of the last three to the first is probably rarely if ever any advantage to them, however, since all of these forms appear to be about equally adapted to the range of habitats in which they grow. Mendelian inheritance also has an important bearing upon the distribu- tion of the various biotypes of a species; for the transportation of a single hybrid seed may carr}^ all of the biotypes which are related to one another in the Mendelian way, however many there may be,* and two pure-bred seeds landed in the same vicinity may lead to the same result. Thus any seed of B. bp. heteris which has been produced by pollination with pollen from B. bp. simplex, or vice versa, or any seed of B. bp. rhomboidea that has resulted from pollination with pollen from B. bp. temiis, or vice versa, will give rise to a progeny in the F2 which will include all four of these biotypes, and self -evidently two pure-bred seeds representing either of these two pairs of elementary species will carry the capacity to produce the same four types in the third generation from the time they find themselves in juxtaposition in a new locality. The same principle would hold if there were 3 pairs of Mendelian char- acters involved, but then 8 biotypes might be carried by a single hybrid seed. This situation would be realized by the material dealt with in this paper if a single seed of B. bp. simplex pollinated by B. heegeri, or vice versa, were taken to a new locality. Transcau (1907) has pointed out how on the same principle any number of biotypes might be introduced by means of a single pollen grain into a new locality where a single biotype had existed before, and to which heavy seeds might find much greater diffi- culty in being transported. Still another important effect of Mendelian inheritance in the promotion of organic evolution is brought out by my crosses between B. bursa-pastoris and B. heegeri, namely, the production of parallel series of biotypes in nearly related species. Up to the time this cross was made B. heegeri y^^s known only in the heteris form, but among the hybrid offspring were 4 distinct pure-breeding biotypes of B. heegeri. It is thus seen that the single mutation by which B. heegeri originated from B. bursa-pastoris doubled the number of possible biotypes of Bursa in the world, provided all such other biotypes behave as do the four under discussion in this paper. It may be added that the facts here shown that the rosette of B. heegeri represents the same Mendelian units that are present in B. bursa-pastoris and that there is only a single fundamental difference between these two species is the best possible proof of the origin of B. heegeri from B. bursa- *Except in the presence of "spurious allelomorphism." BIOTYPES AND HYBRIDS. 55 pastoris, despite the considerable differences of general aspect both of foliag-e and of inflorescence. It is possible that these differences of aspect may rest upon the presence of still other units which have not been taken into account in these studies, but if such should prove to be the case, the gen- eral conclusions must be the same. I wish to acknowledge my appreciation of the facilities which have been placed at my disposal by the Station for Experimental Evolution of the Carnegie Instittition of Washington, and of the faithfulness of my assistants, who have greatly aided me with the technical side of this investigation. My thanks are due to my brother, J. Marion Shull, dendrological artist of the United States Forest Service, for the drawings reproduced in figs. 1 to 23. I particularly desire to acknowledge my indebtedness to Dr. E.N. Transeau, in whose charge the cultures of B. bursa-pastoris X heegeri were left during my absence in California in the spring of 1907. Much of the work of grouping the F, into the appropriate classes was done by him. SUMMARY. (1) Bursa {Capsella) biirsa-pastoris is a composite species made up of at least 4 and possibly many distinct elementary species or biotypes ; 4 of these are described tmder the names Bursa bursa-pastoris heteris, B. bp. tetiuis, B. bp. rhomboidea, and B. bp. simplex. Except for the sup- pression of characteristics due to bad treatment, the fluctuations of these forms are slight. (2) These 4 biotypes cross readily, giving in each case a uniform Fi and typical Mendelian splitting in Fj . They are differentiated from each other by 2 unit-characters, namely, elongated primary lobes of the climax-leaves and a rounded secondary lobe in the distal axil of the primary lobes. When B. bp. heteris is crossed with B. bp. simplex, and when B. bp. tenuis is crossed with B. bp. rJwmboidea, the Fi is intermediate between B. bp. heteris and B. bp. rhomboidea because of the imperfect dominance of B. bp. heteris, and the Fj consists of heteris, tenuis, rhomboidea, and simplex in the ratio 9:3:3:1. (3) The sterile or nearly sterile plants supposed by Almquist to be hybrids were probably not hybrids, as similar forms were noted in guarded cultures, and in all of my hybrids no decrease in fertility was apparent. (4) Bursa heegeri differs from B. bursa-pastoris in the aspect of its rosette and inflorescence, but most notably in the form of the capsule. Its climax-leaves are of the same general type as those of B. bp. heteris. (5) Bursa heegeri may be readily crossed with B. bursa-pastoris. When crossed with B. bp. simplex, an Fi is obtained which more nearly resem- bles B. heegeri in rosette characters, but has the capsule-form of B. bursa- pastoris. In F, there appear the 4 forms of rosette already mentioned, in 56 BURSA BURSA -PASTORIS AND BURSA HEEGERI : the ratio 9:3:3:1, in association with each type of capsule. The ratio of the bursa-pastoris type of capsule to the heegeri type is about 22 : 1. (6) Bursa heegeri possesses the same unit -characteristics in the rosette as B. hp. /leteris, which serves to further confinn its direct derivation from that species. (7) The fact that the various forms of Bursa show Mendelian inherit- ance on crossing' is conclusive proof that they are distinct biotypes, not- withstanding: the fact that their disting^uishing characteristics are readily rendered latent by fluctuation. (8) In the presence of Mendelian inheritance the swamping of a new characteristic does not take place, and the kind of variation which has been supposed to be swamped by crossing' is just the kind that is most certain to be preserved. (9) The recessiveness of a newly arisen form is to be considered an advantage in its struggle with the parent, whenever the former is in any way less adapted to its environment than the latter. This principle appar- ently has no bearing upon these biotypes of Bursa, however, as all appear to be equally adapted to the range of habitats in which they live. (10) Bursa bursa-pastoris hetcris appears to have a more general distri- bution than any of the other forms. This is probably the primitive type from which the other biotypes have been derived by retrogressive mutation. (11) Mendelian inheritance also assists in the distribution of the vari- ous biotypes, since a single hybrid seed or two pure-bred seeds may carry into a new locality as many distinct biotypes as are related to each other by Mendelian characters. A single pollen-grain may do the same provided one biotype is already present in the locality in question. (12) Crosses between nearly related species may g'ive rise to parallel series of biotypes in the two species concerned. Before my crosses between Bursa heegeri and B. bursa-pastoris were made, the former was known only in 1 form, but out of the crosses came 4 distinct biotypes of B. heegeri, corresponding with the 4 biotypes which have been demonstrated in B. b u rsa-pastoris . (13) Each mutation which results in the appearance of a new Mende- lian unit-character doubles the possible number of Mendelian biotypes belonging' to the species in question, however numerous they may be already, except as limited by "spurious allelomorphism." Station for Experimental Evolution, July 11, 1908. BIOTYPES AND HYBRIDS. 57 LITERATURE CITED. AlxMquist, £. 1907. Studien iiber die Capsella biirsapasioris ( L.). Acta Horti Bergiani, 4: No. 6, pp. 92, figs 66, May 15, 1907. De Vries, H. 1901. Die Mutationstheorie. I. Die Entstehung der Arten durcli Mutation, pp. xii + 64S, pis. 8, figs. 181, 1901. Leipzig. (See pp. 477-478.) 1905. Species and varieties: their origin by mutation, pp. xviii +847, 1905. Chicago. (See pp. 582-5S4.) Hus, H. 1908. Fasciations of known causation. Amer. Nat., 42: 81-97, 2 figs., Feb., 1908. Laubert, R. 1905. Notizen iiber Capsella Jieegeri Solms. Verh. Bot. Vereins Provinz Branden- burg, 47: 197-199, 4 figs., 1905- LOTSY, J. p. 1906. Vorlesungen iiber Deszendenztheorien mit besonderer Beriicksichtigung der botanischen Seite der Frage. Erster Teil, pp. 384, pis. 2, figs. 124, 1906, Jena. (See pp. 179-181.) Noll, F. 1907. Ueber eine Heegeri-ahnliche Form der Capsella bursa-pastoris Moench. Sitz- ungsber. Niederrhein. Gesells. Nat.-u. Heilkunde, 5 pp., 1907, Bonn. POTONIE, H. 1906. Capsella hee^eri eine pathologische Erscheinung mit atavistischen Momen- ten? Naturwis. Wochenschr., 21 (n. s., 5): 788-791, 2 figs., Dec. 9, 1906. Shull, G. H. 1907a. Elementary species and hybrids of Bursa. Science, n. s., 25: 590-591, Apr. 12, 1907. i()o-jb. The significance of latent characters. Science, n. s., 25: 792-794, May 17, 1907. 1908. A new Mendelian ratio and several types of latency. Amer. Nat., 42: 433- 451, July, 190S. Solms-Laubach, H. j tt 1900. Cruciferienstudien. L Capsella heegeri ^o\m?, erne neuentstandene term der deutschen Flora. Bot. Zeit., 58: 167-190, pi. vii, 1900. Transeau E. N. 1907. Hybridization a factor in migration and competition. Science, n. s., 25: 269-270, Feb. 15, 1907. Shull Plate I Fig. I. — Bursa bursa-pastoris heteris. Dominant form from the F2 of an artificial cross of B. bp. heteris X tenuis. The grandparents were sibs of the plants shown in text-figures 4 and 9. Fig. 2. — Bursa bursa-pastoris tenuis. Recessive form from the same family as the plant shown in figure 1 . Shull Plate 2 ^^p^ 06130 06130 F'g- I- — Bursa bursa-pastoris rhomboidea. Dominant form in the F2 of a liybrid family repre- senting the natural cross, B. bp. simplex X rhomboidea. The parent was a sib of the plant shown in text-figure 20 and had the same characteristics. Fig. 2. — Bursa bursa-pastoris simplex. Recessive form in the same hybrid family whose dominant form is represented by figure 1. » Shull Plate 3 06197 Fig. 1. Bursa heegeri >< bursa- pasloris simplex. A heteris rosette in the F->. Fig. 2. Bursa heegeri X bursa- pastoris simplex. A tenuis rosette in the F2. Shull Plate 4 06197 Fig. I. Bursa heegeri X bursa-pastoris simplex. A rhomboidea rosette in the F-... Fig. 2. Bursa heegeri X bursa-pastoris simplex. A simplex rosette in the F2. Library u