CIHM Microfiche Series (Monographs) ICMH Collection de microfiches (monographies) [Ml Canadian Inttituta for Historical Microraproductiona / Inatitut Canadian da microraproductiona hiatoriquaa Technical and Bibliographic Notes / Notes techniques et bibliographiques Th« institute has attempted to obtain the best original copy available for filming. Features of this copy which may be bibliographicallv unique, which may alter any of the Images in the reproduction, or which may significantly change the usual method of filming are checked below. □ Coloured covers / Couverture de couleur □ Covers damaged / Couverture endommag^e □ Covers restored and/or laminated / Couverture restaur^e et/ou pellicul^e I Cover title missing / Le titre de couverture manque I Coloured maps / Cartes g^raphiques en couleur □ Coloured ink (i.e. other than blue or black) / Encre de couleur (l.e. autre que bleue ou noire) □ Coloured plates and/or illustrations / Planches et/ou illustrations en couleur □ Bound with other material / Relld avec d'autres documents □ Only edition available / Seuld ^ditton disponlble □ Tight binding may cause shadows or distortion along Interior margin / 1^ reliure serrte peut causer de I'ombre ou de la distorslon le long de la marge int^rieure. I I Blank leaves added during restorations may appear ' — ' within the text. Whenever possible, these have been omitted from filming / Use peut que certaines pages blanches ajout^es lors d'une restauration apparaissent dans le texte, mais, lorsque cela ^tait possible, ces pages n'ont pas 6\6 film^es. □ Additional comments / Commentalres suppl^mentaires: L'Institut a microfilm^ le meilleur exemplaire qu'H lul a i\6 possible de se procurer. Les details de cet exem- plaire qui sont peut-§tre uniques du point de vue bibN- ographiique, qui peuvent nrxxlifier une image reproduiie, ou qui peuvent exiger une modification dans la mMho- de normals de filmage sont indiqu^s ci-dessous. I I Coloured pages /Pageade couleur I I Pages damaged/ Pages endommagtes Pages restored and/or laminated / Pages restaurtes et/ou peincultes Pages discoloured, stained or foxed / Pages dicolordes, tachetdes ou pk^utes I I Pages detached / Pages d^tach^es Showthrough/ Transparence Quality of print varies / Quality in^gale de rimpresston Includes supplementary material / Comprend du materiel suppMmentaIra □ Pages wholly or partially obscured by en^ta slips, tissues, etc., have been refilmed to ensure the best possible image / Les pages totalement ou partlellement obscurcies par un feuillet d'errata. une peture, etc., ont 6\6 film^es k nouveau de fa9on k obtenir la meilleure innage possible. □ Opposing pages with varying colouration or discolourations are filmed twice to ensure the best possible Image / Les pages s'opposant ayant des colorations variables ou des decolorations sont film^es deux fois afin d'obtenir la meilleure image, possible. This item it filmed at tha reduction ratio checked below / C* docunwnt ut (iimi au taux tfa rMuetion indiqui ci^atoua. 10x 14x 18x 22x 26x 30x I I I I I I I I I I I I I I I I / 1 I I I I ~T~ 12x 16x 20x 24x 28x 32x Th« copy filmad h«r« hat baan raproduead thanka to tha ganaroaity af : MacOdrum Library Carleton Unimiity Tha imagaa appaaring hara ara tha bast quality poMibia considaring tha condition and lagibility of tha originai copy and in icaaping with tha filming contract spacif icationa. Original eopiaa in printad papar covars ara filmad baginnbig with tha front covar and anding on tha last paga with a printad or illustratad impraa- sion, or tha bacic covar whan appropriata. All other original copias ara filmad beginning on tha first paga with a printad or illustratad impraa* slon. and anding on tha laat paga with a printad or iiluatratad impraaaion. The last recorded frame on each microfiche shall contain the symbol — » (meaning "CON- TINUED"), or tha symbol ▼ (meaning "END"), whichavar appHaa. Maps, plates, charts, etc., may ba fHmad at different reduction ratios. Those too large to be entirely included in one exposure are filmed beginning in the upper left hand worner, left to right and top to bottom, aa many framea aa required. The following diagrams iHuatrata the method: L'axamplaira fUmt fut raproduit grica * la g«n4ro8iti da: MaoOdrum Library Carlston Uflivsnity Las images suivantes ont M reproduites avec le plus grand soin, compta tenu de la condition at da la nattat« da l'axamplaira filmi. at an conform it* avec lee conditions du contrst de flimaga. Laa axamplairas origineux dont la couverture en papier est imprimte sont film*s en commenpant par la premier plat at en terminant soit par la derniAre page qui comporta una empreinte d'Impression ou dHHuatration. soit par la second piat. saion la cas. Tous las autrea exemplaires originaux sont fllm*s en commenpant par la premiere page qui comporte une empre-nte d'Impression ou d'illustration et en terminant par la dami«ra paga qui comporta una raHa acnpraiiiia. Un das symboles suh^ants apparaitra sur la damiAre image da cheque microfiche, selon le caa: la symboio signifie "A SUIVRE". le symbols V signifie "RN". Laa eartaa. planches, tableaux, etc., peuvent Atre film«8 * dee taux de rMuction diff«rents. Lorsque le document est trop grand pour itre reproduit en un saui eiich*. ii eat film* * partir da I'angia sup*rieur gauche, de gauche * droita. at da haut en bas. en prenant le nombre d'images n*cessaire. Las diagrammaa suivants illustrant la m*thoda. 1 2 3 4 5 6 9* JSr. TaM PREFATORY NOTE The work embodied iti tliis article was done by Mr. J. H. FauU, as a graduate student in Botany, iu the Biological Department of the University of Toronto dur- ing the winter of 190001. E. C. Jeffrey, Lecturer on Biology. r THE ANATOMY OF THE OSMUNDACEAE By J. H. Faull, B.A. THE ANATOMY OF THK OSMUNDACEAE. (KtprimltJ fi-t'm Iht H 'Unical Gattlli, Vol. .V.V.V//** (with I'LATES xiv-xvii) INTRODUCTOKY. The caulinc vascular system of tlic Osmimdaccae has attracted considerable attention on the part oi morphologists, since it is exceptional among the leptosporangiate ferns in exliil'itin^f a ty|)e of strir ture presented by the |)hancrognms. Thus UeHary, the exponent of the "bundle system," states that "collateral bundles" i.re with rare exceptions characteristic of the stems and leaves of the phanerogams, but are likewise found in the Osmundaceae,' and that in their arrangement in the stems of the Osmundaceae they follow the " dicotyledon type." * Later we find Van Tieghem, the first enunciator of the "stelar theory" expressing himself as follows:' La tiKc Osmundts et des Toti/es diffire de celie des autres Fougtres. La stile axile et sans moelle du jeiine URe, au lieu de te diviser en restant grcle, ileineurc simple en s'dlar^issanl ()ro>;ressivemcnt i mesure (jue ia tige grossit; clle prcml une mocHc de plus en plu!j large, Ji la pOriphcrie de lia()ueUe soiit ranK^ iii ccrcle uii certain noinbre de faisceaux lil)c'r«)li^;ncux it bois sdpards, niais .'i libers conlluents, etitourcs d un pdrii ycle comniun et d'un eiuludernie g£n£ral. En un mut la tige de ces |>lantes demeure monos- ttlique & tout age, corame celle de la plupart det Phantrogame*. Plainly enough, therefore, these eminent botanists, starting from very different conceptions, have arrived at the same con- clusion, namely, that the central cylinder of the Osmundaceae resembles that of the phaneroj^ams. It is important to note, however, that heretofore all anatomical researches in this family have been confined to the tropical genus Toilea and the cosniopolitan Osmnnda rei^j inodullatcil mono- stelic central cylinders. Since then more extended researches have been made which have yielded important results. Thus it has Ikch shown thai the central cylintliT of Ophioglossum and of Uotrychium instead of being meduUated monostclic is in reality " gamodesmic ; " « that the central cylinder in the entire family Kiiuisetaceae, some of whos^.- s|)ecies were included in the exceptions, is of the same kind ; ' and that the central cylin Icr of the genus Helminthostachys it alio of the "gamo- dcsmic" type.'' It is true that Strashurjfcr holds' that the internal cndodermis and endodermal sheaths about individual bundles are of intrastelar origin, and not of cortical as is the external cndodermis, and that therefore these exceptions still sUnd; but this objection may be advanta^'eously left for sub- sequent consideration. Of the apparent exceptions, the family Osmundaceae has remained untouched, and 1 have undertaken the present research on this anomalous case, with the primary object of furnishing data that will help determine the proper mori)holo^ical interpretation of its vascular system. The family Osmundaceae is a very limited one in point of numbers, consisting of but two genera. Osmunda with eight species, and Todea with six, and therefore constitutes a very small part of the fern flora of the earth. But this does not seem to have always been the case,' for the Marattiaceae, although overwhelmingly predominant in the Coal period, con- stituted but 4 per cent, of the total (ilicincous flora in the Lower Jurassic, the remainder being composed of Osmundaceae and Cyathcaceac, with the related families Matonineae and Proto- polypodiaccae. As to distribution, the first genus is confined to the northern hemisphere, and the Todeas are with one exception found only in Australasia. Five Osmundns belong exclusively to restricted areas in east Asia and the aiijoining islands ; 0. • PoiRAOi T : Ann. Sd. Nat. Bot. VII. it i 113. 1893- JjKKKKKV: Mem. Huston S.)C. Nat. Hi»t 5:155. 1899. •Karmkk: Ann. Bot 13:421. l!i99. ' STRASBURr.Eli : Hiiiologische Beitriig*. 3:—. 1891. ■ Scott : Studies in fossil tmtany 304. 1 900. [36] FAUU. Tlir. ANATOMY Ol TMI OsMl NDA< CAK Cli'ytuniiinn occurs in the lliiiialayas and North America ; V. (inntimonuii in eastern Asia, North and South America ; and 0. ngiilis in every continent except Xu.stralasia. Of the Todeas, T. barhitra is a native of Australia, New Zealand, and South Africa; and the remnininv,' species, the »o-catled " filmy " Tudcas (Lcptopteris of some authors), belong to oceanic island* in the eastern suuth-tropical region.* Uf these species I have had the op|x>rtunity of studyin \ five, namely, 0. rt'gttlis, 0. t mmimomeit, 0. Cliiylununiii, 7". biirbimi, and T. superba. Nevertheless, in the ' >lIowing page* most atten- tion will be devoted to O. einnamomta, not io much because its anatomy has not previously been ilescribeil, .is bec.iiise the writer, for reasons which will become apparent, believes it retains a more primitive type of skeletal axis than rst.y of the family so far invcstijjateil. The material ul tiie specier of Osmunila stud- ied was collected from several different localities, and in large quantities. Of 0. einn'^momtn specimens from fully a hundred am! fifty plants were preserved and examined, and of each of the others perhaps one-third of that number. The more important points were verified from specimens taken from three different localities. Observations have been mainly restricted to the mature root, stem, and leaf trace. Some young plants of Osmunda were studied, anti tlii- j^rowin^ points of the older stems have been sectioned. Hut the mature stem, especially the region at which it branches, has proved to be of chief interest from the stand* point of questions of comparative anatomy. THE STEM. Gf.nkrai, anatomy. — The mature stems are very s\ . rhi- zomes, exceptionally so in T. barbara, which grow in a direction somewhat oblique to the horizontal. The leaves are in a closely sei tuft at the anterior end, for they are annual and the inter- nod'.s are very short. The broadly winged, overlapping bases with their sclerenchymatous sheaths resist decay long after the remaining portion of the leaf has perished, and these, together with the roots, which are very numerous, greatly add to the bulk of the stem. The stem usually bifurcates once into two • DiBLs: Englerand Hrantl's Natiirlichcn rfluuenfunilicn t<:377. igoo. [371 6 FAl'LL: THE ANATOMY OF THE OSML'NDACEAE branches of equal size, which lie in a horizontal plane. A few specimens of O. regalis were found, however, in which one of the forks was much larger than the other, but the larj^er almost immediately divided again, so that there were three branches of about the same size lying in the same plane. The forking bears no relation to the number of leaves ])roduced, counting from the cotyledons, nor to the age of the plant. Orcasionally there is no branching at all, though maturity has long since been attained, while in rare cases it has taken place comparatively early in the life of the fern. The rhizome exhibits a very characteristic appearance in cross-section {fig. /). The outer portion, the thick external cortex {ex. c), consists of very resistant, dark -brown scleren- chyma, in O. cinnamomea of a rich icd-tinted brown, in O. regalis and the Todcas of a black, and in 0. Claytoniana of a dull brown hue. The cortex is marked by leaf-traces (//), which form a close spiral, and at the nodes by the escaping roots (r). In O. cinnamomea sclcrification of the cortical tissue is later in taking pl.icc than in the other species. The internal cortex (i. t.) is parenchymatous, comparatively narrow, roughly pentagonal, and its colls arc heavily loaded with starch grains. Passing the pericycle and the bast region, which form a complete sheath, the wood (-t) of the stele is seen to be broken up into bundles of various shapes arranged in a circle, and sepa- rated from one another by the so-called medullary rays. These medullary rays extend out from a large pith. The pith or medulla in 0. Claytoniana and T. snperha is apparentlv homo- geneous. In 0. regalis it is often discolored and may contain one or more strands of brown sclerenchyma ; in 0. cinnamomea it is very frequently characterized by some brown sclcrcnchyma- tous tissue, and in T. harbara there is a large axial strand of this supporting tissue. Histology. — But we turn now to accpiire a more intimate ac(]uaintance with the stem as revealed by a study of its histo- logical features. For this purjjose several sets of transverse and longitudinal series were prepared, and a great many microtome sections examined. The material cut included stems of various ages. .As development jtrocceds rather slowly, all the tissues are mature only at a considerable distance from the apex of the plant. m FAOU: THE ANATOMY OF THE OSMUNDACEAE 7 The cortical part of the stem has little of interest for us other than in the respects already mentioned. The scleren- chyma consists of elongated, thick walled cells, with a small lumen containing starch grains. The walls are brownish, and marked by simple pits, which are round or slit-like. According to Strasburger,'" the endodermis is not the innermost cortical layer, but I am unable to verify this. He has made the state- ment that the innermost cortical layer at a certain stage divides by tangential walls to form several layers of cells; of these, the outermost becomes differentiated as the endodermis, and the remaining layers lie between this and the phloem, filling the place of a pericycle. The somewhat elongated cells of the endodermis are marked in every case by the characteristic cuti- cularization of the radial walls, which in transverse section shows as the "radial dot" (Jig. 6, e,e). The "radial dot" is dis- tinctively brought out by treatment with phloroglucin and hydrochloric acid, and also with dilute sulfuric acid. In O. Claytoniana the radial markings are generally not as distinct as in the rest of the species studied, and the cells are reduced in size in comparison with those of the layers in contact 8.c). The contents in this species, too, are meager, consisting of granular protoplasm, a nucleus which as a rule stains a deeper red with saffranin than those of surrounding cells, and a few starch granules as shown by treatment with iodin solutions. Sometimes the endodermal cells of 0. cinnamomea arc likewise apparent by the lack of contents, in contrast to the heavily-laden cells, both ectad and ccntrad. Generally in this species, as in the remaining ones, T. superia excepted, the cells are filled with tannin, so that the endodermis stands out very distinctly. The pericycle is entirely parenchymatous and consists of several layers — in 0. Claytoniana and Todea of two or three, in 0. cinmjinomta of three or four, and m 0. rer;alis of one to three. The cells are elongated, cylindrical, provided with laI^'e nuclei, and filled with finely granular contents, part of which is starch, llacmatoxylin imparts to this tissue a light blue color. Tangen- tial sections show that the orientation of the cells is very irregu- lar {figs. J and 9,p). Immediately opposite the point of origin of a leaf trace, and for a short distance below, the long axes of "Strasburof.r. OA cit. 449. 8 .inX: THt 'NATOMT JF THE OSMUNDACEAE the cells run parallel with ttie long axis of the stem, but for the most part in the remaining regions of the stem there is consid- erable disturbance, though only in tangential planes. This dis- turbance is commonly so marked that the long axis of the cell is at right angles to the stem axis, and between this and the paral- lel position there is every gradation. Therefore in transverse section these cells are either round or more or less tangentially lengthened S. p). This variation in orientation is of interest, as it is connected with a similar phenomenon in layers lying nearer the cauline axis, namely in the phloem region. XvLF.M. — Before dealing with the [)hlocm, however, it will be convenient to describe the xylem. The wood elements are of two kinds, namely, small ringed and spiral elements consti- tuting the protoxylem, and scalariform tracheids which are of later development constituting the metaxylem. Occasionally a parenchymatous cell is found among the tracheids. A transverse section shows, as mentioned before, a ring of variously shaped bundles ; and by tracing these up and down, or by boiling a piece of stem in potash and then removing the softer tissues, there is shown to be a network forming the wall of a hollow cylinder, the strands being the "bundles" of DeBary, and the meshes the spaces occupied by the " medullary rays." Though there is a great deal of regularity in the apparent construction of this net- work, as proved by DeBary and Zenetti in O. regalis, yet a study of development shows that the " bundle theory " is inadequate for giving the right conception of the vascular system. In the young stem of the Osmundaceae the wood forms a completely closed cylinder, and Van Tieghem, basing his conclusions on Todea and 0. regalis, has stated this to be the case for the whole family. I am able to state that the phenomena in the young stem of O. ciniuimomea and 0. Claytoniana arc in accord- ance with his general conclusions in this respect. Now directly above the point at which a leaf trace leaves the stele the wood is not developed for some distance. This gap is filled by parenchyma chiefly, except at the outer part, which is occupied by sieve tubes. There are exceptions in O. ci„„amomca to be described later. Thus a transverse section of the stele, ju-t above a node, shows a ring of wood br ken at one place, the oreak being occupied by the tissues just referred to; in other [40] FAUIL: THE ANATOMY OF THE OSMUNDACEAE 9 words, the stele here has one medullary ray. Fig. 2j shows a transverse section of the stem of 0. Claytoniana through this region. Still further up the internode the ring is complete again. There is the same sort of gap above the second node. However, as the nodes become more fre(]ucnt, that is, as the intcrnodes become shorter, a leaf gap extends through more than one internode, and in a transverse section there is more than one medullary ray, until in the full i^rown stem, where a leaf gap extends through several internodcs, a transverse section shows several gaps cut across, or in other words shows several medul- lary rays. It is therefore evident that the number of medullary rays seen in any transverse section depends on the frequency of the nodes and the length of the gaps. In well nourished stems the number i.s greatest in 0. Cluytoniaiia {fig. //), there usually being about twenty, and in T. barbara {fig. 24) the fewest. In this species the gaps are quite short, so that while the wall may be thin in many [)!accs at any given level, there are not more than two to six medullary rays seen in the cross section {fig. 24). The persistent portions of the cylinder of wood, the "bundles," present .irious contours in cross section, the shape of any particular portion lying between two adjacent gaps, that is, of any strand, varying with the level at which it is cut. Just below where the leaf trace is given off, the wall is hollowed out on the side towards the pith, so that the transverse section of the strand presents a horseshoe shape {fig. 17). The middle of the inner surface of the strand at this level is occupied by protoxylcm, which consists of about a half dozen small ringed and spiral vessels. Following the strand down, it is seen that the arms of the horseshoe thicken on the sides facing one another, especially towards the ends of the arms {fig. ^jccie« studied, we naturally seek an explanation of this peculiar dis- position of the phloem. In his memoir on sieve tubes Janczew- ski," who could hardly have been prejudiced by any stclar theories, noted that isolated sieve lubes occur occasionally here and there in the medullary rays of 0. rtgalis. The writer has found undoubted cases of the same thing in O. cinnamomea. Two such eminent botanists as DeBary and Strasburger have disagreed as to the topographical distribution of the layer of metaphloem sieve tubes in 0. regalis. The former states •* that the sheath is continuous, while the latter states that he puts himself in opposition to DeBary on this point, for he considers tiie phloem to be interrupted opposite the medullary rays. Strasburj^er does not say for what reason he considers the cells opposite the medullary rays not to be sieve tubes. My own observations on O. regalis are precisely in accord with those of DcHary and Janczewski. The cells opposite the medullary rays differ in no way from the sieve tu'>es opposite the xylem strands. I have found the same to be true of the other species studied, with the additional observation that isolated sieve tubes occur sometimes in the tissues filling the leaf gaps of 0. cinnamomea. To this last observation I have two others to add, namely the occurrence of an internal phloem in whi the sieve tubes form a more or less continuous rin^ {fig^- ^i, and 22^, and in rare cases the union of external and internal phloem through a leaf jjap. In a certain rich, moist situation about a dozen well nourished plants of 0. cinnamomea grew, of \,rhich, on examina- tion, five showed the phenomenon of a continuous layer of inter- nal phloem. Search in an adjoining locality resulted in finding specimens which showed the same feature. To •extend the range of observations, I visited a peat bog some twenty miles distant from Toronto, where I knew the cinnamon fern grew, and secured specimens characterized by the same peculiarity. Fig. 21 shows a transverse section of a stem found in this last locality. "Janczewski: op. cil.bb. "•DkBaRY : of. cil. 360. wStpasbi'RGCB : of. rit. 449. [48] FAUU: Tm ANATOMY OT THB OSMUNDACBAE 17 The sieve tubes of this internal phloem are ai typical a* those of the external, and except for their position not be distinguished from them. They do nut always form a continuous ring as do the sieve tubes oi the external ihloem, but are often in more or less detached groups, embeddcil in small celled parenchyma. The layer of sieve tubes is from one to three cells thick. It should be added that internal phloem occurs only near where the forking of the stem takes place. 0. einnamomea shows likewise two other features which are constant throughout every part of the stem, and at once distin- guish it from other species : ( I ) an internal endodermis, and (a) several layers of parenchyma between this and the xyleni. Internal endodermis. — The internal endodermis possesses the characteristic radial tiot, thouj^h sometimes nut as clearly distinguishable as in the external endodermis (i. e.,Jig. 2). Its cells are usually larger than those of the latter, but are filled with similar contents, most frequently tannin 15)- 1^ further to be noted that it bends outwards opposite the leaf gaps {Jii:s 10, etc.), and not infrequently connects through them with the external endodermis. I have examined scores o{ stems of the cinnamon fern, and in every specimen there was an internal endodermis. On the contrary, it seems to be invariably absent from the other species studied. As the central cylinder of the family Osmundaceae has heretofore been classed as monostelic, the existence of an internal endodermis in one of the species is therefore a matter of considerable moment, especially if it be regarded as a real phloeoterma. Between the internal endodermis and the xylem there is a cylinder of elongated parenchyma, rich in starch and prot()])lasm, and from two to seven cells in thickness. This layer is continu- ous with the medullary rays. In 0. regalis, 0. Claytoniana, T. barbtirn, and T. suptrbu a similar but thinner layer it found as a rule, and the cclLs a-e always smaller and richer in contents than those of the medulla on which they border. The mkiiuli..\. — The medulla is very lar<;e in this family, particularly so in 0. Claytoniana, and consists of large-celled parenchyma. Most of the cells are partly filled with large starih granules, but frecjuently some of them contain tannin, especially in T. barbara. A brownish fluid may occur in inter- [49] Il Pm,' ■ Tmi amatomy or rm osmunoaciai cellular •pscei, and in 0. rrgn/is within the celti themielvef. In these rcjjards there is often a strikini( resemhlance between the |iareiicliyina of the medulla and that of the internal cortex in the lame plant. But there yet remains to be described a still more siifnificant phenomenon, namely, the occurrence in the pith of brown sclcrenchyma of the same kind as is found in the external cortex {fi^. 14 and ao). This is probably a primitive feature, and in this, as in many other respects, 0. dnmtmomca proves to be most interesting. Out of forty-four I'icccs of stem, chosen at random, and representing a corresponding number of different plants of this species, twenty-five of the exam[)lcs showed brown sclerenchyma in one or both ends. It occurs as a central strand, varying in size from a few cells to almost the limits of the pith, or as several small strands irregularly arranged. Fig. 14 is a phutograph of the transverse section of a stem in which there is a large axile strand, and fig. /j of one in which the scleren- chyma is entirely absent from the pith. Further it has the peculiar habit of being present at o e level, but perhaps not at another ; so it is likely to be found in nearly every plant if the stem be sectioned from end to end. This same habit is character! '"c of il.s appearance in 0. regalis {fig. .=0), but more often it is n t present at all. That brown sclerenchyma occurred in the pith of 0. regalis did not escape the observant DcBary,~ but elsewhere I find no reference to this fact. -Strangely enough, however, out of thirty-five or forty plants harvested from one locality there was not a trace of sclerenchyma to be found in the medulla of any of them, while in one region not far distant 25 per cent, showed this phenome- non, and in another a still higher per cent. Parenchyma is the sole constituent of the medulla of O. Clay- ti>mana (Jig. 17). This is prob.ibly true of T. superbu too. Fig. as is a cross section of T. barbara taken too near the growing point to show sclerenchyma, but farther down the medulla was occupied by a large strand of this tissue (fig. 24). Thus medullary brown sclerenchyma is usually present in 0. cittnamomea, in O. regalis not uncommonly, and in O. Claytoniana not at all. In T. barbara it also occuis, but apparently not in T. siif'crbii. It is j)erha])s significant that such series can be DlHakv: .'J.', p. igo. (50) PAUll : THE ANATOMY OF THI OtMUmMClAB •9 arranjjcd, but of ^'reatcr imjiortancc is the fact that the occur- rence in thu Osmumiacfiic of brown -.clcri-ncliytnatous tissue, apparently within the caulinc central cylinder, has no parallel among eriitting ferns. The fork. — There yet remains to be described the anatomy of one particular (Kirtton of the stem, the part in the region of bifurcation. It has been stated that it is peculiar to the stem of the Osmundaceae to fork once, and that in a hori/.ontal plane. We shall treat first of the phenomenon in 0. cinr-vnomca. Trac- ir , .e main stem forwards, it is seen to become flattened and then to become constricted in a median vertical ()lane. Imme- diately anterior to the point of bifurcation of the vascular axis, there is a wide ramular gap in the central cylinder of each branch [Jit;, /o). Sections of the main axis immediately beluw the fork show two bands of phloem, one on the upper and one on the lower internal surface of the central cylinder {/g. ij) . ^ections passing? throu^'h just in front of the region of bifurcation show similar bands of phloem along the inner wall of the central cylinder of each branch {Jig. //). Cases have been described above, in which there is a complete cylinder of inter- nal phloem instead of the two isolated bands just referred to (Jig. 3i) . The internal and external phloem connect through the ramular paps (Ji^;. ii). Likewise the internal and external endodcrmis are in textural continuity through these gaps, so that there is free communication between the cortex and the pith {fig. lo). Sometimes the cortex lyinu; between the two branches con- tains brown sclerenchyma which is continuous through the ramu- lar f'aps with strands of the same tissue occurring in the medulla of the branches. Frc(|ucntly in less vigorous plants a transverse section of the main axis posterior to the point of ramification shows \ di.iriiond-shaped piece of cortex surrounded by cndo- dermii {fi^. 12). Posteriorly this included piece of cortex becomes continuous with the medulla of the main axis {Jig. ij) , and anlciiorlv with the general cortex (_/??■. //). Twenty -'ive forks of 0. cinnamomea were selected at random and sectioned. Twelve of them presented the phenomenon of typical wide ramular gaps. Six of them were of the reduced kind just ueseribed. In five c:ibcs there were gaps in the xyiem [5«1 20 faull: the anatomy of the osmundaceae only, cortex and medulla never becoming continuous; and in two even the xylcin did not open u|) {fig. i6). For reasons to be outlined later, the writer believes the wide gaps to be the most primitive. 0. regnlis presents a much degenerated form of ramular gap, for here only the xylem opens {fig. ig) . In 0. Claytonmna the defreneration is carried still farther, for as a rule there are no branch-gaps at all {fig. /8). In T. barbara the xylem alone may open up. The phenomena of the fork may be thus summarized : { I ) Complete ramular gaps occur only in 0. cinnamomea. (2) Internal phloem occurs only in 0. cinnamomea. It is found in the branches just above, and in the parent axis just below the point of bifurcation of the central cylinder. (3) The internal phloem may form an entire cylinder. (4) Where gaps are comjjlete, the cortical and medullary tissues connect through them. (5) Thus sclerenchyma of the cortex is sometimes continu- ous with sclerenchyma in the medulla of the main axis, and of the branches. (6) O. cinnamomea presents the following forms of ramular gaps arranged in order of degeneration, (- ii l/i.ri, and is continuous with that of the stem. In T. harhara it frecjuently breaks into two or three groups. I52J FAULL: THE ANATOMY OF THE OSMUNDACEAE 31 Surrounding the wood is a layer of parenchyma, which on the concave side of the xyiem quite fills the space between the arms of the horseshoe. The phloem consists of a crescentic band of sieve tubes, one to three cells thick on the external side of the leaf trace (/>/<) , and a smaller band on the opposite side (j>lt). The protophloem consists of small elements which form a ring, broken only on the concave side of the xylem. Here the ring is completed, however, by the inner band of metaphloem. In 0. cinfuimomea and T. barbara isolated proto- phloem cells have been observed by the writer on the side of the inner band of metaphloem towards the stem axis. On the convex side the protophloem is separated from the metaphloem by paren- chyma. There are no "quergestreckte Zellen." The pericycle consists of two or three layers of cells, ancl is bounded by a well developed endodermis continuous with that of the stem. With reference to the attachment of the leaf trace to the cauline vascular axis Zenctti has given a very careful and accurate description." Strasburger has held" that the stele of the petiole of 0. ngtilis is a collateral bundle. He h is considered the inner band of metaphloem to be a parenchymatous tissue. However, the cells of this band prove to be characteristic sieve tubes, and aie continuous with sieve tubes in the stem opposite the medullary rays. The leaf traces, therefore, are undoubtedly concentric. Several botanists have arrived at the same conclusion for 0. In summary, the most important features of the leaf trace are : (i) the absence of a pith, (2) the endarch xylem strand. (3) the concentric tyjje of stele, (4) the absence of " <|uergc- streckte Zellen," and (5) the cylinder of protophloem completed on the inner face by a band of metaphloem. Tin: KOOT. The roots have a definite relation to the leaves, both in posi- tion and in numbers. Two roots invariably originate from the base of evcrv leaf trace, or from the central cylinder immediately below. They come off at the same level, one opposite each arm of the horseshoe-shaped strand of xylem {fig- iS) in every case "Zenetti : op. til. 69. « Scott : op. eit. 319. "Strasuurger ! op. nt 448. mZknetti : op. cit. 66. 1531 33 FAUU: THE ANATOMY OF THE OSMUNDACEAE where there are just two roots to a leaf. They grow almost directly outwards, anil so in a transverse section of the stem are cut longitudinally. In such a section it is seen, likewise, that the cortical tissues of the stem and root are entirely independent of each other, and that, therefore, the root is of endogenous origin. This fact is true of the secondary roots also. The cortex is exceedingly thick, forming by far the main bulk of the root, and consists of large celled sclerencnyniatous tissue. The cortical cells diminish in size towards the periphery, and become thicker walled. In T. btirbara, hovv 'er, there is a discontinuous ring ci exceedingly thick walled brown sclercn- chymatous cells immediitely surrounding the vascular axis. The endodermis, which is continuous with that of the stem and leaf, is very ])ronounced in all of the species, and is at once noted by the radial dot, and by the fact that its cells are filled with tannin. In the second particular, exception must be gener- ally made of 0. Claytoniana. The stele is comparatively small, and is typically protostelic, since there is no pith. The wood has a narrow elliptical form, consisting mainly of very large scalariform trachcids. At each end of the ellipse there arc a few small proto.xylem elements, which are especially evident in the young root, and which have no connection with the proto.xylem of the stem or leaf. The root, therefore, is diarch. There are likewise two bundles of phloem alternating radially with the bundles of xylem. In all of the Osmundas, however, I have observi- l triarch steles in the larger roots, which exception is of comparative frequence in 0. cinnatnomcd. The phloem consists of two flat bundles or bands. These bands are made up chiefly of thin wallod sieve tubes which are of the same kind as occii in the stem. None of them arc "<)ucrgestreckt." The phloem is separated from the xylem by three or four rows of jjarenchynia, and from the endodermis by a two rowed parenchymatous pcricycle. DEVKLOI'ME.NT OF THE TISSUES FROM Till- CKOWINT, 1 HNT. In discussing this subject there are two points in ji.iriicuiar which will receive special consideration : ( i ) the statements of Strasburger and Zenetti regarding the origin of the endodermis, and (2) the real nature of the "qucrgestreckte Zcllen." [54] FAUU: THE ANATOMY OF THE OSMUNDACEAE 33 The determination of the relation of the tissues to the apical cells seems of little concern, and moreover in the study of the apical region of the growing point there are serious difficulties. Having described these for 0. regalis. Professor Bower aptly remarks ; The meristem being thus at times irregular, and the subdivisions of the segments being variable, it is to be expected that the study of it (the apical region of the growing point) in longitudinal section would present difficulties, and I have not been able to trace any definite and characteristic mode of segmentation. Longitudinal sections cut from a considerable number of stems show that a conical apical cell is present. The relations of the sur- rounding tissues, and their reference to regularly succeeding segments are difficult to recognize. To these observations Cii the extreme apical end of the growing point we have nothing to a'unc/um vi^^itiitiottis this is approximately true of all the cells of the stem. At this earliest stage one would hesitate to say, because certain cells were corradial, that they were therefore division products of the same mother cells; so Zenetti's con- clusion, based on this sole argument, scarcely seems < anclusive, even grantinjr the correctness of his observation. If, too, such a conclusion were correct there would be the curious anomaly of certain phloem and cortical tissues having a common origin. Evidently the study of transverse sections cannot settle the matter. To attempt to follow these layers upwards is obviously only possible in median longitudinal sections. But in the stems of the Osmundaceac the leaf traces are exceedingly numerous, and at the growing point arc closely packed together, and appear before the tissues of the cauline central cylinder become at all differentiated. Hence, no matter what be the plane of section, the endodermis cannot be traced continuously very far anteriorly to the point at which it is differentiated, for a' leaf trace is certain to intervene ; and I found it quite out of the question to pick out an undifferentiated endodermis on the side of the leaf trace turned towards the apex. Therefore, every attempt failed to refer the endodermis and the rows of cells "occupying the j.lace of the pericycle" to the .same initial layer. The typ 1 protophloem, and the " quergestreckte Zellen " begin to be differentiated sinuiltaneously with the appearance of the protoxylem. They are best examined in tangential sections. Their walls at this time become pitted, and their contents much less granular than those of the surrounding cells. Here, as in the maturer parts of the stem, there app.-ar to be no differences between the typical protophloem and the "quergestreckte Zel- len." Their rehation to tile leaf traces seems to explain their irregularity in orientation. Immediately below the point of ori- gin of a leaf trace they are arranged with their long axes parallel to the long axis of the stem, and there is a gradual transition to l56J FAULL: THE ANATOMY OF THE OSMUNDACEAE 25 the tanj^cntial position. More than this, the laterally placed protophloem cells of the leaf traces can be directly traced into the " quergestreckete Zellen " of the stem. There seems little douLit, tluTL-fi)rc, as to their nature. To summarize observations : ( I ) The " cjuergestreckte Zel- len " are sieve tubes, as has been demonstrated above ; (2) they become differentiated at the same time is the tvpical proto- phloem, and (3) occupy the same relative position; (4) they resemble the protophloem cells in form ; ( 5) their orientation is not unift>rm ; (6) tliev Jiass imperceptibly into the lontfitudi- nally orientated protophloem cells of the leaf traces. Hence there seems no reason to regard them as anything else than pro- tophloem. CONCLUSIONS. The question now remains, how to interpret the vascular sys- tcr-i uf the Osniunilaceac. To do this more intcUigib! v, it will be '> cU to recapitulate the main fibrovascular theories. We shall begin with that of Sachs and DeBary. These botanists regarded the bunille as the unit, and the vas- cular system as a more or less simple complex of bundles embedded in ground or fundamental tissue. Developmental studies have shown that this theory is inadequate, for the unit is wrong. The hypothesis which at present obtains v\is supplied bv Van Ticghem and Strasburgcr. In this conception'^''''3°' the stole is the unit. The primitive form of stele, the monostelc, such as occurs for example in most roots and in the stems of Ivcopods, is a solid central strand of xylem, surrounded by a sheath of phloL ii, and marked off from the cortex by the differentiateti internal cortical layer, the endodermis. Of this there are many modifi- cations, of which mention is made of the most im])ortant. Hy the repeated bifurcation of the monostele, the polystelic tv[)e is presented, as in Primula and Pteris, each segment being in every respect a steli'. If these stoics fuse laterally, thus forming a ring with internal and external phloem, the gamostele is pro- duced as illustrated by Marsilia. Again, when parenchyma "Van 'I'lhUiHr-M : Tr.iil(5 ile liul.iiiiiiuc 073, 765. •» Van TlliiiiiEM : Sur la polystiJlie. .\nii. Sci. .Nat. Hot. VII. 3 ; 27s. "Van TlEaHEM : l<oents de Botanniue 1:84, 179. l57] 36 FAULt: THE ANATOMY OF THE OSMUNDACEAE segregates in the axis of the monostelc, and the vascular ring is broken into strands by cctad extci.sions of this ])ith (the medul- lary rays), we have the medullated monostelic type, such as is common in phanerogams. It is to be noted that the medullary and cortical nics arc considered by both these botanists to be of morpholoirjcally different value. Now by the bending in of the endoderniis of the medullated monostele between the bundles, and the fusiui: of the ends of adjacent groups on the centrad side of the bundle, so that each bundle has its cndoder- nial sheath, and medulla and cortex become continuous, the schizostelic or astelic type results. Of this phenomenon Ranun- culus d Equisetum afford examples. A modification of this ty])e, the gamodesmic-schizostelic, is produced by the lateral fusion of these endodermal sheaths, so that there is a common internal and a common external endodermis. If the internal endodermis degenerates, as it does in E. arvenst, then there is evidently a simulation of the medullated monostele. It is fair to add that Strasburger dissents 3' from the last two types des- cribed, the astelic and the gamodesmic, for he regards the endo- dermal sheaths about the bundles in the first of these, and the internal endodermis in the second, as not mori)hologically phloeotermal, but originating from specialized stelar cells. The researches of Gwynne-Vaughan^' and Jeffrey" have shown that the phenomena said to lead up to j)olystely do not occur in Primula and I'teris. If the polystelic conception falls, obviously gamostcly goes too. P'urther, astely has been shown, where it occurs in Equisetum and Ranunculus, to be preceded by the gamodesmic ap])earance. Later the internal and external endodermis may fuse between the bundles, but in no case is there an inward loojjing of the endodermis. Finally, the stelar origin of the pith of the medullated monostele has been disputed, and the question raised as to whether the medullary and cortical tissues are in reality morphologically different. In other words, is the medullated monostelic tyjie primitive, as its simplicity might indi- cate, or has it resulted by degeneration from more complex types ? '■Si Asm RCER : of. tit. p. 442. J'CwYNNE-VAi r.HAN : I'oIy»tely and the genus PrimuU Ann. Bot. 11:307. 1897. njf 1 1 K^ V : M'lriili.ilDgy of the central cylinder of angioiperms. Trans. Canad. ln:*l. 6 : — . (i 4*J; i'>uu. [5»] FAULL: THE ANATOMY OF THB OSMUNDACEAE »7 It is interesting to note that Potoni^ had discussed this last question from the stan(J|)()int of fos ' botany, and concludes^* that it seems evident in the case of certain groups, such as the cycads, that the simple results from the complex (for example, the cycads from the Mcdulloscac). Ilcncc for these groups at least he is inclined to reject this idea of segregation of parenchyma in the center of the protostele to form the meduUated monostele, but holds that the medullated nionostelic type has probably arisen by degeneration from his " pericaulom." Since this peri- caulom was produced, according to his theory, by the lateral fusion of leaf bases in the stem surrounding the originally solid stele, the " urcaulom," the medullated monostele has been derived from a form of central cylinder such as Van Tieghem has described as polystelic, preceded or accompanied bj- the dis- appearance of the enclosed urcaulom. The paleontological evidence, however, appears not to be conclusive, for in the very group that Potonie cites, the cycads, so eminent a palcobotanist a- ^' D. H. Scott takes a directly opposite view. He points out ' hat the vascular system of the Medulloseae was typically polvstelic, while in the recent cycads there is but one vascular cylinder, and that hence "we should involve ourselves in unneces- sary complications if we endeavored to derive the simple, primary structure of the cycadean stem from the more elaborate organi- zation of a MeduUosa. It is far more natural to suppose that the monostelic cycads arose from monostelic ancestors." In 1897, Dr. li. C. Jeffrey put forward another view of the vascular system ,3^ based upon a study of the young sporophyte. Here, too, the stele is the unit. According to this conception there are two jjrimitive types of vascular axes ; the first the same as V; ghem's primitive type, and designated "protostelic ;" the s . - i- i one in which there is a hollow cylinder, or "siphono- stcl- sc external wall abuts on the cortex, and whose inte wall encloses the medulla, and which possesses internal as well as external phloem. This is the "amphiphloic siphono- stelic" type, called by Van Tieghem the "polystelic." The xPoTONii: Die Metamorphose der I'flanzen im l.ichte pal.x-ontologi&cher Thataachen 22. MScoTr : Stuilies in fossil botany 395. 1900. ^Jbtl-KkV : TraUb. Uril. AuMi. luiuiilo. 1897. [59] 38 FAUlt! THE ANATOMV OF THE OSMUNDACEAE commonly called "astclic" modification results from the amphi- phioic type by a deffcncration of the internal phloem, and the medullated monostclic type of Van Tieghcm is derived from the astelic by the loss of the internal phloeoterma or endodermis. A study of development from the seedlinj^ is likely to show how these and other modifications in the stellar structure have been derived from the primitive types. Attention is also called to certain portions of the wall of the siphonostele in which the vascular tissues do not develop. These places lie above the points of exit of branch traces, an-i of leaf traces, and are known as ramular and foliar gaps respectively. Through these j,'aps the tissues outside and inside connect. In transverse section, the connecting tissues seen constitute the medullary rays, and the segments of the woody cylinder with adjacent phloem and parenchyma the bundles. A fact of great phylogenetic impor- tance in dealing with "gaps" was further pointed out, namely, that in small leaved plants, as in the Lycojiodiaceae, Etjuisc- taceae, etc., only ramular gaps occur. These plants are grouped in the division Lycopsida, and their stelc.f are said to be cladosiphonic. In all other vascular plants there is a gap for every leaf. These constitute the large leaved plants, the Pteropsida, and their steles are said to be })hyl!osiphonic. As a matter of theory, it is suggested that the siphonostele arose from the protostele for mechanical causes in the Lycopsiria to support the branches, and in the Pteropsida to support the leaves. Potonic also explains the origin of his second primitive type the "pericaulom." the homologue of the siphonostele, on mechanical grounds. In the light of these theories we can now apply ourselves to an interpretation of the anatomy of the vascular system of the Osnuindaceae, and likewise note if the facts already dealt with throw any light on the theories. First, we are in a better position now to decide whether the internal endodermis of O. cinnamomca is phloeotermal or not. It has been noted that in similar cases, that is, in gamodcsmic stems, Strasburger has denied the phloeotermal character of the internal endodermis. With regard to the internal endodermis the following facts have been observed : I . There is present the characteristic cuticularized "radial dot." [60] FAULL: THB ANATOMY OF THE OSMUNDACBAE »9 I 2. The structure and contents of the cells are materially the same iis of the external endoilermis. 3. The sheath is continued into the portions which in some i'dividuals present the phenomenon of internal phloem, just as in any form called by Van Ticj^hcm and Strasburgcr ganiostelic. In thegamostclic type the phloeotermal character of the internal endodermis has been admitted. 4. It generally connects with the external enduilcrinis through ramular gaps, and by no means rarely through foliar gaps. When this occurs, there is no point at which it could be said that the one stops or the other begins. Having verified these facts in a great many cases, I am there- fore of the opinion th.it the internal and the external endodermis are homologous tissues. Second, are the medullary tissues morphologically equivalent to the cortical ? Aj^ain we recapitulate observations. 1. They do not differ in structure or in contents. 2. The medulla very often contains brown sclerenchyma, at least in three species studied, a tissue which, in other ferns, never constitutes a part of the stele. 3. Medulla ■'nd cortex connect more frctjuently than not through the 'ar gaps in 0. cinnamomea, and occasionally thiouj^h foli s; and neither is there a transition in the nature of the cunnecting tissues, nor any line at which we can say, the cortical tissues He externally to this and the medullary tissues internally. 4. The cortical and the medullary brown sclerenchyma some- times fuse through ramular gaps in 0. cinnamomea. 5. Portions of stem of 0. cinmimomea have been found which are of the "ganiostelic " type of Van Tieghcm. The medulla in gamostclcs is granted to be morphologically a cortical tissue. The conclusion is evident for 0. ciimiimontiii at least, and if it be granted that the medullary tissues of this species are morphologically equivalent to the cortical tissues, then biological prinrijjles alone would demand a like conclusion for the other species. Third, of what type is the vascular system of O. cinnamomea ? Again the f.icts must form the basis (or .1 decision : 1. The young stem of 0. cinnamomea jjossesses an entirely (61) 30 FAUU: THI ANATOMY OP THB OSMUNDACIAB closed hollow vascular cylinder, sheathed with phloem and broken only imniLiliatcly ;iliovc the exit of a leaf trace ; and at a level higher uj> the cylinder is entirely closed again. There ia a medulla and an internal endodermis. 2. In oilier plants the leaves arc more frcijuent, and the i^aps extend through several internodes ; but yet the cylinder is the unit. The cylinder of phloem is quite rarely broken, exce|)t where branching takes place. 3, There is an internal endodermis which is persistent throughout the entire central cylinder of the stem. 4- As a rule the internal endodermis bends out opposite leaf gaps. 5. There is an internal phloem in portions of some plants. 6. Not only does the cylinder of external phloem remain practically unbroken, but opposite leaf gaps there is on the inner side a proliferation of sieve tubes. In O. n>^<;/u Janczewski found isolated sieve tubes in the parenchyma filling the leai gaj) ; and the same thing is true of O. cinnamomea. According to Van Tieghem's stelar theory, the last two facts Can be explained only by considering the central cylinder of the Osmundaceac to be "gamostelic." The ccntrad exten- sions of the phloem opposite the medullary rays could then be ex])lainLd by a.ssuniing that steles had united laterally, with the di.sappearance of phloem on the medullary side, but with the partial persistence of phloem on the radial planes. This would also explain the occurrence of internal phloem, the union of iiitcrnal and external endodermis, and the homology of medul- lary and conical tissue. But from the study of development there is not a shred of evidence to prove that there has been a union of steles. In fact, such a study .shows distinctly that there is but one stele in the stem of 0. cinnamomea from the very first. Van Tieghem's observations on O. regalis have already been (juoted (see Intkoductiox) ; so we cannot describe the cauline vascular system as "gamostelic," if this name implies \x union of steles. There remains yet another interpretation, namely, that the v, ular system of the stem of 0. cinnamomea is a siphonostele in which some degeneration frc have closed somewhat, so that medulla and cortex r.irely connect except through ramular gaps. Also the phloem I'ornis an almost unbroken cylinder, and the ccntrad proliferations op|>osite the medullary rays are the vesti^^ial relics of connection between external and internnl |)hloem. The internal phloem has also disappeared in greater part. With Mich a conception of the caulinc vascular system of 0. cinnamomai, the ccntrad accumulation of sieve tubes opposite the medullary rays, the occasional presence of sieve tubes in the meJullarv ravs, the fact of the internal i)hloem, the connection of medulla and cortex through ramular and foliar gaps, the presence of sclcrenchyma in the medulla, the bending out of the internal endodcrinis into the leaf gaps, and the facts of develop- ment, all become intelligible. Fourth, which of the species studied possesses the inost prim* itivv- tvpe of central cylinder ? After a fairly comprehensive study there is one feature that stands out prominently, the great similarity and uniformity of vascular structure in the varit)us species of Osmunda andTodea. According to Solms-Laubach the stems of fossil remains of this family, of which none earlier than the Tertiary have been found, do not present anv striking differences from the livini^ represen- tatives. Paleobotany, therefore, offers no solution to the prob- lem. In spit^of the conservatism of the central cylinder, there are, however, minor anatomical differences. On the basis of these alone, without referring to the young sporophvtes, I think there is sufficient warrant for placing O. cinnainonua at one end of the series. i>ossessing as it does an internal endodermis, inter- nal .phloem, and wide ramular gaps. It is difficult to sav which .^pccics is to be placeil at the other end of the series. In view of the fact thai 0. Claytoniana never has sclerenchvma in the medulla, that there are small or even no raniuiar gaps, no internal [63l PAUUi THi ANATOMY OP THl OtMirmMCMI •clercnchyma. ami even a degenerated external endodermis. we may not be far istray i i piittinj,' it in the |..)sition farthest from O. annamom.i.. Now of these two. which retains a cen- tral cylinder more nearly p. imitivc ? If O. ngalis has a medul- Uted monostelic cent-. I cylimlcr. as has hitherto been clai-ued for it. then 0. Cl,iytom,i,ui has also, and therefore, according to Van Tieghem, a more primitive form than that of O. cinnantomta. Assuming the correctness of this for the moment, it will be in order next to see if such phenomena as presented by 0. cinna- momea could be derived according to Van Tiegh-m's hypothesit from such a simple meduUatcd monosteltc form as that of O. ClaytoHtana. The i.iiloem sheath must have broken into bundles, and the endodermis must have looped in between the bundles, and con- nected around them on the centrad side. With the formation of this astelir tyj.e some of the cortex would have been included in the medulla, in evidence of which the sclercnchyma in the pith would stand as proof. Then next the bundles must have fused laterally to produce the gamo.lesmic type in which there IS an external and an internal endodermis. Grantinj; that the central cylinder could be so plastic in a single species, there are left yet to be explained the continuous sheath of phloem, the proliferation of sieve tubes opposite the medullary rays, tli<- occurrence of isolated tubes in the medullary rays.'the occur- rence of internal phloem, and the phenomena of the ramular gaps. 1- urther. there are no facts in development that point to such a series of changes. Turiiin- now to the other alternative, namelv, the possibility that a ann,imomt;tYi:xs the more primitive form of central cyl- inder, it will be grantf^d that by the degeneration of internal phloem, endodermis. ir.d medullary sclercnchyma. and by the closing of the ramular gaps the central cylinder such as we find m O. a,,ytoni,im would result. U proof that such degeneration could have taken place, it is t., be noted ( i ) that in O. amuwio,ma Itself, It has been pointed out that the amphiphloic con.lition is localized, that the internal endcniermis has already begun to degeiuT.Ue tli.it medullary sclercnchyma is not a constant fea- ture, and that closed steles above the point of bran. hing are not at all uncommon; and {2) in further proof, analogous cases of I64I FAUU: THC ANATOMY Of Tttl OMUINDACIAI 93 def(encrition within the Mme genu* >re frequent. Thus within the lectins lvnii-.cttim two N|ircics such .is /:. ttnu nsf ,iiul /:. hiimiilf may be chosen, the tirst long considered mcduUatcd munostclic and more primitive, the second gamodcamic and considerably modified. Hut .i study of di-vclopmciit and of nod.il por- tions of tlic stem has shown that h. tinunst has a reduced cen- tral cylinder, the product of detrencration from a gamodeAmic type, and th:it therefore H. Iiii mdh- is nearer the primitive. Sim- ilar cases of degeneration haye been pointed out by Van Ticghem, PoirauU, and Jeffrey, in the genera Ophioglossum, Hotrychium, Ki|uist tiiiii, R.iiiunculus, etc. Very lately Hoodie, "-i' has called .itteiition to an interesting series of central cylinders in the family Schizacaccae. Aneimia Phyilitidis has a ring of se]>arate l)iiiidles, each with a banil of xylem surrouiuied by a phloem, pencycle, and endoderinis of its own ; A. Alixictnui has a complete ring of xylem in the internodes with external and internal cylinders of phloem anil endodermis ; Schizaea has a ring of xylem surrounding a central pith, but no internal phloem or endoilerinis. It is likely that here, too, the Schi/aea type is derived from the Aneimia type by ilegeneration. In the Ilymcn- ophyllaccae likewise, every grade is found from the case in which the phloem of the solid stele forms a complete ring to !h:'t in which it is developed on one side only. After ex.imining a number of rompiir itively young specimens of Cliiyfoniitrut, I am somewhat doubtful if the stu''y of the development of this species will throw any further ligi." on the subject of mi)rplioli>gv; but for 0. nxtiiis I am more hope- ful, Nevertheless, aside from further developmental proofs, I incline to the view that 0. cinnamonuii possesses the most primitive type of central cylinder. I again recapitulate the reasons : 1. The opposite view demands a very plastic central cylinder in one species alone, not differing very greatly in habit from the others. 2. There would still remain phenomena that the opposite view could not explain. " Bnoni.E : Stem ttructure in Schizwaceae, etc. Brit. Asm. Dover, iSqq. ^ KooDLE : On the anatomy of the Hymenophyllaceae. Ann. Hot. 14 : 4SS. 1900. 1*51 34 F^ULl: THE ANATOMY OF THE OSMUNDACEAE 3. There are no facts of development even in analogous cases to support the op;)ositc oi)inion. 4. The view adopted here demands only slight changes, and those are of degeneration, to explain all the phenomena. 5. There are precisely similar analogous cases of degenera- tion. 6. Within the species O. cinnamomea itself, every phase of degeneration except the entire disaj^pearance of internal endo- dermis is observable in suitable specimens. When wc attempt to orient the other species amongst them- selves, the task is more difficult, and of little imi)ortance. As already indicated, a closer study of development may afford more precise proofs. In the mature stems we have seen that O. regalis occasionally has sclerenchyma in the medulla, that there are ramular gaps, thouirl, usually small, and that the external endoilermis is well developed. In O. Claytonmna, on the other hand, sclerenchyma is never found in the medulla, ramular gaps are infrequent, and the external endodermis shows indications of degeneration. In neither of these i^j-ecics is internal endodermis or internal phloem present. Tiic probability, therefore, is tiiat in the genus Osmunda there is a series, 0. ciniuimomea possessing the most primitive type of central cylinder and O. Claytoniana the most degenerate, 0. regalis occupyiu',' a mitldle position, but nearer to the latter. It is merely interesting to note in passing that Professor Campbell concluded 39 from his study of the pro- thallia of^ 0. Claytoniana and O. cinnamomea, that the gametophvte of the former was more specialized in many particulars, in other words, was less primitive in ty|)e than the latter. Ftfth, does a study of the vascular system help to determine the phylogenetic [wsition of the Osmundaceae ? It was stated at the beirinning of this paper that botanists have regarded the Osmundaceae as possessing an anomalous form of central cylinder among the Filicales, their reason being thai it seemed to present more of the features of a central cylinder such as is typical for dicotyledons, that is, a medullated monostele in Van Tieghem's terminology. In determining the position of the family, therefore, in any natural system of "Cami'BKLi.: On tlio pruihallium and embryo of O, Claytaniana ami O. cimna- mamta. Ann. Bot. 6 : 49. 1661 « FAULL: THE ANATOMY OF THE OSMUNDACEAE 35 classification, it was hopeless to try to reconcile this single (licotvlcdonous character with the remainiiijf filieinean characters, and so the vascular system in the family .-^s regarded as anomalous. It is fair to note that Zenctti dissentt from the pr- v ''ling view, and evidently for the reason that 1 I'.Mclieu s jn j value to the nature of the central cylinder ! ■ 'ii 'hi: nhyh genetic standpoint. Hence he sc)U'.,'ht to fun! the same type amongst the vascular cryptogams. He rejected the ordinary fern type because it is "polystelic," and the lycopod type because there is no pith.obvioi siv overlooking Si/aj^i/u/Zif /ih-riL^d/i!. Phylloglos- sum, etc. So finding no living form with w iiicii comi)arison ould be established he turned to paleophytology. Among the I.e|)i- dodendraceae he found the prot.ityiie sought for, especially in such of these fossils as L. If. cit. [671 36 FAUIL: THE ANATOMY OF THE OSMUNDACEAE SUMMARY OF OBSERVATIONS. I . An internal cnilodermis has been demonstrated in Osmunda antuitnomea, but in none of the other species examined. This inter- nal endodermis is in textural continuity with the external cndo- dermis throu^rh branch ga,.s, and sometimes through foliar gaps. 3. Internal phloem has been found in 0. cinnamomea in the region of branching. This is continuous with the cxter. .1 phloem throuirh ranuilar jjaps. 3. The external phloe n of the Osmundaceac forms a contin- uous cylinder, a fact which De Bary has stated for 0. ngalis; and IS not broken opi)osite the medullary rays as Strasburger has affirmed of the same species. Isolated 'sieve-tubes have been found in the medullary rays of 0. cinnamomea. 4. The xylem forms a cylinder broken only by foliar and ramular gaps. 5- Brown sclerenchyma has been shown to be usually present in the medulla of O. chw„momc.7, not uncommonly in regalis and not at all in O. Claytom.ma. It occurs likewise in Todea borhara. but has not been observed in T. suf>erbt7. 6. The medullary and cortical tissues of the Osmundaceac are histologically cjuivalent. Brown sclerenchyma, which is not an mtrastelar tissue in other ferns, occurs in both medulla and cortex; and in O. dnminwmc,^ the brown sclerenchyma of the medulla is in continuity with that of the cortex. 7- In 0. cinnamomea the typical ramular gap is one through which internal and external endodermis, internal and external phloem, cortex, and medulla connect. Kvery st,age of degen- eration has been observed in O. cinnamomea, however, down to the completely closed steles. O. rcgalis has a gap in the wood only, and O. Claytonuiim usually none. S. The so-called " <|uergcstreckte Zellen" pointed out by DeBary m O. rcgalis, and more fully described hy Zenetti h ive been found in all the species studied. They .are sieve iubes I'ossessm- all the characteristic features of sieve tubes even that of callus plu-s. Their irregularity of orientation is sh.ired by the other peripheral tissues of the central cylinder, and is apparently due to disturbance caused by the exit of the large leaf traces. Q. Callus plugs have been demonstrate.! in the sieve tubes. (68) « FAULL : THE ANATOMY OF THE OSMUNDACEAE 37 10. A study of the growing point has further shown that the " querj^estreckte Zellen" and the typical protophloem are of the same kind ; but it has failed to verify Strasburger's statement that -..le pcricycle and the cndodermis ari-e from a common maternal layer. 1 1. The phloem forms a continuous sheath in the leaf. 12. The root possesses a protostclic, diarch, occasionally triarch, vascular axis. SUMMARY OF CON'CLUSIONS. 1. The internal cndodermis in 0. cinnamomea is to be regarded as phloeotcrmal in nature, a fact denied by Strasburger in homologous cases. 2. The medullary and cortical tissues seem to be morpho- logically eijuivalent. 3. Observations on the anatomy of the Osmundaceac have been confined heretofore to the cosmopolitan 0. regalis, and the subtropical Todcas. F"rom these observations it was concluded by Van Ticghem that this family possessed a type of central cylinder anomalous among the vascular cryptogams, a .^pe (the medullated monostelic type) peculiar to the phanerogams. The writer dissents from this view. It appears to be the case that the central cylinder of O. cinnamomea is not medullated monostelic, for the nicdiilla is obviously extrastclar. Further, it cannot be regarded as gainodesmic on account of the topo- graphical distribution of the phloem. The most obvious inter- pretation seems to be that it is a degenerate form of the amphi- phloic siphonostelic type of central cylinder (polystelic type of Van Tieghem) . 0. cinnamomea, O. regalis, 0. Claytoniana form a scries arranged in order of degeneration of their central cylinders, and the same is true of T. harbara and T. superba. The present research was carried on in the Biological Depart- ment of Toronto University under the direction of Dr. E. C. Jeffrey, to whom I wish here to e.xpress my obligations for his advice throughout. My thanks arc due to Professor R. Ramsay Wright for the facilities afforded in the department. For some of the material used I am indebted to Mr. Oakes Ames, Assistant Director of the Botanical Gardens, Harvard University ; Sir (691 38 TAULL ; THE ANATOMY OF THE OSMUNDACEAE William Thistleton Dyer, Director of the Royal Gardens, Kew ; Dr. Brodie, Toronto ; and Mr. R. B. Thomson, B. A. Univf.rsity of Toronto. m. r., medullary ray. PLA TE x/y. Fig. I. Transverse section of the stem of Osmunda cinnamomea. Fig. 2. Transverse section of part of central cylinder of O. cinnamomea. Fig. 3. Tangential section of O. regaU%. Fir,. 4. "Quergestreckte Zelle" of T. barbara, showing sieve plates and c.ilhis I'lugs. I'u;. 5. Transverse section of leaf trace of O. ClaytOHiana near the grow- ing point. Fig. 6. Transverse section of part of central cylinder of O. cinnamomea. Fig. 7. .Sieve tubes of T. barbara, showing sieve plates, "globules bril- lants," and callus plugs. Fig. 8. Transverse section of part of central cylinder of O. Claytoniana. PLA TE XV. Fig. q. Tangential sectioi; of Todea barbara, showing " quergestreckte Zellen." KiG. !o. Transversa' section of 0. immediately above point of ramihc.uiiiii, showint; open branch gaps. Ki(i. 11. Transverse section uf O. iinnaniomea through nearly the same region in .mother plant. Fig. 12. Transverse section of same plant as in fig. 11, but lower down. Fig. 1 3. Transverse section of same plant as in fig. i3, but lower down. Fig. 14. Transverse section of the central cylinder of O. cinnamomea, showing internal endodermis and brown sclerenchyma in the medulla. FlO. 15. Transverse section of central cylinder of (). lifnamomen, show- EXri..\.N.Vri().\ OK I'l.AIKS .XlV XVll. AhireviatioHt uttj. cp, callus plugs. c, cnrtc.x. e. c, external co'tex. ;. <■., internal cortex. e, endodermis. e. ("., external endodermis. (. e., internal endodermis. ll, leaf trace, medulla. /. pericvcle. //;, phloem. /. /*/;., protophlocm. fx, pr()tox)lem. gu, " ('uergestreckte Zellen." r, root. s. s. s., strands. sc, sclerenchyma. X, xylem. P/.A TE xvr. Facu: the anatomy of the osmundaceae 39 FlO. i6. Transverse section o{ the stem of O. cinnamomea in the region of forking, showing absence of ramular gaps. Fig. 17. Transverse section of the stein of O. Claytottiana. Fio. 18. Transverse section of the stem of O. Claytoniana in the region of forking. Fio. 19. Transverse section of the stem of O. rtgalis in the region of forking. Fio. 20. Transverse section of the central cylinder of O. regalis, show- ing brown sclerenchyma in the medulla. FLA TE XV//. Fig. 21. Transverse section of the centra! cylinder of O. cinnamomea, showing internal phloem. Fig. 22. A part of the central cyliiuler nf O. minumomt-a shown in fi,i;. 9t more hifjhly magnified. Fig. 23. A trai^verse section of the young sporophyte of O. Claytoniana, showing one foliar gap, und the corresponding leaf trace opposite. Fio. 24. Transverse section of the stem of T. barbara, showing brown sclerenchyma in the medulla. Fio. 25. Transverse section of a part of the stem of T.barlara nearerthe growing point. Fig. 26. Transverse section of I.yi;inoii,iulron Oldluimium, showing a leaf gap, a leaf trace opposite, strands of sclerenchyma in the meciiiUa, .and strands of primary xylem centrad of the cylinder of secondary xylera. [71] FAl'LL on OSMl NDACKAi: BOTAHICAL GAZBTTB, XXXtl PLATE XV FAULL on OSMUNDACEAE. mOTAHICAL GAMMTTM, XXXII PLATS XVI FAULL OSMUNDACEAE. FAULL on OSMl'NUAtF.AK.