CHAPTER XXXIII.. Cordaitales.

CORDAITALES.

A. =POROXYLEAE=.

=Poroxylon=. Renault.

In 1879 Renault briefly summarised the anatomical features of some silicified vegetative shoots from the Permian of Autun for which he instituted a new family, the Poroxyleae. The more complete account contains a description of two species, Poroxylon Boysseti and P. Duchartrei: the latter was afterwards recognised as a stem of Heterangium. Renault considered this new genus to be closely allied to Sigillaria and Sigillariopsis and pointed out its resemblance to Cordaites. Additional species have since been described but as yet the genus has not been found outside France in Permo-Carboniferous strata of Autun and the St Étienne district. The results of a more detailed investigation of the anatomy of the genus were published by Bertrand and Renault in 1882 and since then Bertrand, Renault, and Scott have added to our knowledge of this interesting type. In several respects Poroxylon stems present a striking resemblance to Lyginopteris, but the recent discovery of the genus Mesoxylon has given greater significance to the characters in which Poroxylon agrees with representatives of the Cordaitales. Our knowledge of the genus, though exceptionally full with regard to the anatomy of vegetative shoots, does not include any precise information as to the reproductive organs.

The slender cylindrical stems, not exceeding 2–3 cm. in diameter in specimens so far recorded, bore large broadly linear leaves similar in form and venation to those of some species of Cordaites which were attached singly to slightly swollen nodes separated from one another by internodes several centimetres long. The base of the rather fleshy lamina passes imperceptibly from the narrow lower portion into a tangentially expanded petiole which forms a decurrent ridge on the stem. Axillary buds frequently occur. Little is known of the leaf-impressions, but if Grand’Eury is correct in his identification of certain specimens from French Stephanian beds as the leaves of Poroxylon, the lamina reached a length of 1 met. and a breadth of 15–20 cm. In habit the stems probably resembled some of the larger-leaved Bamboos. The only evidence bearing on the nature of the reproductive organs is furnished by Grand’Eury who believes that some Rhabdocarpus seeds and bractless inflorescences associated with the leaves assigned to Poroxylon belong to that genus.

The single cylindrical stele has a relatively large solid pith, the perimedullary region being characterised by the occurrence of a row of primary crescentic strands of centripetal xylem of exarch type, though not improbably in some cases slightly mesarch, varying in size and shape and forming single or paired bundles. These strands represent the xylem of collateral leaf-traces similar to those of Lyginopteris but differing in the absence of well-defined centrifugal elements: the curved form of some of the xylem strands gives them an appearance similar to that of the leaf-traces of Lyginopteris. The leaf-traces, except in the lower part of their course through the pith, are double and pass through several internodes before the centripetal tracheids die out. The secondary xylem (fig. 463) is manoxylic and very similar to that of Lyginopteris though rather less parenchymatous. The secondary phloem and cambium are often very well preserved. No endodermis and no distinct pericycle has been recognised. The cortex is parenchymatous and, like the pith and to some extent the phloem, contains numerous secretory sacs; in the outer cortex the presence of hypodermal strands is a prominent feature. At an early stage in the growth of the stem a deep-seated phellogen forms secondary tissue both externally and internally and decortication ensues.

The bundle of each leaf-trace is accompanied by an arc of secondary centrifugal xylem as it passes through the secondary wood and this is retained in the leaf except in the finer veins. After entering the petiole the leaf-trace branches and an arc of bundles is produced, the concave side facing the upper surface of the thick lamina (fig. 464, A). Further reference is made to the structure of the leaves in the description of Poroxylon stephanense. The specimens of roots so far described are characterised by a diarch plate of primary xylem and two masses of secondary vascular tissue separated by two medullary rays opposite the protoxylems. Bertrand mentions the occurrence of roots of P. stephanense with more than two protoxylem strands. The phellogen was produced in the pericycle as in the roots of recent Gymnosperms. It is suggested by Lignier that some silicified rootlets from Grand’ Croix (Loire) described by him as Radiculites reticulatus and at first compared with roots of Sequoia may belong to some Cordaitalean plant, possibly Poroxylon.

Poroxylon Edwardsii Renault.

This species affords a good illustration of the generic characters already summarised. The strap-like leaves are fleshy and the occasionally forked, parallel or slightly divergent, veins are embedded in a homogeneous mesophyll with hypodermal strands of mechanical tissue. The pith consists of parenchyma in vertical series with scattered secretory sacs and differs from that of Cordaites and Mesoxylon in the absence of transverse discs. There are 13 primary-xylem strands close to the inner edge of the secondary wood: the centripetal tracheids are scalariform or have multiseriate pitting like that in the secondary xylem. The structure of the leaf-traces is clearly shown in fig. 464: the double trace seen in fig. 464, C, has two protoxylem-strands accompanied by some parenchyma, and these are almost enveloped by the metaxylem tracheids which abut on the secondary wood. At this stage in its course, that is just before bending outwards, the centripetal xylem reaches its maximum development and the trace forms a prominent and broad twin-strand in striking contrast to the two narrower and tangentially extended strands shown in fig. 464, E, D. Each of these strands with a single protoxylem-group would at a higher level assume the broader and more compact form and contain two protoxylems as in fig. 464, C. The tracheids of the secondary xylem have 4–7 alternate rows of contiguous alternate pits on the radial walls: the medullary rays are 2–3 cells broad and may be 60 cells deep. According to Renault several small oblique pits occur on the radial walls of the ray cells. The secondary phloem, separated by a normal cambium from the xylem, forms a broad band of sieve-tubes with lateral sieve-plates like those in Medullosa anglica alternating with tangential rows of parenchyma. The cortex is relatively narrow and in older stems is chiefly occupied by secondary tissue formed from deep-seated phellogens.

Poroxylon Boysseti Renault.

The stems of this species agree closely with those of P. Edwardsii, the chief difference being in the structure of the secondary phloem which does not show the regular concentric alternation of sieve-tubes and parenchyma.

Poroxylon stephanense Bertrand and Renault.

This the oldest species, from Stephanian beds at Grand’ Croix, differs in no essential features from the other representatives of the genus. It is from a study of the leaves of this type that Bertrand and Renault have obtained most of the facts with regard to the anatomy of Poroxylon foliage. In the median region of the fleshy leaf the bundles are characterised by a comparatively large amount of centripetal xylem accompanied by a considerable development of secondary centrifugal tracheids: the bundles are connected laterally by both centripetal and centrifugal xylem and thus at certain levels in the lamina the vascular tissue has the form of a continuous plate (fig. 464, A, B). The veins become independent on branching and near the edge of the lamina they consist only of primary elements. Secretory sacs of elongated form are scattered in the homogeneous mesophyll, and thick stereome-strands underlie the epidermis. The epidermal cells are rectangular and rows of stomata occur on both surfaces.

B. =CORDAITEAE=.

=Cordaites=. Unger.

A preliminary statement with regard to nomenclature may serve to remove possible misconceptions in connexion with the application of the generic name Cordaites. It has been the general practice to apply this name to certain forms of linear leaves which are particularly abundant in Carboniferous and Permian strata in Europe and North America, and in recent years a few palaeobotanists have substituted Cordaites for Noeggerathiopsis as the more suitable designation for Permo-Carboniferous specimens abundant in the rocks of Gondwana Land. It has been customary to assign to Cordaites certain reproductive shoots, seeds, and stems described under the generic names Cordaianthus, Cordaicarpus, Cordaicladus, Cordaioxylon, etc. Stems agreeing anatomically in their main features with those of recent Araucarineae have long been attributed to Cordaites, but a few years ago a new type of stem was discovered which, though almost identical with that of Cordaites, is distinguished by the character of the primary xylem. For this new type the name Mesoxylon was proposed. Nothing is known as to the reproductive organs borne on Mesoxylon stems, but the leaves are externally at least indistinguishable from those referred to Cordaites. It is therefore obvious that when we apply the name Cordaites to leaves or other plant-organs, under that designation are undoubtedly included specimens belonging both to Mesoxylon stems and to stems with the characters of Cordaites (Cordaioxylon). Further research may enable us to subdivide Cordaites into more precisely defined types distinguished by well-marked morphological characters, but at present the only course would seem to be to restrict the term Mesoxylon to petrified stems exhibiting the features of that genus and to retain Cordaites as a comprehensive designation in accordance with the general account of the genus given in the following pages. This widely distributed and mainly Palaeozoic genus is especially well represented in the coalfields of France where in some localities it contributed largely to the formation of seams of coal, and it is chiefly from the researches of French Palaeobotanists that our knowledge of its morphology is derived. Cordaites has shared the fate of most other abundant fossil plants in the distribution of its disjuncta membra among several genera and classes, but on the whole the information that is now available enables us to reconstruct the complete plant with a greater degree of confidence than is usually attainable.

Cordaites may be described as a forest-tree closely resembling in habit and probably in size the recent Conifer Agathis, more especially such species as A. macrophyllus, A. vitiensis and others with leaves considerably longer than those of the Kauri Pine (A. australis). The main stem reached a considerable height before giving off scattered branches bearing spirally disposed, sessile, and often crowded leaves like the foliage of Agathis. The absence of any evidence of a two-ranked arrangement of leaves on lateral branches suggests a general tendency towards a vertical rather than a horizontal direction of growth. The sessile and closely set leaves for the most part of leathery texture vary considerably in length and breadth in different types (figs. 466–472): in some the broadly linear lamina with its parallel veins and perfectly constructed I-shaped girders (fig. 465) reached a length of nearly 100 cm., in shape like the blade of a straight broad-sword or the leaves of a Yucca, torn by the wind into strips; in other forms the lamina is shorter and more obovate, while in some the leafy shoots must have looked like slender stems of the smaller-leaved Bamboos. There is no proof that young vegetative branches with their spirally rolled leaves were protected by bud-scales, but some oval triangular scales (fig. 468, C), occasionally found in association with larger foliage-leaves, may have served that purpose. The branches from which leaves had recently fallen at the time of fossilisation are characterised by transversely elongated oval scars, occasionally showing a slightly curved row of pits like the marks of leaf-traces on the scars of a Horse Chestnut, sometimes terminating a feebly projecting decurrent leaf-cushion (fig. 466, C). The leaves persisted for a comparatively long period as in Araucaria imbricata, and on older leafless branches the scars are transversely stretched; the leaf-cushion loses its individuality and eventually the development of secondary cortical tissue causes the exfoliation of the superficial bark.

In the form and structure of the fertile shoots Cordaites parts company with Agathis; the trees bore no cones in the ordinary sense, but unisexual inflorescences—whether on one plant or on different individuals is uncertain—were produced in the axils or from a supra-axillary position as compound spikes or compact racemes. Both the longer female shoots and the shorter and more compact male branches are constructed on a similar plan. The ovulate inflorescence may exceed 30 cm. in length (fig. 479); a stout axis bears two-ranked linear bracts subtending short lateral bud-like shoots with one or several sessile or stalked ovules (fig. 480) between the sterile scales. The seeds are platyspermic and agree much more closely with those of Cycads and Gingko than with the seeds of Conifers. The male inflorescence is on a smaller scale, in habit not unlike the elongated male shoot of Cephalotaxus pedunculata and some other Conifers; each bract subtends a small oval bud composed of imbricate scales and highly modified microsporophylls borne singly or in clusters (figs. 481, F; 482). A microsporophyll consists of a comparatively long pedicel bearing at its apex a few long microsporangia. The term microsporophyll implies a morphological interpretation which is not accepted by all palaeobotanists, some of whom prefer to regard the microsporangia as stamens or microsporophylls reduced to their simplest terms and sessile on an elongated flower-stalk.

The stem agrees very closely in its more important features with that of an Araucaria or an Agathis: the primary xylem forms the inner surface of the thick cylinder of secondary wood, merging gradually into it as in recent Conifers; there are no separate bundles of primary centripetal xylem. The medullary rays are narrow: in other words the secondary xylem is of the pycnoxylic type. The pitting of the tracheids is Araucarian and, as in Agathis, the leaf-traces arise as twin-bundles. The pith is larger than in the Araucarineae and more homogeneous in structure; it shares with the pith of Juglans and some other recent plants an almost constant tendency to assume a discoid structure. Anatomically the leaves agree more closely in the structure of the vascular bundles with Cycads than with Conifers though there are points of contact with both of these classes. The roots branch freely and their horizontally extended arms (figs. 468, A; 478) suggest growth in swampy ground; anatomically they conform to the recent Gymnospermous type and there is good evidence that in some cases fungal mycelia lived symbiotically in the cortex of coralline rootlets.

Sternberg figured some leaves of Cordaites from Carboniferous rocks in Bohemia under the generic name Flabellaria in the belief that they belonged to a Palm. Brongniart substituted a new name Pycnophyllum on the ground that Corda had disproved the supposed relationship with Monocotyledons. The name Cordaites was instituted by Unger, his definition being based on leaf-form as well as on stem-anatomy. It has recently been proposed to revive the forgotten designation Pycnophyllum, but the reasons given are hardly likely to induce botanists to discard the familiar generic name which perpetuates the memory of Corda. As already pointed out, the name Cordaites, even though employed in what has always been regarded a legitimate sense, is no doubt often given to specimens of some other allied member of the Cordaitales which can only be recognised as such in the case of more completely preserved material. The naming of wood of the Cordaitean type, but which may equally well belong to another genus, raises a difficult question: if there is satisfactory evidence from collateral sources that the wood is that of a Cordaites Grand’Eury’s name Cordaixylon or Schenk’s form Cordaioxylon may be used, though there seems to be no adequate reason against the use of the name Cordaites. If there is no confirmatory evidence available and it is impossible to say whether the wood is that of a Conifer or a Cordaites, or some other plant with the same type of secondary xylem, Endlicher’s term Dadoxylon is most conveniently employed. The confusion liable to follow from the use of the two generic names Dadoxylon and Araucarioxylon for wood of the same type differing only in geological age is an argument in favour of extending Dadoxylon to all specimens having certain anatomical characters, which cannot be certainly assigned either to the Araucarineae or the Cordaitales, irrespective of geological age. The term Cordaicladus sometimes applied to branches is hardly necessary, but the subgeneric names Eu-Cordaites, Dory-Cordaites, and Poa-Cordaites, instituted by Grand’Eury for different forms of leaf, are frequently employed and serve a useful purpose as descriptive terms though the characters which they connote are of small importance and by no means always well defined or constant. For inflorescences it is customary to adopt the name Cordaianthus suggested by Grand’Eury as a substitute for Antholithus and some other terms. The same author uses Rhizo-Cordaites for roots.

The nomenclature of seeds is more difficult: in a few instances seeds occur in organic connexion with Cordaitean shoots, but there is no doubt that many platyspermic Palaeozoic seeds preserved as detached fossils belong to Cordaites or some other member of the group. The difficulty is that in the present state of knowledge we cannot definitely determine in many cases whether a seed is Cordaitean or whether it belonged to a genus of Pteridosperms. For this reason the account of several seeds that were probably borne on Cordaites or some allied genus is given in a later chapter devoted to Gymnospermous seeds. There is no doubt that under the generic names Cardiocarpus, Cordaicarpus, and Samaropsis are included true Cordaitean seeds, though it would be incorrect to say that all the seeds so named belong to members of the Cordaitales.

Cordaites reached its maximum development in the Carboniferous and Permian periods; the genus or some closely allied types persisted into the Triassic and Rhaetic periods, and there is reason to believe that the group was represented in some post-Rhaetic floras. The genus is one of many remarkable examples of the high degree of specialisation attained by Palaeozoic plants. The complex mechanisms represented by Cordaites and similar types give force to the conviction that we cannot hope to penetrate below the higher branches of the genealogical tree which had its roots in a period of the earth’s history inaccessible to botanical investigation. The plants of the present age are to a large extent the result of evolutionary tendencies more correctly described as the result of degeneration or simplification than as the latest phase in a series composed of a succession of types gradually growing in complexity. Cordaites is essentially a generalised type, a composite product of an age characterised by an activity in the elaboration of the complex from the simple. Botanical records furnished by the geological series available for investigation furnish evidence of the sorting of characters among gradually diverging races and of changes in plant-organisation tending towards simplification and increased efficiency.

Cordaites, using the generic designation in a wide sense, occurs in Carboniferous and Permian strata in Europe, North America, and China; it is recorded from several localities in Russia and Siberia for the most part from Permian rocks, from Permo-Carboniferous (Lower Gondwana) beds in India, Australia, South Africa, and South America. Wood agreeing generally in the structure of its secondary tracheids with that of Cordaites is represented in Devonian rocks, and there can be no doubt as to the existence of Cordaitalean plants in pre-Carboniferous floras. It is represented in the Rhaetic flora of Tonkin and has recently been discovered in strata probably of Rhaetic age in Mexico.

+Leaves.+

It is important to recognise the fact that leaves included under the generic name Cordaites were in many cases not borne on stems or branches with the anatomical characters of Cordaites. Scott in his account of the genus Mesoxylon says, ‘I feel no doubt that most of the British specimens of Cordaitean leaves really belong to Mesoxylon, which is a much commoner type of stem in the Coal Measure petrifications than that of Cordaites itself.’ Some of the Cordaitean leaves were probably attached to stems of the Poroxylon type and it is not improbable that, as investigations are extended, additional genera of vegetative shoots will be discovered provided with leaves similar at least in external characters to those which it is customary to refer to Cordaites. In the present state of our knowledge we cannot make use of anatomical characters as criteria by which to distribute the foliage of the Cordaites form among the genera Cordaites, Mesoxylon, and Poroxylon, using these names as designations of certain types of anatomical structure. The specimen reproduced in fig. 465 is in all probability a piece of a leaf of Cordaites principalis, but on anatomical grounds Miss Benson has made it the type of a new species, C. Felicis, and more recently Scott has brought forward evidence supporting the view that it is a leaf of Mesoxylon. As, therefore, neither impressions nor petrifications of Cordaitean leaves can in the great majority of cases be referred with confidence to their respective genera of stems, pending fuller information the only course would seem to be to use the name Cordaites in a comprehensive sense indicating in special cases where evidence is available the more precise systematic position of the specimen. The classification of Cordaitean leaves proposed by Grand’Eury is based partly on the form of the lamina and in part on the equality or inequality of the ‘veins.’ The actual veins, which are embedded in the fairly thick mesophyll, do not directly affect the superficial ribbing on the carbonised impression of the leaves and, as seen in fig. 465, the most prominent hypodermal strands of supporting tissue which would appear as the main veins or primary ribs on an impression do not correspond in position with the vascular bundles. Although in some cases the largest stereome-strands coincide with the veins, forming the upper and lower parts of I-shaped girders the centre of which is occupied by the veins, this is by no means always the case. Grand’Eury has drawn attention to the difference between the upper and lower surface of some carbonised leaves: in C. crassifolius (fig. 468, D) there are five to seven finer ribs between each pair of primary ribs on one face while the other shows ridges and grooves with a rib corresponding to each. Attention is called on a later page to the variable character of the ribbing even on different parts of the same lamina. The lower surface of the leaf, seen in section in fig. 465, would show a number of approximately equal ribs, or possibly primary ribs (midway between the veins) separated by two interstitial ribs, while on the upper face there would be three rather smaller secondary ribs. In a section of a leaf called by Renault C. crassus, a specific name used also by Lesquereux for an impression of a leaf originally described by Goeppert as Noeggerathia crassa, there are deep stereome-strands between the veins next the lower epidermis alternating with single smaller strands, while on the upper surface the hypodermal strands occur only immediately above the veins. In a section figured by Felix from North Germany as C. robustus, the hypodermal stereome forms continuous bands; on the upper face the bands are uniform in thickness but next the lower epidermis they form a series of ribs.

Grand’Eury’s subgeneric terms Cordaites, Dory-Cordaites, and Poa-Cordaites have therefore very little value as regards differences in the ribbing of leaf-impressions: the large size of leaves included in Dory-Cordaites and the more acute apex of the lamina as compared with the obtuse apex of smaller leaves of Cordaites are features of limited application and of minor importance as diagnostic characters. The name Poa-Cordaites is, however, usefully employed for the narrower linear leaves with an obtuse apex.

The structure of a Cordaites leaf is clearly shown in fig. 465; the lamina is approximately 1 mm. thick and there are about 30 veins in a breadth of 2 cm. Strong I-shaped girders with the webbing composed of thick-walled cells divide the mesophyll into rectangular compartments: the intervening hypodermal strands differ in number and size on the two faces. The epidermis is not preserved: specimens of other leaves show that the stomata occur in rows on the lower surface. The mesophyll shows no differentiation into palisade and spongy parenchyma, and in this respect the leaf agrees with many other forms; but in some leaves the palisade-tissue is well developed, as in C. lingulatus Ren. The central region of the lamina consisted of lacunar tissue, portions of which are preserved, with a more compact sheath of parenchyma enclosing each vein. In some leaves there is a narrower sheath of thick-walled cells more sharply contrasted with the mesophyll. The vascular bundles agree in structure with those in the rachis of a Cycadean frond more closely than with the veins of an Araucarian or other Coniferous leaf. The xylem consists mainly of centripetal elements which form a deltoid strand with the protoxylem at the apex, and in close association with this is a larger or smaller amount of narrower centrifugal tracheids: in the section shown in fig. 465 the centrifugal xylem may extend all round the centripetal tracheids, but it usually forms an irregular arch with its base attached to the sides of the larger tracheal strand, cp, separated, except at the base of the arch, by a small amount of conjunctive parenchyma from the centripetal xylem. The phloem is not preserved and is represented only by a few patches, ph, below the centrifugal tracheids. Dr Benson in her account of this type of leaf gives additional details and compares the anatomical features with those in other species. The dual nature of the xylem like that characteristic of recent Cycads has usually been regarded as a definite feature of Cordaites leaves; Dr Stopes, on the other hand, interprets the narrower tracheids (occupying a position similar to those in fig. 465) in some sections of a leaf identified with C. principalis from Grand’ Croix, as an inner sheath of transfusion elements (‘primitive transfusion tissue’) possibly derived from the centripetal xylem with which it is clearly connected at the sides precisely as in fig. 465; but in the Grand’ Croix leaf the phloem is enclosed within the sheath of narrower tracheids and not external to it as it is in the section shown in fig. 465 and in a section of C. lingulatus figured by Dr Stopes. It is, however, difficult to recognise any fundamental difference between the ‘inner transfusion tissue’ and centripetal xylem. The cells of the outer sheath in Dr Stopes’s specimens of C. principalis have bordered pits on their walls and this character is mentioned also by Renault in other specimens.

Prof. Lignier has described the structure of fragments of adult leaves from the Stephanian of Grand’ Croix (Loire) which he refers to Cordaites lingulatus, and the same author gives an interesting account of the anatomical features of a bud of the same species. The bud, which resembles in general appearance that of Dolerophyllum (fig. 430, p. 133) is 3 cm. long, oval in transverse section—as the result of compression—and consists of four convolute leaves and a piece of a fifth. The outer leaves have 75 to 80 veins: the inner laminae are sinistral in their curvature while the three outer leaves are dextral. In the second, the first in which the tissues are recognisable, the small desmogen-strands afford some evidence that the phloem preceded the xylem in the order of differentiation as is often the case in recent plants. The first tracheids occur almost in the centre of the desmogen-strand and to these are added the other tracheids of the centripetal xylem, the oldest elements being spiral, the next scalariform and the later tracheids reticulate. The centrifugal xylem is formed at a later stage, and at about the same time are differentiated the elements called by Dr Stopes the inner sheath and by Lignier the ‘bois diaphragmatique.’ Lignier also describes the development and structural features of the other tissues of the young leaves and compares the anatomical features of the French leaves with those of Cordaites Felicis described by Prof. Benson.

The main features of Cordaites leaves are (i) the presence of two kinds of xylem in the veins, the larger centripetal tracheids, or chief water-conducting elements, and the narrower tracheids, in some cases attached to the sides of the centripetal xylem, in others forming free groups, usually between the protoxylem and the phloem, but sometimes enclosing the phloem; (ii) the frequent presence of a well-defined sheath of cells round each vein composed of comparatively thick-walled elements comparable with the transfusion-tracheids in Conifers; (iii) the presence of lacunar tissue in the centre of the mesophyll and in some cases of transversely extended tracheids similar to those in some Podocarp leaves; (iv) a well-developed system of stereome-strands and I-shaped girders. The structural features on the whole suggest a xerophilous type, and the frequent absence or feeble development of palisade tissue points to diffused rather than to brilliant sunlight.

The considerable range in size and form among Cordaitean leaves as well as the obvious dependence on conditions of preservation or growth of such a relatively unimportant feature as the presence or absence of the so-called false or interstitial veins—the variability of which has been demonstrated in several instances—renders specific determination exceedingly difficult. The following species are briefly described rather with a view to illustrate the nature of the characters employed by authors than as implying the existence of so many well-defined types.

Cordaites principalis (Germar).

This species was founded on a large specimen from the Coal Measures of Wettin showing a fan-like cluster of longitudinally torn and partially overlapping leaves spread out in the position that would be assumed on the compression of a shoot with a close spiral phyllotaxis. This form of Cordaites is the most abundant in the British Coal Measures. The broadly linear lamina is characterised by an obtuse apex (fig. 466, A), a tendency to split into strips, close-set parallel ribs, the stouter ribs or veins separated from one another by 2–3 or it may be as many as 5 finer ribs or interstitial ‘veins.’ A statement by Weiss that in Germar’s type-specimen the longitudinal ribbing of the lamina is very imperfectly preserved confirms the scepticism that is justly felt as to the validity of this character as a satisfactory specific criterion.

The incomplete example shown in fig. 466, A, is 19·5 cm. long and has a maximum breadth of 3·5 cm., but the complete leaf was much larger and tapered gradually to the comparatively broad and slightly concave or amplexicaul base. The narrow elliptical proximal end of a specimen figured by Kidston from the Middle Coal Measures of Yorkshire is 2·8 cm. broad indicating that the tangentially expanded leaf-scars on a branch recently deprived of its foliage must have been a conspicuous feature. In his synonymy of this species Kidston includes Knorria taxina, a species founded by Lindley and Hutton on a piece of stem from the Coal Measures of Newcastle. The type-specimen, as Mr Howse states, is much larger than the published drawing and closely resembles in the decurrent leaf-bases with broad apices the piece of stem represented in fig. 466, C, which Kidston identifies as C. principalis. Geinitz refers to this species the seeds named Cordaicarpus Cordai (Gein.), but there is no evidence of connexion. Kidston points out that this seed is rare in Britain: he believes that Cordaianthus Pitcairniae (Lind. and Hutt.) is probably the inflorescence of C. principalis.

Cordaites principalis occurs in both Carboniferous and Permian strata. The leaves described by Lesquereux from Pennsylvania as C. Mansfieldi agree closely with C. principalis. Another similar or possibly identical form is represented by C. Ottonis Gein.

Cordaites borassifolius (Sternberg).

The leaves of this species, originally referred to Flabellaria, resemble those of C. principalis but differ in the ovate-lanceolate and less obtuse apex and in the presence of only one or rarely two finer striations between the stronger ribs. Corda’s drawing affords a good illustration of the crowded spiral disposition of the foliage comparable with that on an Agathis shoot. The lamina is usually 4–8 cm. broad but in exceptional cases may reach a breadth of 12 cm. The species occurs in the Coal Measures, especially in the Westphalian series and in Permian rocks. Feistmantel unites with this type Cordaianthus Pitcairniae (fig. 480, A), but as in other cases there may be a confusion between C. borassifolius and C. principalis. Leaves described by Lesquereux as C. communis are, as White says, not distinguished by any well-marked characters from this species. White figures some good examples of C. borassifolius from Missouri, reaching in one case a length of 40 cm., showing on the lamina the fructifications of a fungus, Hysterites cordaitis Grand’Eury. The leaves described from Canada and the United States as C. Robbii Daws. are closely allied to if not identical with Sternberg’s type. Among other species differing in no definite character from C. borassifolius is C. lancifolius described by Schmalhausen from the Permian of Russia.

Cordaites lingulatus Grand’Eury.

The leaves of this species are characterised by the obovate lamina and bluntly rounded or almost truncate apex; it affords a good illustration of the uncertainty of the ribbing as a diagnostic character. The lamina of a well-preserved specimen from the Blanzy coalfield described by Zeiller reaches a length of 35 cm. and a breadth of 10–11 cm. decreasing to 4 cm. at the base (fig. 467). In the lower part of the lamina Zeiller describes the ribs as unequal in prominence, the stronger ones being separated by 1–3 finer ribs, while in the middle and upper portions the ribs appear to be of equal size. Some of the finer ribs are due to folding of the lamina and are not represented, as are the ribs due to the presence of stereome-strands, by dark streaks in the detached cuticle.

Reference has already been made to the anatomical features of leaves of this species described by Lignier and other authors.

Cordaites grandifolius Lesquereux.

The leaves so named by Lesquereux, from the Coal Measures of Pennsylvania, are distinguished by the elongate cuneate lamina, which reaches a length of 38 cm. with a narrow base and a slightly rounded truncate distal end, 16 cm. broad, characterised by a few broad and shallow crenulations. By contrast with some American specimens in Dr Kidston’s collection Lesquereux’s figures convey an imperfect idea of the size of the leaf. A large leaf from the Coal Measures of Belgium described by Cambier and Renier as a new species of Psygmophyllum, P. Delvali, is perhaps identical with the American type; the lamina of sub-triangular form has approximately the same dimensions; the veins are numerous and repeatedly forked. The leaf is much longer than any known Psygmophyllum and the veins are much more numerous than in P. majus Arb., the largest representative of that genus. Palaeobotanists who have seen the type-specimen inform me that they have no doubt as to the Cordaitean nature of the Belgian specimens, which may be designated Cordaites Delvali. It is, however, not impossible that Psygmophyllum and Cordaites are allied genera: our knowledge of the former is limited to unimportant characters.

Cordaites (Dory-Cordaites) palmaeformis (Goeppert).

This Permian type, originally described by Goeppert as Noeggerathia palmaeformis, is characterised by numerous slender veins: according to Weiss there may be as many as 3–5 in 1 mm. The leaf is broadly lanceolate; it tapers gradually to an acute apex reaching a length of 80 cm. and a breadth of 10 cm. In habit the young foliage-shoots resemble those of C. principalis and C. borassifolius. Grand’Eury records the frequent association of Samaropsis seeds with this species; it occurs in Upper Carboniferous and in Permian strata and is recorded from a few British localities.

Cordaites (Poa-Cordaites) microstachys Goldenberg.

Weiss first figured this species from drawings supplied by Goldenberg at whose suggestion the name C. microstachys was adopted. The type-specimen consists of a slender axis bearing numerous narrow linear leaves and a few imperfect fertile axillary shoots. A specimen is figured by Kidston from the Upper Coal Measures of Radstock: it is a rare type in Britain. The species is readily distinguished from C. principalis and similar forms by the narrow lamina which varies considerably in length, rarely as long as 30 cm. and not exceeding 1 cm. in breadth. The apex is obtuse and the ribs are either equal in strength or 1–2 finer striae may alternate with the stronger ribs. The base of the lamina is 3–4 mm. wide and the leaf-scars have a slightly arched upper margin and an almost straight lower edge. The foliage of this species, generally regarded as identical with C. linearis Grand’Eury, bears a close resemblance to that of the Mesozoic genus Phoenicopsis from which it is distinguished by the occurrence of the leaves in bunches.

C. gracilis Lesq. is a similar type. The shoot on which Lesquereux founded his genus Desmiophyllum may perhaps be an example of Poa-Cordaites. Poa-Cordaites tenuifolius Schmal. from the Permian of Russia may be identical with C. microstachys.

As examples of other forms of leaf referred to Cordaites, though as in other cases without any proof of connexion with branches having the anatomical features of the genus, reference may be made to Cordaites circularis Grand’Eury from Gard (fig. 468, B) and a smaller leaf from the same locality compared with C. Lacoei (fig. 468, C) Lesq. Cordaites circularis is characterised by the almost orbicular lamina traversed by slightly spreading veins; it recalls some of the larger Cyclopteris pinnules of Pteridosperm fronds and is indistinguishable from some leaves assigned to the genus Dolerophyllum.

The species C. Lacoei was founded by Lesquereux on some detached specimens 3–12 cm. long and 1·5–5 cm. broad; it is by no means certain that a specimen referred by Grand’Eury with some hesitation to this species is Cordaitean.

The generic name Scuto-Cordaites was proposed by Renault for a specimen from Commentry consisting of a flattened branch bearing a few imperfectly preserved leaves. The surface of the branch shows semicircular leaf-scars on decurrent, spirally disposed leaf-cushions and bears a certain resemblance to a slender stem of a Clathrarian Sigillaria. The leaves of the type-specimen of Scuto-Cordaites Grand’Euryi appear to be broadly linear, 13 cm. long, the breadth gradually increasing from the base: a short distance from the proximal end the lamina is broken up into narrow segments; the veins are ·5 mm. apart with finer striations between them.

Some specimens from Pennsylvania made by Dawson the type of a new sub-genus and named Dictyo-Cordaites Lecoi agree in shape and arrangement with some species of Cordaites, but differ in an occasional anastomosis of the veins as in Psygmophyllum flabellatum. It is, however, impossible to determine the true nature of the fossils from the published figures.

+Cordaitean leaves from India, the Southern Hemisphere, and Siberia. Noeggerathiopsis, Feistmantel; Rhiptozamites, Schmalhausen; Euryphyllum, Feistmantel.+

{Cordaites aequalis Goeppert. {Cordaites (Noeggerathiopsis) Hislopi (Bunbury).

In 1845 Goeppert instituted the species Noeggerathia aequalis (fig. 469) and N. distans for incomplete broadly linear and obovate leaves, from Siberian Permian strata, having a contracted base and equal parallel veins. The specimens so named are no doubt specifically identical. Goeppert’s species N. aequalis has recently been carefully investigated by Zalessky who agrees with Kosmovsky in identifying it with Noeggerathiopsis Hislopi (Bunb.) and Rhiptozamites Goepperti Schmal. Schmalhausen had previously pointed out the probable identity of his species with Noeggerathia palmaeformis Goepp. (= Cordaites). The question of specific identity of these leaves from different localities and of other hardly distinguishable forms is of secondary importance; the main point is that they are all examples of Cordaitean leaves, Cordaites or some allied genus, and point to the existence of this group of Gymnosperms during Permo-Carboniferous times in Siberia, China, India, Australia, South Africa, and S. America, also in the Rhaetic floras of Tonkin and Mexico. The fragments from Devonian strata at Iguana Creek, Australia, named by McCoy Cordaites australis are probably pieces of the rachis of some large frond.

Wieland recently discovered Cordaitean leaves exhibiting a wide range in size and shape in the Mixteca flora of Mexico in the lower members of a series which extends from the ‘upper borders of the Rhaetic’ through the Liassic to the lower beds of the Inferior Oolite. These leaves are referred to Noeggerathiopsis Hislopi, and it is clear from an examination of photographs received from Dr Wieland, one of which is reproduced in fig. 470, that the Mexican Cordaites cannot be specifically distinguished from Bunbury’s type as represented by specimens described from India, South Africa, Siberia, Tonkin, and elsewhere.

The occurrence of Noeggerathiopsis is also recorded by Newberry from the Rhaetic series of Honduras.

Noeggerathiopsis. This genus was founded by Feistmantel for some leaves from Lower Gondwana rocks in India originally described by Bunbury as Noeggerathia (Cyclopteris?) Hislopi (figs. 470–472) and regarded by him as probably Cycadean. Several authors have added to our knowledge of this widely spread southern type and in many localities the leaves occur in association with platyspermic seeds of the Samaropsis or Cordaicarpus type, pieces of stems with Cordaitean leaf-scars, and petrified wood agreeing in the structure of the secondary xylem with that of European species of Cordaites. In some Permo-Carboniferous sandstones at Vereeniging, South Africa, stumps and spreading roots (fig. 478) resembling those described from France by Grand’Eury (cf. fig. 468, A) have also been discovered. A remarkable occurrence of roots and prostrate stems of some forest-tree was recorded some years ago in the bed of the Vaal river near Vereeniging where the surface of a seam of coal was exposed over an area of more than two acres. Large branched roots (fig. 478) spreading over the coal for a distance of several feet and thick stems 40–50 ft in length with very few branches and but little decrease in diameter afford a striking picture of a forest-floor. The frequent occurrence of Cordaites (Noeggerathiopsis) Hislopi in the associated strata suggests a reference of the stems and roots to that species. Moreover the structure of the secondary xylem of some petrified pieces of stem sent to me by Mr Leslie from Vereeniging agrees closely with that of a European Cordaitean stem.

The leaves of Cordaites (Noeggerathiopsis) Hislopi vary considerably in size, in some cases reaching a length of 80 cm. (fig. 471); the lamina tapers gradually from a short distance behind the obtuse apex to a relatively narrow base: in venation and form the leaves are very similar to those of C. principalis and other European and North American species. The specimen from India represented in fig. 472 shows several spathulate leaves attached in a close spiral to a branch. As White and Zalessky have shown, the stronger ribs are separated by less prominent striations indicating the presence of two sizes of hypodermal strands. The obvious resemblance between Noeggerathiopsis Hislopi and species of Cordaites has long been recognised and many authors have included Feistmantel’s genus in the Cordaitales. Prof. Zeiller preferred to retain the name Noeggerathiopsis as a precautionary measure, chiefly on the ground that the stomata appeared to be less definitely arranged in rows and more scattered than in the European leaves of Cordaites, and because of the absence of interstitial veins. We have as yet little information as to the arrangement of the stomata, but in view of the irregularity in stomatal grouping in recent leaves this feature is, perhaps, of minor importance. The presence of interstitial ‘veins’ has now been established in Indian and South American leaves. In a paper published in 1908 the name Cordaites was substituted for Noeggerathiopsis and Zalessky’s recent work supports this step. The description by Zalessky of the ribbing in Goeppert’s species Cordaites aequalis from Siberia shows how uncertain and variable a character the venation is even in different parts of the same leaf.

Cordaites Clerci Zalessky.

This species (fig. 469, D) was instituted for some small lanceolate or spathulate leaves from the Petschora basin (Adzva River) reaching a length of 6 cm. and a breadth of 1 cm. It is separated from Cordaites aequalis on the ground that the veins are more numerous, as many as 44 in a breadth of 1 cm.

A recent investigation by Miss Holden of the carbonised cuticles of some Indian specimens, sent to Cambridge by the Director of the Indian Geological Survey, and a comparison of them with preparations made from European Cordaites leaves, have revealed certain distinguishing features which support Zeiller’s view that the Gondwana-Land leaves, though similar superficially to those of Cordaites, are probably distinct. It is, however, impossible in many cases to obtain any information with regard to epidermal characters, and though it would seem probable that had we a fuller knowledge of the Indian and southern hemisphere plants represented for the most part by leaf-impressions well-defined distinguishing features would be recognised, the comprehensive name Cordaites may conveniently be retained on the ground that in the absence of well-preserved cuticles no satisfactory distinguishing features are exhibited by the impressions of Noeggerathiopsis.

=Phylladoderma.= Zalessky.

Phylladoderma Arberi Zalessky.

Zalessky founded this genus on some Permian leaves from the Petschora basin (Adzva River) which closely resemble those of Cordaites but are characterised by a coarser venation. The lanceolate lamina reaches a length of 18 cm. and a breadth of 4·2 cm.; the veins are 2 mm. apart and occasionally forked near the base of the leaf. The epidermal cells have straight walls and stomata are abundant on the lower surface. As Zalessky says, the systematic position of the leaves is uncertain though they are probably Cordaitean. The coarseness of the venation is a feature of minor importance and hardly worthy of generic recognition.

Rhiptozamites Schmalhausen.

This genus was instituted by Schmalhausen for leaves from beds in the Kusnezk basin regarded by him as Jurassic. These strata are now recognised as Permian and homotaxial with those from which Schmalhausen subsequently recorded the same species. The leaves, though smaller than many of the Indian and South African specimens of Cordaites (Noeggerathiopsis) Hislopi, may belong to that species. Zeiller and others definitely assigned the Russian leaves to Cordaites.

Euryphyllum. The Indian leaves for which Feistmantel proposed this name are, as several writers have pointed out, in all probability referable to Cordaites.

The general conclusion to be drawn from this imperfect summary of an extensive literature is that the employment of the generic names Noeggerathiopsis, Rhiptozamites, Euryphyllum, and others has tended to exaggerate the difference between the European and Southern botanical provinces during the Permo-Carboniferous period.

Scale-leaves, seeds, and stems.

The occurrence of small scale-like leaves of the type represented in fig. 468, C, in association with Cordaites (Noeggerathiopsis) Hislopi in India, Brazil, Siberia, and elsewhere may mean that these organs are scales of large foliar buds. The occurrence of several forms of platyspermic seeds, in some cases apparently identical with European forms and sometimes distinct types, in close association with Cordaites (Noeggerathiopsis) Hislopi has already been mentioned. Examples of such seeds are described in Chapter +xxxv.+ under the genus Samaropsis.

There are very few satisfactory examples of Cordaitean branches from the southern hemisphere. Schmalhausen figures good specimens from Siberian rocks from which his Rhiptozamites leaves were obtained. Branches with spirally disposed leaf-scars figured by Zeiller from the Rhaetic of Tonkin closely resemble Cordaicladus. Feistmantel’s drawing of a fossil from the Karharbari series, compared by him with a Fern rhizome, may be a Cordaitean branch, and the same author describes a stem from New South Wales as Caulopteris Adamsi which bears a close resemblance to a branch of Cordaites. Similarly a leafy shoot described from India by Zeiller as Araucarites Oldhami may be compared with branches of the Poa-Cordaites type.

+Stems.+ i. Pith-casts.

=Artisia.= Sternberg.

A character to which authors tend to attach excessive importance as a diagnostic feature is the almost invariable tendency of the parenchymatous pith of Cordaites to break up on contraction into transverse diaphragms, thus producing what is known as a discoid pith. In the stem shown in fig. 473 the pith is represented by a more or less cylindrical cast characterised by fairly regular transverse ribs and narrow grooves; in the upper part of the fossil the peripheral tissue of the pith is preserved in the form of narrow plates projecting from the inner face of the wood. As Renault pointed out, this type of pith is the expression of certain conditions of growth and is not a satisfactory distinguishing feature of any particular genus or family. The same tendency to form a discoid pith is characteristic of Mesoxylon, and it occurs also in some other Palaeozoic genera. Corda long ago figured a stem attributed by him to Lomatofloyos with a typical discoid pith, and a similar pith is recorded in a stem of Dicranophyllum. Among recent plants Juglans regia affords perhaps the most familiar instance of an identical form of pith: the same type occurs in the white Jasmine, in Ceropegia peltata, and some other flowering plants. An interesting case is that of the tree Groundsel, Senecio praecox D.C., of Mexico: in this plant, which grows in arid districts, the pith serves as a water-store and as the water is drawn off the thick turgescent discs contract and form thin transverse diaphragms separated by wide spaces, as is also the case on drying in some succulent Euphorbia stems. It may be that in Cordaites the medullary region also served as a water-reservoir and the depth of the medullary discs would vary according to the state of their contents.

The earlier writers regarded the pith-casts as stems with scars of amplexicaul leaves: Artis described specimens from the English Coal Measures as Sternbergia, one of which he stated to be 6 ft long; a few years later Sternberg proposed the name Artisia and this has been generally used on the ground that Sternbergia is the name of a recent flowering plant. A specimen of Artisia transversa (Art.) from the Coal Measures of Yorkshire is shown in fig. 466, B, and similar specimens varying considerably in diameter up to about 10 cm. are abundant in European and American Coal Measures. The prominence and depth of the transverse ridges, the presence or absence of anastomoses between adjacent discs are, as Zeiller says, of very doubtful value as specific characters. Dawson in 1846 spoke of Artisia as probably the pith of a tree, a view suggested to him by Mr Dawes. In 1851 Williamson published a description of some specimens in which a pith-cast, Artisia approximata Lind. and Hutt., was enclosed by wood showing very clearly Cordaitean characters. Further demonstration of the true nature of Artisia was supplied by Grand’Eury from St Étienne material. If the generic name Artisia is applied to all pith-casts showing the transverse ridges and grooves like those seen in fig. 466, B, it must be remembered that it is not safe to assume a connexion with Cordaites or Mesoxylon. A Liassic species described by Lignier from France as Artisia alternans is quoted by authors as evidence of the persistence of Cordaites into the Jurassic period; but in view of the fact that the discoid type of pith is not by any means confined to Cordaites or even to the Cordaitales the occurrence of Artisia is in itself of no great botanical significance.

It is also true that a discoid pith is not an invariable attribute of stems closely allied to the genus Cordaites; but if these reservations are made the use of the generic term Artisia serves a useful purpose.

ii. Petrified stems.

=Dadoxylon.= Endlicher.

Palaeobotanical literature contains numerous descriptions of Palaeozoic petrified wood occasionally enclosing an Artisia pith-cast described under such names as Dadoxylon, Cordaioxylon, Araucarioxylon, etc., and regarded as portions of Cordaitean stems. It is, however, certain that much of this material belonged to stems other than those of Cordaites. Recent research has demonstrated the insufficiency of the secondary xylem alone, however well preserved, as a safe guide to generic position: stems identical in the structure of the secondary xylem differ in that of the primary portion of the stele, and it is on the characters of the latter tissues that several genera have recently been founded. Mesoxylon affords a striking example of the importance of the primary xylem as a distinctive feature. As Gothan points out, the species of Calamopitys recently made the type of a new genus Eristophyton would, in the absence of the primary xylem, probably be regarded as Cordaitean. It is important to recognise the limitations imposed by the imperfection of the material; we cannot in most cases determine whether a specimen should be referred to Cordaites or Mesoxylon, and while it may be described as probably Cordaitean in affinity there remains the possibility that some of the Palaeozoic plants with secondary wood like that of Cordaites, if their reproductive organs were known, would not be included in the Cordaitales. Goeppert’s species Araucarites Tchihatcheffi, which Renault quotes as Cordaites, has recently been assigned to a new genus Mesopitys because of certain distinctive features of the primary xylem. Additional examples might be quoted pointing to the tendency of recent and more thorough investigation to establish the fact that the occurrence of Permo-Carboniferous wood of the Araucarian type does not necessarily denote the existence of Cordaites. The question of nomenclature is necessarily raised in this connexion.

In recent years it has been customary to assign Palaeozoic wood with Araucarian pitting to the genus Dadoxylon, while wood of the same general type from more recent strata is by many authors referred to Araucarioxylon. This arbitrary distinction based on a difference in age is open to serious objection. Fossil wood of the Araucarian type is widely scattered in strata ranging from Carboniferous to Jurassic periods; it also occurs in later formations. The fact that on the one hand Araucarian plants, as recognised by cones and foliage-shoots, are especially characteristic of Jurassic floras and occur more rarely in Rhaetic and Triassic floras, and on the other hand that Cordaites and its allies reached their greatest development in Permo-Carboniferous times, renders it probable that in the majority of cases a distinctive name based on geological age would be in accordance with botanical differences. But we have no satisfactory data as to the upper limits of the Cordaiteae or the lower limits of the Araucarineae: in all probability the two families overlapped and co-existed for more than one geological period. It is, moreover, the plants from formations where overlapping occurred that are the most critical from a botanical standpoint. The age-distinction is therefore at best an artificial one and may be seriously misleading. Potonié and Gothan have emphasised the desirability of adopting the name Dadoxylon for all wood of the Araucarian type irrespective of age. If a particular specimen can be correlated definitely with Cordaites or some other genus it should be so designated, but the fragmentary nature of the records usually precludes this simple course. The most logical plan is to use the name Dadoxylon for all woods with Araucarian characters if there is no sufficient reason for employing a less provisional term. If the evidence clearly points to the Araucarineae the generic name Araucarioxylon should be added in parentheses after Dadoxylon, but whether or not this is done, a statement as to the geological age of the fossil will in itself be some assistance in enabling the student to form an opinion on the balance of probability in favour of a Cordaitean or an Araucarian affinity. The course suggested by Gothan, namely to add Cordaites after Dadoxylon if an Artisia pith is present, is rendered inoperative now that we know that a discoid pith occurs in more than one genus. In this chapter we are concerned primarily with Cordaites and with such stems as may fairly be regarded as Cordaitean: examples of fossil wood from later formations are dealt with in another place. A distinction between Araucarioxylon and Cordaioxylon stems has been based by Felix on the nature of the pith-casts; those of the Artisia type he refers to Cordaioxylon, while Palaeozoic stems with Tylodendron pith-casts are assigned to Araucarioxylon. This distinction can, however, only be made in the comparatively few cases in which the pith-cast is preserved. Its validity is, moreover, open to question. A Tylodendron (= Schizodendron) cast shows on its surface the characters of the inner face of the secondary xylem, projecting spindle-shaped areas representing the inner ends of medullary rays and a reticulum of grooves formed by the more resistant and prominent inner edges of the rows of tracheids (fig. 746). A pith-cast of a stem in which the destruction by decay of the medullary parenchyma had not extended to the edge of the xylem-cylinder might show transverse diaphragms. The occurrence of Tylodendron casts means that decay had extended to the surface of the wood. But in view of the occurrence of Tylodendron casts in stems that are not those of Cordaites a short account of the genus is given on another page.

The main features of the stem of Cordaites have already been enumerated. The stele agrees with that of Araucaria and Agathis and especially with Agathis in the double nature of the leaf-trace. Williamson in 1877 described pieces of wood from the English Coal Measures and the Lower Carboniferous of Scotland which he referred to Dadoxylon but without any specific name. These include the Coalbrookdale stem in which he had previously demonstrated the connexion between Artisia and Dadoxylon. The structure of the xylem is like that in D. Brandlingii and the specimens may belong to that species. The most interesting fact recorded by Williamson is the occurrence of double leaf-traces, a feature which led him to suspect a remote generic affinity to Ginkgo. This double trace may be an important diagnostic feature but unfortunately the majority of descriptions of species of Dadoxylon throw no light on the character of the foliar bundles.

Thomson and Allin have recently pointed out that a double leaf-trace occurs in a stem from the Permian of Kansas described by Penhallow as Pityoxylon chasense and referred to that genus because of the supposed occurrence of resin-canals in some of the medullary rays: the canals are apparently leaf-traces traversing broad rays in the secondary wood.

The primary xylem of Cordaites is in direct continuity with the secondary tracheids and does not form mesarch strands as in Mesoxylon. The pith is usually discoid. The pitting on the tracheids is a character of special importance: while it is true to say that as a rule the number of pits on the radial walls of a single tracheid is larger than in the Araucarineae, this is not always the case. In Araucaria there are occasionally as many as five rows of alternate polygonal pits (fig. 691, A) and in some Palaeozoic Dadoxylons there are only one or two rows. The very broad zone of transitional elements at the inner edge of the xylem-cylinder is a characteristic feature shared by the Araucarineae; the spiral protoxylem-tracheids are succeeded by scalariform elements and these, by the gradual anastomosing of the transverse bars, pass into tracheids with multiseriate pitting. In this broad zone we probably have a primitive feature, an epitome in a single stem of the course of development of multiseriate from scalariform pitting. In some Palaeozoic species with wood of the pycnoxylic type and agreeing generally with typical Cordaites the bordered pits are sometimes separate and circular, and opposite pits occasionally replace the usual alternate arrangement. Another feature on which stress has been laid is that in Cordaites the pits occupy the whole breadth of the tracheal wall; but this, though frequently the case, is by no means a constant feature. In Dadoxylon Newberryi the pits tend to form groups, leaving unpitted areas, as in the genus Coenoxylon. In the stem of Dadoxylon materiarum Daws. represented in fig. 475 the pits do not always cover the whole of the tracheid-walls: this stem is also instructive as an example of the different appearance presented by pitted tracheids according to the state of preservation. In some places an oblique pore is well shown while in others only the outer border of the pit is seen. Gothan has described a specimen in which some of the pits are circular and occupy only the central area of the xylem elements: separate circular pits occur also in D. Pedroi Zeill. (fig. 476). Similar departures from the normal are illustrated by recent species of Araucarineae. The absence of a torus is another feature shared by Dadoxylon and true Araucarian wood. Annual rings other than incomplete and spasmodically formed rows of narrower tracheids are not as a rule present, and in this respect also Araucaria affords a close analogy. Thomson has figured a transverse section of a root from English Coal Measures in which rings of growth are well defined; and other instances are recorded. In an Australian species named by Arber D. australe, there are well-marked rings of growth, and this is equally the case in some Indian wood of Permo-Carboniferous age, more nearly allied to Mesoxylon than to Cordaites, and in a Dadoxylon of similar age from South Africa. On the other hand the statement that annual rings occur in Palaeozoic wood is often incorrect, partial rings having been confused with regular concentric cylinders of summer elements. Dawson and Matthew described rings in D. ouangondianum, and Goeppert and Stenzel, who examined the Canadian material, refer to circles like annual rings; but Penhallow states that there is no evidence of true growth-rings.

The medullary rays are uniseriate and consist of thin parenchymatous cells with unpitted walls; they vary considerably in depth, usually comparatively shallow but in some cases 40 or 50 cells deep. In recent Araucarineae the rays are generally shallower. The absence of special receptacles, other than occasional resiniferous tracheids, for products of secretion is a feature common to Dadoxylon and the Araucarineae. The phloem presents no features of special interest, but our knowledge of this tissue is comparatively meagre.

Among other examples of large Dadoxylon stems some of which no doubt bore Cordaitean foliage—though as a rule we have insufficient information as regards anatomical characters to enable a decision to be made between Cordaites and Mesoxylon—reference should be made to the imposing array of silicified trunks in the grounds of the Chemnitz Museum. These were obtained from Lower Permian strata at Hilbersdorf near Chemnitz from beds overlain by porphyry tuff and resting on quartz porphyry, the volcanic material which furnished the siliceous solutions. Several large pieces of wood were found in association with stems of Medullosa and Psaronius, leaves of Cordaites, Artisia pith-casts, and Cardiocarpus seeds with specimens of Walchia, Gomphostrobus and other plants. Sterzel describes a stem 16·5 met. long and 1·5 met. in diameter; on the main trunk the branch-scars are scattered but on some branches there is a tendency to a whorled arrangement. This and many other stems are referred to Araucarioxylon (or Dadoxylon) saxonicum, a species first described by Reichenbach as Megadendron saxonicum. In one specimen Sterzel states that the bordered pits are generally in 1–2 rows, though rarely in 3–4 rows, on the radial walls of the tracheids which they do not completely cover: the medullary rays reach a depth of between 20 and 30 cells. It is noteworthy that the stem 16·5 met. long has a pith-cast of the Tylodendron type.

Dadoxylon (Cordaites) Brandlingii (Lindley and Hutton).

1831. Pinites Brandlingii Lindley and Hutton, Foss. Flor. Vol. +i.+ Pl. +i.+ 1850. Araucarites Brandlingii Goeppert, Foss. Conif. p. 232, Pls. +xxxix.–xli.+ 1890. Cordaioxylon Brandlingii Schenk, in Schimper and Schenk, p. 853, fig. 408.

This species was founded on ‘a fossil giant of the vegetable kingdom’ discovered at Wideopen near Newcastle in Carboniferous strata on the estate of Mr Brandling. The stem, 72 ft long and far from complete, showed an irregular and not a whorled distribution of branch-scars. It is noteworthy that in D. medullaris (Goepp.), a Permian species from Saxony, the branch-scars, while for the most part irregularly scattered, in one case showed an approach to a whorled disposition as in recent Araucarias. Witham gave a fuller account of the structure of the stem than is included in the original description, and the species has been described by many later authors from both Permian and Carboniferous localities. The pith is discoid and the broad transitional region at the inner edge of the wood is a characteristic feature. Thomson points out that there is a tendency to a retention of the scalariform type of pitting in the region of the medullary rays. There are 1–5 rows of pits on the radial walls of the tracheids. The rays may reach a depth of 40 cells; they are usually one cell broad. It has recently been shown that as many as six vascular strands may form one leaf-trace instead of the customary pair, a feature suggesting comparison with Metacordaites Rigolloti Ren. with its five foliar bundles. Other species agree very closely with D. Brandlingii and it is impossible to determine with accuracy the precise specific limits of stems agreeing generally with this type; but for the sake of emphasising the variation in anatomical structure it is worth while to draw attention to a few more or less divergent forms from different geographical areas.

Dadoxylon protopityoides Felix.

An interesting feature in this Westphalian type from Germany is the occurrence of transversely elongated pits on the tracheids associated with those of normal form closely simulating the pits in the xylem elements of Protopitys.

Dadoxylon nummularium White.

In this Brazilian wood from Permo-Carboniferous beds the medullary rays are very numerous, mostly uniseriate and 1–30 cells in depth. The pits on the tracheids are in 1–2 rows and are often contiguous. In another type, D. meridionale, described by the same author, the pits are strictly uniseriate and generally contiguous. As White says, the absence of the pith and cortex and of any evidence as to the structure of the primary xylem renders impossible any definite expression of opinion as to the affinity of these and many other species.

Dadoxylon Nicoli Seward.

Dr Arber in naming this species, from the Newcastle (Permo-Carboniferous) Series of New South Wales, Dadoxylon australe, does not mention Crié’s earlier account of some wood from New Caledonia under the name Araucarioxylon australe. The latter generic name according to the usage adopted in this volume should be superseded by Dadoxylon, and this necessitates a fresh specific name for Arber’s specimens. The name Nicoli is suggested in place of australe, as the sections on which Arber founded his species form part of the Nicol collection in the British Museum.

The xylem shows distinct rings of growth, a feature also seen in Indian stems of approximately the same geological age and recorded by Shirley in wood from Queensland which needs more careful examination. The bordered pits, usually multiseriate and contiguous, are not infrequently in 1–2 rows and separate. The uniseriate medullary rays are very numerous as in White’s Brazilian species D. nummularium, and as a rule 6–12 cells deep. Some well-preserved specimens from Permo-Carboniferous strata in Natal and Zululand have been described by Warren as Dadoxylon australe Arb., showing interesting anatomical features, but the material almost certainly includes more than one specific type and would repay more detailed investigation.

Dadoxylon materiarum Dawson.

This species was described by Dawson from Carboniferous strata in Nova Scotia and afterwards referred by Penhallow to the genus Cordaites. In the transverse section reproduced in fig. 474, A, the tracheid-walls have been reduced in thickness by partial decay, but some of the bordered pits are clearly shown on the radial walls; the pits usually form 2–4 contiguous rows (fig. 475) in some cases with an oblique pore while others are represented either by the outer border of the pit or by the pore only. The narrow medullary rays are as a rule uniseriate and may be 60 cells deep (fig. 474, B). Dawson states that some specimens have large Artisia pith-casts, a fact that formerly would have been regarded as proof of the Cordaites nature of the wood, but in the absence of evidence with regard to the nature of the primary xylem it is impossible to say whether the stem is Cordaites or Mesoxylon.

Dadoxylon sp.

Some wood received from Mr Leslie, collected at Vereeniging, South Africa, in Permo-Carboniferous rocks, shows well-defined rings of growth. The pits form either a single row, a double, alternate and contiguous row, or rarely three series on the tracheid walls. The medullary rays are usually uniseriate and 1–30 cells deep.

Dadoxylon Kayi Arber.

This species is represented by some large trunks, in some cases with a diameter of 40 cm., discovered by Mr Kay in the Coal Measures of Worcestershire. The pith is very small and shows no indication of a discoid structure, but owing to its poor preservation no sections could be obtained of this region. The secondary wood is characterised by the large number of uniseriate medullary rays 1–27 cells in depth; the tracheids have usually two or sometimes three rows of alternate and contiguous bordered pits on the radial walls. Arber regards the absence of a discoid pith as a fatal objection to a reference of the stems to Cordaites and speaks of them as affording further evidence of the occurrence of Coniferae in the higher Coal Measures of the Midlands. It is, however, impossible to determine the position of the species in the absence of any data with regard to the structure of the perimedullary region, and without such information we are hardly justified in regarding Dadoxylon Kayi as a member of the Coniferales.

Dadoxylon Pedroi Zeiller.

This species from Upper Carboniferous or possibly Lower Permian strata in Brazil has a pith 3·8 cm. in diameter composed of parenchyma with scattered secretory sacs and characterised by the occurrence of three equidistant bays projecting into the cylinder of wood (fig. 476, A) which extend through the length of the specimen (6 cm.): these, as Zeiller suggests, may be connected with the departure of leaf-traces or branches. The xylem is entirely composed of centrifugal elements and shows a broad transitional zone (fig. 476, B) including spiral, scalariform, and reticulate tracheids, but the bordered pits are less numerous and less crowded than in many species of Dadoxylon. The rays are 1–2 cells broad and reach a depth of 50 cells. The most striking features are the solid and not discoid pith with its three rounded bays and secretory canals, also the smaller number and frequently circular form of the pits on the tracheids. Zeiller considers that the stem is that of some Cordaitean plant though probably not a true Cordaites. White questions the advisability of adopting the generic name Dadoxylon and suggests the possibility, though without any satisfactory evidence, that it is the stem of a Gangamopteris. Failing further information, there would seem to be no sufficient reason for the institution of a distinctive generic name.

Dadoxylon permiense (Renault).

This Permian species from Autun differs from typical examples of the genus in the differentiation of the pith into a central thin-walled region contracted into transverse diaphragms surrounded by a cylinder of stouter tissue and in the greater breadth of the medullary rays. The tracheids have 3–4 rows of pits of the usual type. Spirally disposed, decurrent, leaf-bases occur on the surface of the stem, and the cortex includes secretory canals and strands of hypodermal stereome. A small number of veins pass up the median part of the lamina which in this respect and in its greater thickness differs from that of Cordaites leaves. Renault speaks of the rays as a cycadean feature, but they are only two cells in width and shorter than in recent Cycads.

Dadoxylon spetsbergense Gothan.

In this species from Spitzbergen, of doubtful age though probably Palaeozoic, there is no xylem-parenchyma and the medullary rays are from 2 to 5 cells deep; the bordered pits occur in 1–2 or rarely 3 rows on the radial walls of the tracheids; they are alternate but not flattened and characterised by their small size (7μ high); they do not cover the whole face of the tracheids. It is pointed out that in many Palaeozoic and Mesozoic Dadoxylons the pits are larger than in recent species (16–17·5μ as compared with 9–12μ) while in D. spetsbergense they are still smaller. The large size of the medullary-ray cells is another noteworthy feature, also the absence of annual rings, a character possibly connected with conditions of growth in northern regions. It is, however, pointed out by Nathorst that the fossil was not found in situ and, as he says, it may have been carried by currents from a more southern locality.

=Metacordaites.= Renault.

Metacordaites Rigolloti Renault.

Renault founded this species and genus on a stem from Autun which, like D. Pedroi, differs in certain respects from stems usually attributed to Cordaites. The pith is solid and contains secretory ducts and cells; the tracheids have often a single row of pits, and multiseriate pitting is much less common than in Dadoxylon. The medullary rays are generally 1–6 cells deep. A striking feature is the occurrence of groups of five vascular bundles penetrating the secondary wood in V-shaped groups, each group being regarded as a multiple leaf-trace, a type recently recognised by Thomson in D. Brandlingii. In one of Renault’s figures a larger scar, presumably a branch-scar, is shown immediately above a group of foliar bundles. The genus Metacordaites is considered by its author to be intermediate between Conifers and the Cordaitales, but nearer to the former. This conclusion is, however, based on insufficient evidence, as nothing is known of the reproductive organs.

+Roots.+

In 1871 Williamson gave an account of a petrified plant from the Lancashire Coal Measures which he named Dictyoxylon radicans, but he afterwards came to the conclusion that the specimens so named were portions of the subterranean axis of some other plant, possibly Asterophyllites, and proposed a new generic term Amyelon. In 1874 he brought forward fresh evidence in support of connecting Amyelon radicans with Asterophyllites or Sphenophyllum, genera which Williamson believed to be very closely related. It has since been recognised that Amyelon is the root of Cordaites or of some closely allied member of the Cordaitales. Our knowledge of Cordaitean roots is based chiefly on the work of Williamson and Renault, and more recently Osborne has added new facts of considerable interest. In the larger roots the primary xylem may be diarch or there may be as many as four or five protoxylem groups (fig. 477). The primary tracheids are spiral or scalariform and the space, s, separating them from the surrounding secondary xylem seen in fig. 477, B, was no doubt originally occupied by conjunctive parenchyma. The secondary wood is composed of tracheids, with contiguous bordered pits identical with those in the xylem of the stem, and narrow medullary rays. The section, 4 mm. in diameter, represented in fig. 477, A, shows a tetrarch primary xylem strand enclosed by secondary wood composed of rather thin-walled elements succeeded by a zone of phloem including some secretory sacs, and beyond this is a cylinder of periderm, p. In a section of a root figured by Renault from Autun the periderm is separated from the stele by a broad band of parenchyma which appears to be cortical, but in the British specimens the deep-seated origin of the periderm is clearly shown: Osborne states that it arises in a layer immediately outside the endodermis. In one of the specimens figured by Williamson the secondary wood shows clearly marked irregular concentric lines simulating rings of growth, but there is no evidence of any regularly recurring variation in the diameter of the xylem-elements. From the descriptions of Williamson and Osborne it is evident that the roots of Cordaites were profusely branched and, as the latter author has shown, the method of branching points to the formation of coralline roots like those of recent Cycads, some Conifers and Dicotyledons. Osborne found that the cortex of small rootlets is composed of two zones, an outer parenchyma without cell-contents and an inner parenchymatous tissue characterised by the occurrence in some of the cells of tangled masses of fungal hyphae almost always unseptate. In some cases the hyphae bear terminal vesicles similar to those observed on fungal hyphae in the cortex of Podocarpus roots. Osborne makes out a good case for regarding the fungus as symbiotically related to the tissues of the lateral roots, a relationship identical with that in many existing trees, particularly Myrica and Alnus. It is suggested that the formation of the coralline root-tubercles is a feature consistent with the view that Cordaites lived in saline marshes, a physiologically dry habitat favourable to the occurrence of mycorhiza.

Reference has already been made to the habit of Cordaitean roots in the general account of the genus (figs. 468, A, 478). The specimen shown in fig. 478 may be a root of Cordaites (Noeggerathiopsis) Hislopi, but nothing is known as to its structure.

+Reproductive Organs.+

=Cordaianthus.= Grand’Eury.

We have as yet no definite knowledge of the nature of the reproductive organs of Mesoxylon and Poroxylon, but having regard to their close resemblance in other respects to Cordaites, particularly in the case of Mesoxylon, the presumption is that some of the seeds and fertile shoots attributed to Cordaites may belong to other members of the Cordaitales. Despite the abundance of Cordaites, or at least of material assigned to that genus, and the comparative frequency of fertile shoots in actual connexion with foliage-shoots, the practical identity of Mesoxylon and Cordaites leaves precludes any confident use of the latter name in a strict sense.

In 1822 Brongniart described a small bud-like fossil of Tertiary age as Antholithes liliacea, and this generic name in the form Antholithus became widely used for fertile shoots or flowers from different geological horizons. As knowledge became more precise other names replaced Antholithus, though Renault retained it for some inflorescences from Commentry which could not definitely be included in Cordaites. Lindley and Hutton employed the genus for a specimen, now recognised as a Cordaitean fertile shoot, from the Coal Measures which they called Antholithus Pitcairniae, the specific name being chosen to indicate a possible affinity to the Bromeliaceous genus Pitcairnia. A few years later Morris described a similar inflorescence as A. anomalus. In 1872 Carruthers substituted Brongniart’s term Cardiocarpon for Antholithus and called A. Pitcairniae Cardiocarpon Lindleyi and Morris’s species C. anomalum. The specific name Lindleyi has been widely adopted, but there would seem to be no adequate reason for disregarding the priority-rule. It is, however, customary to use Grand’Eury’s term Cordaianthus for all Cordaitean inflorescences. Goeppert suggested the name Botryoconus for an inflorescence similar to C. Pitcairniae and for this C. E. Weiss substituted Noeggerathianthus on the ground that he considered Goeppert’s specimen to be the male inflorescence of Noeggerathia. Grand’Eury resuscitated Botryoconus for some spikes from the Gard coalfield connected by him with Dory-Cordaites. The nature of the seeds borne by the inflorescences has largely influenced authors in the choice of a generic name: Carruthers used Cardiocarpon while Zeiller speaks of Samaropsis Pitcairniae. The genus Cardiocarpon was founded by Brongniart for compressed cordiform seeds, but it was not until later that their gymnospermous nature was recognised. Further reference to the nomenclature of seeds of the Cardiocarpon type will be found in Chapter +xxxv.+ The correlation by Grand’Eury and other authors of different species of inflorescences and species of Cordaites is frequently based on association, and in the absence of more satisfactory evidence the safer course is to deal with Cordaitean fertile shoots in a general sense.

(a) Ovulate shoots. These are represented by a considerable number of forms in both European and American localities. In rare cases the compound shoot reaches a length of 30 cm. (fig. 479), but it is usually much shorter; the lateral compact buds may be more or less widely separated: the seeds have long pedicels (fig. 480, A) or appear to be sessile (fig. 480, B) and there may be one or several seeds on a single lateral shoot. The seeds are platyspermic and, as seen in fig. 480, A, in some inflorescences they show very clearly the Samaropsis features. It would, however, be unsafe to assume that all Samaropsis seeds were borne on Cordaitean plants. Among other types of seed referred by authors to Cordaites are Cardiocarpus, Cordaicarpus, Sarcotaxus, Taxospermum, Diplotesta, and Leptocaryon. But in most cases there is no evidence of actual connexion between seeds and vegetative organs, and while it is possible to state with confidence that many of the seeds represented by impressions described as species of Samaropsis and Cordaicarpus are undoubtedly Cordaitean, it is certain that not all seeds referable to these genera were borne by Cordaitalean plants. Cordaitean seeds are characterised by certain morphological features recalling those found in recent Cycads and in the seeds of Ginkgo as illustrated by species of Cardiocarpus and some allied types. As most of the Palaeozoic seeds known in a petrified state cannot be assigned to their parent-plants they are dealt with in a separate chapter.

Cordaianthus Pitcairniae (Lindley and Hutton).

This type of inflorescence is considered by Kidston to belong to the tree which bore leaves known as Cordaites principalis, but if this is the case it is probable that the stem possessed the anatomical characters of Mesoxylon.

The portion of an inflorescence shown in fig. 480, A, from the Middle Coal Measures of Yorkshire, illustrates the occurrence of the bud-like fertile shoots and the stalked Samaropsis seeds. A species described by Renault from Commentry as Cordaianthus acicularis may be identical with the British species.

Cordaianthus Volkmanni Ettingshausen.

The example of this species seen in fig. 480, B, shows the relatively small size of the lateral buds, presumably unexpanded, compared with the large subtending bracts.

=Petrified specimens of Cordaianthus.=

Our knowledge of the structure of Cordaianthus is based on the researches of Renault, supplemented by those of Prof. Bertrand to whose kindness I owe the photographs reproduced in fig. 481. The inflorescences described by Renault are referred by him to different species, but in the following brief account these are treated from a generic standpoint. The tangential section of Cordaianthus Williamsoni Ren. shown in fig. 481, D, was originally figured by Renault and more recently by Bertrand; it shows the spirally disposed leaf-traces in the lower part of a stout axis, and at the sides some vascular bundles are seen passing up into the bracts. A very small proportion of the bracts subtend ovules; two are seen at a and b, and at c is the tangentially cut micropylar canal of a third borne near the apex and covered by the terminal cluster of bracts. The ovule a, separated by a narrow space from its short stalk, consists of a thick single integument—not two as stated by Renault—extended at the apex as a micropylar canal: the apical extension is more completely shown in the tangentially cut ovule b. The central body is much contracted and the two spaces, s, at the base are regarded by Bertrand as cavities in the integument separated from one another by a central strand of conducting tissue which gives off two bundles to the integument, one at each end of the long axis of the seed (fig. 481, A, v). The dark patch, n (fig. D), is the upper and broader end of the shrunken nucellus the apex of which extends upwards as a slender beak, and this originally no doubt fitted into the micropyle. Fig. 481, C, shows a female inflorescence in transverse section; the stele consists of a ring of bundles separated by broad medullary rays and enclosing a comparatively large pith: the leaf-traces are seen in the cortex and one is cut through as it bends out into a bract which is not yet free from the axis. Two ovules, seen in section at a and b, are represented by the bilaterally symmetrical and compressed integument enclosing small pieces of nucellar tissue. Fig. 481, E, is a transverse section of an inflorescence at a higher level and above the apex of the axis: there are four large ovules and one aborted ovule, a. Bertrand describes two vascular bundles in the integument of the ovule a, one at each end of the long axis.

Fig. 481, A, B, Cordaianthus Grand’Euryi Ren., shows a longitudinal section of the nucellus, 1·5 × ·7 mm., and part of the integument of an ovule at the time of pollination, which was probably aided by the secretion of a drop of mucilage as in the ovules of recent Conifers. The integument, separated by a broad space from the nucellus, is cut in the plane of the two vascular strands, v. From the centre of its broad upper surface the nucellus projects upwards as a beak, b, and this originally engaged with the micropylar canal formed by the integument: the lighter patch below the beak is the pollen-chamber (fig. 481, B, pc) containing two microspores, and two more, p, are seen above the nucellar beak. In another species described by Renault, C. Lacattii, the nucellus fills the space bounded by the integument.

* * * * *

(b) Staminate inflorescences. The male inflorescence, though smaller, is similar in habit to the ovulate shoot: the secondary branch consists of a short axis bearing crowded, spirally disposed, bracts, and the actual flowers are represented by single stamens or groups of 2–3 highly specialised microsporophylls. Each microsporophyll consists of a long filament with a central vascular strand bearing at its apex 3–4 long microsporangia (fig. 481, F, m) which open longitudinally as seen in fig. 482, A. The microsporangia are 2·5–3 mm. long covered by dark palisade cells and thin-walled parenchyma, shown as indistinct patches in the photograph. Some of the elliptical and comparatively large microspores are seen in fig. 482, B; the exine is finely punctate and inside are the remains of a few thin cells in which presumably spermatozoids were developed. The microspores shown in fig. 482, B, have a maximum length of 0·1 mm.: Renault describes some as 0·9 mm. long while others are much smaller. Fig. 481, F, is a transverse section of a staminate inflorescence showing near the centre five groups of microsporangia, each sporangium having the form of a curved incomplete dark band indicating that dehiscence has occurred.

=Mesoxylon.= Scott and Maslen.

This generic name was instituted for stems obtained by Messrs Lomax from the Lower Coal Measures of Lancashire previously referred by Scott to Cordaites and Poroxylon. Further investigation showed that while agreeing closely with those genera they possessed certain distinctive features demanding recognition. The name chosen suggests the intermediate nature of the stems. The more striking features may be summarised as follows: In the largest specimens so far described the stem, including leaf-bases, reaches a diameter of 5 or 6 cm.; the large pith consists in the central region of diaphragms of parenchyma separated by horizontal spaces produced by splitting and shrinkage consequent on the failure of the tissue to keep pace with the general growth of the stem. The secondary xylem is of the Araucarian type and has narrow medullary rays varying in depth from 1 to 25 cells. The leaf-traces are represented by twin-bundles which fuse in the downward direction, the level at which fusion occurs being regarded as a specific character. The presence of centripetal xylem is an essential feature of the traces: the occurrence of single or double traces consisting of centripetal elements and, externally, a large amount of centrifugal xylem is an important feature in which Mesoxylon differs from Cordaites. The double leaf-traces divide after emerging from the secondary wood and each strap-like leaf receives several collateral bundles (fig. 483, C). An axillary bud may occur at the base of each leaf (fig. 483, A, b). The phloem, including sieve-tubes and secretory sacs, is succeeded by a broad pericycle, and the comparatively narrow cortex is traversed by successive bands of periderm. In the outer region of the cortex the presence of radial bands of fibres is a characteristic feature. The reproductive organs are unknown. The anatomical features are well illustrated by M. Sutcliffii first described by Scott, who provisionally placed it in Poroxylon, and afterwards more fully investigated by Maslen.

Mesoxylon Sutcliffii Scott.

The average diameter of the stem is 3 cm.: the section reproduced in fig. 483, A, has a maximum breadth of 3·5 cm.; the leaf-bases cut at different levels give an irregular contour to the surface like that of a Lepidodendron. An axillary bud, either reproductive or vegetative, is seen at b consisting of a short axis bearing crowded bud-scales. The leaves are crowded and according to Maslen have a phyllotaxis of ⁸⁄₂₁: the lamina is linear like that of Cordaites with 16 collateral bundles in the petiole. The presence of a meristematic band at the base of the lamina affords evidence of a deciduous habit. The large size of the pith is a striking feature with its central tissues in the form of transverse diaphragms and a narrower peripheral zone of solid parenchyma (fig. 483, A, a). The secondary wood of the stele is composed of tracheids with 2–3 contiguous alternate rows of bordered pits on the radial walls, but none on the tangential walls. In the stem shown in fig. 483, A, the secondary wood is preserved only in patches. Numerous blunt teeth varying in prominence project into the pith; these consist chiefly of serially disposed centrifugal tracheids distinguished by their spiral and scalariform structure and by the medullary rays which are broader than those in the more external xylem. Further reference is made to these perimedullary strands in the description of the leaf-traces. The medullary rays are uniseriate and usually 1–6 cells in depth: beyond the secondary wood is a cambium and a cylinder of secondary phloem (fig. 483, D, ph²) consisting of tubular elements, presumably sieve-tubes, and elongated secretory sacs. The pericycle is composed of several rows of rather large and short cells and has an ill-defined outer boundary. A succession of arcs of periderm-like tissue and phellogen, which may invade the pericycle and phloem, forms a prominent feature in the cortex; radially placed bands of fibres similar to those in Lyginopteris and other genera occur in the outer cortex. At the edge of the pith the more prominent projections of xylem are arranged in pairs (fig. 483, B) and as each pair travels downwards the component strands gradually fuse. Each bundle of a double trace consists internally of an arc of centripetal xylem, the elements of which are arranged in rows (fig. 483, B, cp), with a single protoxylem group in the middle of the inner face, px. It is not clear whether any primary centrifugal tracheids are present, but there are indications that such are occasionally represented. In most cases the primary xylem of the leaf-traces is exarch, but the existence of mesarch bundles is not improbable. The bulk of each foliar bundle is formed of a fan-shaped mass of secondary centrifugal xylem (fig. 483, B, cf) and an island of parenchyma occurs next the protoxylem. There is no clearly defined boundary between the outer or centrifugal xylem of the leaf-traces and the tracheids of the stem-wood; the latter may consist exclusively of tracheids with bordered pits or the inner rows of the xylem-cylinder may be of the scalariform or spiral type. Differences shown in transverse sections of the inner portion of the xylem are due to the circumstance that in certain parts of the inner face of the secondary wood leaf-traces are unrepresented, while in other places the dwindled remains of the outer, centrifugal, portions of a trace are still recognisable. As each double leaf-trace passes down the pith the bundles fuse and the single strand retains for a time some centripetal xylem; this gradually disappears and at a lower level the centripetal xylem also dies out. The space enclosing the obtuse apices of the bundles shown in fig. 483, B, was originally occupied by thin-walled tissue which accompanied the trace in its outward course. In Mesoxylon Sutcliffii the leaf-strands pass almost horizontally through the secondary wood, bend outwards in the phloem and follow a steeply ascending course to the leaves. In fig. 483, B, a double leaf-trace is seen at the inner edge of the secondary wood with the centrifugal xylem, cf, continuous with that of the stele: fig. 483, D, lt, shows a leaf-trace in the pericycle where one of the bundles has divided and the other is tangentially extended and partially divided. The branching is carried further in the cortex, as seen in fig. 483, C, where the trace is represented by a curved row of six bundles, lt, and at a higher level further subdivision may occur. The leaf-bundles are collateral and in the leaf retain both centripetal and centrifugal tracheids. In the section shown in fig. 483, C, the oval stele of an axillary shoot is seen at s subtended by the row of collateral bundles: the stele has a fairly large pith surrounded by a zone of secondary xylem with broad medullary rays.

* * * * *

Among other species of Mesoxylon mention may be made of M. Lomaxi and M. poroxyloides. M. Lomaxi Scott and Maslen generally resembles M. Sutcliffii but shows the following distinctive features: the leaves are more scattered and less crowded; the twin-bundles of the leaf-traces fuse immediately on entering the pith, thus appearing for the most part as single and not double strands in the perimedullary zone; the centripetal xylem is well developed, the medullary rays are deeper and the outer cortex has shorter bands of mechanical tissue.

In Mesoxylon poroxyloides Scott and Maslen, the twin-bundles of the traces unite soon after reaching the pith as in M. Lomaxi, the secondary tracheids have only two rows, or sometimes a single row, of bordered pits and the tracheids are rather smaller than in M. Lomaxi (20–40μ as compared with 30–60μ) and the medullary rays are shallower. There is a particularly broad zone of spiral and reticulate transitional tracheids at the inner edge of the wood as in Cordaites and in Dadoxylon Pedroi (fig. 476). The leaves of this species are believed to be represented by the type described by Dr Benson as Cordaites Felicis (fig. 465), but, as already suggested, it is very probable that many or possibly nearly all the leaves from British Coal Measures described as Cordaites may belong to Mesoxylon.

The chief interest of the genus Mesoxylon is its close resemblance in certain characters to Cordaites and Poroxylon: the presence of strands of centripetal xylem in the perimedullary region is an important feature in which Mesoxylon differs from stems assigned (under the generic name Dadoxylon) to Cordaites. Mesoxylon differs from Poroxylon in having a discoid pith like that of Cordaites, but a more important difference is the absence in the leaf-trace xylem of Mesoxylon of bordered pits of the Araucarian type, whereas in Poroxylon Araucarian pits occur in both the centripetal and centrifugal tracheids. In Poroxylon the secondary xylem is manoxylic; in Mesoxylon, as in Cordaites, it is pycnoxylic.

=The range of Cordaites and a consideration of other imperfectly known genera.=

An increased precision in knowledge derived from anatomical investigation often tends to demonstrate the untrustworthiness of criteria based on external features previously employed with confidence. This inevitable though, from the point of view of the systematist, inconvenient result of intensive study is well illustrated by the recent discovery of the stems named by Scott and Maslen Mesoxylon. A separation of Cordaites from Mesoxylon, which no doubt extended far beyond the British area, is possible only if well-preserved petrified material is available. The leaves of Mesoxylon, so far as our imperfect knowledge of them enables us to express an opinion, are constructed on a plan almost identical with those of Cordaites and, as already stated, it is almost certain that many of the impressions referred to Cordaites were borne on Mesoxylon stems. An additional source of confusion is supplied by the Cordaites-like leaves of Poroxylon. It is evident, therefore, that even within the limits of the Carboniferous and Permian formations the recognition of true Cordaites leaves must often be attended with considerable risk of error. Apart from the possible confusion between the foliage of Cordaites and Mesoxylon there are other difficulties as regards detached leaves which depart more or less widely from the typical Cordaitean form. Leaves such as C. circularis (fig. 468, B) and C. grandifolius emphasise the lack of any thoroughly satisfactory dividing line separating single pinnules of Cardiopteris or Cyclopteris on the one hand and leaves of Psygmophyllum on the other from Cordaites. The petrified buds described as Dolerophyllum have been quoted by several authors as examples of unexpanded shoots of Cordaites though anatomical evidence warrants a generic separation. In the case of species founded on leaves described in this chapter as Cordaites it should be remembered that further research may necessitate an alteration in nomenclature.

Among the species included in Cordaites is Noeggerathiopsis Hislopi (figs. 470–472), a type widely spread in India and in other parts of Gondwana-Land: if the change of generic name is accepted it involves the extension of the geographical range of Cordaites from Northern Europe and North America to the southern botanical province. We have as yet no proof of the existence of Cordaites in the Arctic regions. The range in time of Cordaites or of the Cordaitales has generally been stated to be from the Upper Devonian to the Permian. It is, however, by no means certain that the genus flourished before the Carboniferous period, though it is clear that closely allied types must have lived in pre-Carboniferous floras. The strata in New Brunswick from which Dawson recorded his supposed Devonian Cordaites have been shown to be Upper Carboniferous in age. As regards the length of time during which the Cordaitales existed we have no decisive evidence. In recent years the tendency has been to extend their range into the Mesozoic era, and there are several pieces of evidence in favour of this. There is no doubt that considerations of age based on the arbitrary divisions of the geological scale sometimes insinuate themselves too thoroughly into questions connected with the duration of plant-types whether represented by families or genera. We have been accustomed to regard Cordaites as a genus confined to the Palaeozoic period, a type which with many others carried on the tradition of Upper Carboniferous forests to the Permian floras and then made way for the precursors of Mesozoic types. There is, however, no valid reason for supposing that Cordaites and other Palaeozoic genera did not survive as less prominent members in succeeding floras. It must be admitted that evidence in support of Mesozoic Cordaitales is not above suspicion, though the probability is that Cordaites or some allied genera still flourished in the earlier stages of the Mesozoic era. The data on which this opinion is based cannot be fully discussed in a general treatise, but a few of the facts may be briefly considered. Zeiller and other authors have expressed the view that the Cordaitales were not exclusively Palaeozoic. In addition to Cordaites (Noeggerathiopsis) Hislopi recorded from Rhaetic floras, other possible representatives of the group are illustrated by specimens included in such genera as Yuccites, Bambusium and Krammera.

=Pelourdea= gen. nov.

The name Yuccites was given to some detached, broad, linear leaves from the Bunter sandstone of the Vosges which were compared with the foliage of Yucca and classed among Monocotyledons. The authors of the genus also described a cylindrical cast as a Yuccites stem, including both stem and leaves in Yuccites vogesiacus. The supposed stem, as Fliche has shown, is a pith-cast and is appropriately named by him Endolepis vogesiacus. The Vosges leaves are assigned by this author to the genus Cordaites, a change of name which may eventually be justified though as yet based on insufficient evidence. There are objections to the institution of a new name in place of Yuccites, but it is undesirable to retain a designation suggesting false ideas with regard to affinity. A new name Pelourdea (after M. Pelourde of Paris, whose recent death deprives Palaeobotany of an able and promising investigator) is therefore proposed for leaves of the Yuccites type which in form, venation, and spiral phyllotaxis agree with those of Cordaites but cannot confidently be assigned to that genus or even to the Cordaitales. For linear leaves, especially from Jurassic strata, resembling those of Phoenicopsis the name Desmiophyllum is employed: these are very similar to those of Pelourdea; they are characterised by their fairly uniform breadth and afford no indication of their arrangement on the supporting axis.

Pelourdea vogesiaca (Schimper and Mougeot).

The linear-lanceolate leaves described by Schimper and Mougeot as Yuccites vogesiacus and transferred by Fliche to Cordaites are probably specifically identical with specimens described by Mr Wills from Lower Keuper rocks in Worcestershire. The English leaves were described by Arber as Zamites grandis,—the name Zamites vogesiacus having been previously used by Schimper and Mougeot,—on the ground that the supposed leaves were probably pinnae of a cycadean frond, a view in agreement with an opinion previously expressed with regard to similar leaves from Stonesfield. A later discovery by Wills of specimens, on which the drawing reproduced in fig. 484 is based, of the same type of leaf showing the foliar nature of the fossils necessitated the abandonment of the pinna-hypothesis, and the original name Yuccites vogesiacus was resuscitated. The leaves reach a length of 50 cm. and a maximum breadth of 6·5 cm.; the lamina is entire, lanceolate or linear-lanceolate, the apex acuminate, and the lower part rather abruptly contracted and attached by a broad crescentic base; veins numerous, parallel, and occasionally forked. Fliche records the occurrence of a small Artisia-like pith-cast and pieces of stem with leaf-scars (4 × 3 mm.) in association with leaves of Pelourdea vogesiaca in Triassic strata in Lorraine. An imperfectly preserved specimen described by Fliche as Cordaianthus Minieri resembles an inflorescence of Cordaites. It consists of an axis 15 cm. long, the lower part forming a peduncle, and on the upper portion are linear bracts subtending oblong bodies which may be lateral fertile shoots.

Pelourdea hadroclada (Halle).

Dr Halle recently published an account of some imperfect leaves and stem-fragments from the Rhaetic of Scania which he named Phyllotenia (?) hadroclada, the generic name provisionally adopted having been proposed by Salfeld for some rather obscure remains from the Corallian of Germany. It appears to have escaped the notice of both authors that Saporta in 1894 had adopted the designation Phyllotenia for some examples of broad parallel-veined leaves from Lower Cretaceous rocks in Portugal very similar to Velenovský’s Krammera mirabilis. Some other generic name must therefore be used. For the Rhaetic species the name Pelourdea would seem appropriate. The type-specimen consists of an axis 10–12 mm. in diameter with spirally disposed transversely elongated leaf-scars bearing sessile linear leaves similar to Poa-Cordaites; none of them are complete, the largest is 6 cm. long and 5–7 mm. broad with 8–12 parallel veins. An examination of the original specimens in the Stockholm Museum satisfied me that Dr Halle is justified in the opinion that they may be fragments of some Cordaitalean plant and that he was well advised to avoid the use of the name Cordaites. Salfeld’s species, Phyllotenia longifolia, may be an imperfectly preserved example of Phoenicopsis, but the material is too incomplete to be identified with any degree of confidence.

Pelourdea Imhofi (Heer).

The Triassic leaves from Switzerland described by Heer, and more recently by Leuthardt, as Bambusium Imhofi, were referred by Fliche to the genus Cordaites. The lamina is ensiform, 25 cm. long with a maximum breadth of 2·4 cm. Leuthardt’s photograph of aerial stems and rhizomes of this supposed Monocotyledon are far from convincing.

Pelourdea keuperiana (Compter).

The leaves from the Lower Keuper of Thuringia assigned by Compter to Cordaites without adequate evidence resemble those of P. vogesiaca, but there is no evidence as to their manner of attachment; they are 30–40 cm. long and from 1·5 to 2 cm. broad.

Pelourdea megaphylla (Phillips).

This species was first described by Phillips from the Middle Jurassic Stonesfield Slate and afterwards referred to Zamites: the leaves bear a striking resemblance to foliage of the type Cordaites borassifolius; the lamina is 30 cm. long and attains a breadth of 3 cm., the apex is acuminate and slightly contracted towards the broad concave base. My former comparison of these Stonesfield leaves with the long pinnae of Ceratozamia mexicana seemed to be supported by Phillips’s type-specimen of Palaeozamia longifolia. It may be that the supposed pinnae in Phillips’s type are spirally disposed leaves: if this is the case the specimen may be a fragment of a Podozamites; its specific identity with the larger detached specimens, though probable, cannot be demonstrated. Some leaves figured by Zigno from Jurassic rocks of Italy as Yuccites Schimperianus may be identical with P. megaphylla.

Pelourdea mirabilis (Velenovský ex Corda +MS.+).

The generic name Krammera, suggested by Corda, was employed by Velenovský for large Cordaites-like leaves from the Lower Cretaceous of Bohemia, for casts of cones regarded by him as stems, and for fruit-like bodies. The leaves, previously described as Flabellaria chamaeropifolia Goepp., Dammara albens Presl, etc., bear a close resemblance to the large broadly linear leaves of Cordaites; the lamina reaches a length of 40 cm. and between the veins occur 1–4 finer striations. The fossils identified by Velenovský as stems bearing crowded imbricate scales, which he regarded as the persistent bases of Krammera leaves, are probably cones; they agree very closely in size and shape, also in the form of the scales, with cones of Agathis and some other recent Conifers. As the designation Krammera was instituted primarily for cones and not leaves the name Pelourdea is substituted for it.

=Niponophyllum.= Stopes and Fujii.

Niponophyllum cordaitiforme Stopes and Fujii.

The generic name Niponophyllum was proposed for some petrified specimens of leaves or possibly leaflets from Upper Cretaceous beds in Japan which, though not definitely assigned to a group or family, are considered by the authors of the genus ‘to lie [anatomically] somewhere between Cordaites and Cycadeoidea’ ‘with a closer similarity to the former than to the latter if we compare the whole Cordaites leaf with our blade.’ The data on which this conclusion is based are, however, insufficient to justify a reference of Niponophyllum to the Cordaitales or indeed to lend any substantial support to the opinion that the Japanese specimens are anatomically more akin to Cordaites than to other plants. The type-species is represented by two specimens of leaf-fragments about 0·4 mm. thick and from 6 to 9 mm. broad containing from 21 to 33 vascular bundles; the upper part of the mesophyll is composed of palisade tissue and the stomata appear to be confined to the lower epidermis. Each bundle is accompanied by an =I=-shaped girder, and small patches of sclerenchyma occur next the upper epidermis between the girders; there are no resin-canals: the vascular bundles are collateral, the xylem is said to be almost entirely centripetal and exarch, but in the absence of evidence afforded by longitudinal sections the details of structure cannot be definitely determined. A comparison is made with Cycadean leaves and with leaves of Araucarineae and Podocarpeae, also with Cordaites; another type with which Niponophyllum may be compared is Desmiophyllum Solmsi.

The genus is interesting as an example of a petrified gymnospermous type of leaf characterised by the absence of resin-ducts and transfusion-tracheids, the possession of collateral, apparently exarch, bundles enclosed in a double sheath; but the data supplied are insufficient to enable us to allocate the specimens to a position within the class.

A specimen described by Schenk as Eolirion primigenium from Lower Cretaceous beds in the Carpathian mountains closely resembles in habit a foliage-shoot of Poa-Cordaites; the leaves are narrow and linear with obtuse apices and attached, apparently, in a close spiral. Schenk assigns the plant to the Monocotyledons, but its systematic position must be left unsettled.

* * * * *

The list of Mesozoic specimens resembling Cordaites leaves might be extended. Apart from some Triassic and Rhaetic examples which may well be Cordaitalean, there are many others which, though similar in form and venation to Cordaites, are in all probability more closely related to Agathis and other genera; the species Dammarites Bayeri recently described by Zeiller from the Upper Cretaceous of Bulgaria is a case in point. The Araucarian character of the wood of Cordaites precludes any satisfactory discrimination between Mesozoic Araucarian stems and those of Cordaitalean species, at least in the case of such material as is usually available.

=Titanophyllum.= Renault.

Titanophyllum Grand’Euryi Renault. The remarkable leaves on which this genus and species are founded were discovered in the Commentry coalfield; they occur as detached specimens and cannot be correlated with any known stem. Renault suggests that the Autun stems referred to Colpoxylon may have borne the Titanophyllum leaves, but this correlation rests only on the dimensions of the stems and the occurrence of transversely elongated scars on the surface. The lamina is thick and coriaceous, 70–75 cm. long and 20–25 cm. in breadth; the veins are parallel but not branched; numerous longitudinal striations on the upper surface indicate the presence of hypodermal stereome-strands; stomata are abundant on the lower surface and the more or less rectangular cells in the neighbourhood of the stomata appear to be papillose (fig. 485, A, B). The distal region of the lamina is often torn into strips (fig. 485, A); the approximately rectangular leaf has a broad elliptical base 9–10 × 3–4 cm.

Dr White describes a specimen from the Lower Coal Measures of Missouri as ? Titanophyllum Brittsii which he speaks of as the thick base of a leaf similar to that described by Renault but, as White says, no formal diagnosis is possible without more satisfactory material. Such evidence as is available suggests that Titanophyllum is a type of Cordaitalean leaf probably closely allied to Cordaites.