CHAPTER XXXV.

PALAEOZOIC GYMNOSPERMOUS SEEDS.

Seeds are abundantly represented as fossils from Carboniferous to Post-Tertiary deposits. The importance of fossil and sub-fossil species in the later geological series has been demonstrated by the investigations of Mr and Mrs Clement Reid and a few other workers in this neglected field. In cases where it is possible to assign seeds to their parent-plants the descriptions of casts, impressions, or petrifactions are added to the account of vegetative organs; but it frequently happens that seeds are preserved only as detached specimens many of which have little or no value as botanical records, while others that afford striking examples of the possibilities of petrifaction as a means of preserving the most delicate structures, are of great importance. In Volume +ii.+ an account was given of such Palaeozoic seed-bearing organs as _Lepidocarpon_ and _Miadesmia_, and the genera _Lagenostoma_, _Sphaerostoma_, and _Trigonocarpus_ are dealt with in this volume under _Lyginopteris_, _Heterangium_, and _Medullosa_. Certain seeds afford some evidence as to the systematic position of the parent-plants though insufficient to warrant more than a surmise as to the nature of the vegetative organs: in several cases it is only from the resemblance of detached seeds to types that on the strength of more or less convincing evidence are referred to definite parent-plants that any conclusions can be drawn with regard to precise systematic position. In view of the occurrence of several different types of seeds that retain their morphological features, but cannot be assigned with any degree of certainty to genera founded on vegetative organs, a special chapter is devoted to a comparative study of selected examples with the object of directing attention to data bearing on evolutionary problems. The chief interest of Palaeozoic seeds to the botanist lies in the facts they contribute towards the elucidation of questions connected with the promotion of the megasporangium and megaspore of the Pteridophyta to the higher stage represented by the integumented megasporangium (nucellus) and single megaspore that in the main fulfil the definition of a seed[790]. ‘With the evolution of the seed,’ as Oliver says, ‘the plant rose at a bound to a higher plane, and this structure in its perfected form has become the very centre of the plant’s existence[791].’ We can as yet form a very partial conception of the successive stages in the adoption of the seed-habit, but since 1855, the year in which Hooker and Binney[792] published their paper on the structure of _Trigonocarpus_, ample proof has been furnished of the importance of Palaeozoic seeds from the standpoint of affinity between recent Gymnosperms and extinct seed-bearing plants, and of the _modus operandi_ of evolutionary tendencies. A cursory examination of Palaeozoic seeds suffices to demonstrate their resemblance to those of recent Cycads and the seed of _Ginkgo biloba_; but while it is clearly with these Gymnosperms that the majority of the seeds described in the following pages are most closely allied, the extinct types possess many distinguishing features that throw light on some at least of the factors concerned in the production of the modern type. In many of the Palaeozoic seeds the nucellus stands free within the integument, to which it is attached only in the chalazal region, in contrast to the lateral union between integument and nucellus in the ovules of recent Cycads. It has been suggested by Oliver[793] that the seed of the Conifer _Torreya_ affords a clue to the interpretation of this difference and that the lower part of the seed in Cycads and _Torreya_ represents a later intercalation in the basal portion of the ovule, the ancient type having a free nucellus in contrast to the nucellus of modern seeds which is free only at the apex. It has been objected[794] that there are no adequate grounds for assuming the addition of an intercalated zone or of the elongation of the ovule that this implies, the more probable view being that the lateral union of nucellus and integument represents congenital fusion in the ancestral type, a fusion comparable with that of the coherent petals of a gamopetalous corolla. In the presence of a pollen-chamber most of the Palaeozoic seeds agree with those of recent Cycads, but in the extinct forms it is usually a more highly developed structure. The name pollen-chamber was given by Brongniart[795] to the pollen-containing cavity in the free region of the nucellus in the petrified seeds from St Étienne in ignorance of the use of the same term by Griffith[796] in a posthumous work published in 1852 for the nucellar chamber in _Cycas_[797]. The genus _Stephanospermum_ (fig. 494, A) illustrates the means by which the pollen-chamber was liberally supplied with water and thus adapted to the requirements of fertilisation by motile gametes. The pollen-chamber and its vascular supply paved the way for siphonogamy, that is the development of a pollen-tube for the more direct transmission of the male sperms. The highly developed mantle of tracheal tissue at the periphery of the nucellus in _Stephanospermum_, represented on a reduced scale by the separate vascular strands of other seeds, may be compared with the tracheal investment to the nucellus in the recent Dicotyledonous genus _Cassytha_[798]. The presence of a nucellar vascular system in several Palaeozoic seeds is a feature in which they differ from those of recent Cycads with the exception of _Bowenia_. The retreat of the vascular supply from the immediate neighbourhood of the pollen-chamber in recent Cycads may, as Oliver points out, be correlated with the evolution of the pollen-tube—the substitution of siphonogamy for zoidiogamy. The diagram reproduced in fig. 492 represents a synthetic type based on such seeds as _Stephanospermum_ and _Cardiocarpus_ which illustrate an arrangement of conducting tissue frequently found in Palaeozoic seeds: the main strand gives off a pair of bundles in the sarcotesta in the principal plane, as in _Cardiocarpus_[799]; from the tracheal mass in the chalazal region numerous bundles pass up the nucellus as far as the floor of the pollen-chamber. The nature of the vascular supply in this generalised type and in individual genera should be compared with that in the seeds of recent Cycads[800] described in Chapter +xxviii.+

[Illustration: Fig. 492. ‘A conjectural synthetic type of seed embodying the characters of such a seed as _Stephanospermum_ with those of a _Cardiocarpus’_; _n_, nucellus; _pc_, pollen-chamber. (After Oliver.)]

Recent research has revealed the not unexpected fact that in such Upper Carboniferous petrified seeds as have been investigated—a small proportion of the large number produced in the Palaeozoic forests—there was a remarkable range in the mechanism connected with pollination and the maturation of the microspores. A comparison of the genera _Physostoma_, _Lagenostoma_, _Conostoma_, _Trigonocarpus_, _Stephanospermum_ and others reveals the occurrence of very different though not unrelated structural features especially in the apical region of the seed. These seeds, including _Physostoma_, probably the most archaic type, represent a stage in evolution already far removed from the starting-point: the diversity of plan recalls the variety in the form of the chloroplasts in the Green Algae, and in both cases we are in touch with an experimental phase representing a tentative advance towards greater efficiency.

In its differentiation into an outer fleshy region, the sarcotesta, a stony layer, the sclerotesta, and in many cases an inner flesh, the Palaeozoic seeds resemble recent Cycads: in both extinct and modern seeds the balance of evidence would seem to be in favour of attributing a single rather than a double origin to the integument.

Among the numerous types of Palaeozoic seeds are several which invite comparison with the fruits or carpels, apart from the seeds, of Angiosperms. Impressions of _Samaropsis_ seeds (figs. 502, B–K; 503; 504) bear a close resemblance to the laterally expanded fruits of the common Crucifer _Thlaspi arvense_; the ribbed testa of _Hexagonocarpus_ (fig. 506, H) and other genera recalls the fruit-wall of _Alstroemeria_; the recently described Lower Carboniferous seed _Thysanotesta sagittula_ Nath. (fig. 506, F) simulates a carpel of _Erodium_. These and similar instances of a close parallelism in external features between organs that are not homologous, though in themselves of no morphological significance, are at least interesting as illustrating the plasticity displayed by reproductive structures, which in the Palaeozoic period marked a morphological achievement comparable in its importance with the still greater achievement represented by the highly specialised fruits of the modern Flowering plants. The range in form and surface-features of Angiospermous fruits was foreshadowed by Palaeozoic seeds. Structural types and in some cases, superadded to these, features which may reasonably be supposed to have facilitated dispersal had been acquired by the seeds of Palaeozoic plants in forms that in a much later period were adopted by fruits even to a greater degree than by seeds. Characters useful in seed-dispersal, that are now shared by fruits and seeds, are illustrated by the fleshy and possibly edible seeds of extinct Gymnosperms, the plumes and hairy beak of _Gnetopsis_ (fig. 494, E) and _Thysanotesta_ (fig. 506, F) suggestive of feathery stigmas and other appendages. The lacunar sarcotesta of _Aetheotesta_, the thick endotesta of _Pachytesta_ (fig. 497), and the air-chamber of _Codonospermum_ (fig. 498), are strictly comparable with aids to buoyancy in fruits of existing Flowering plants. The mucilage-hairs and superficial cells in _Physostoma_ (fig. 494, I) and _Stephanospermum_ may be compared with the thick mucilaginous investment of the megaspores of recent water-ferns and with similar tissues of some Angiospermous seeds.

The bionomics of Palaeozoic plants is a subject worthy of more serious attention than it has so far received. The search for morphological characters that may have facilitated the wanderings of widely distributed genera and species and a closer investigation of physiological-anatomical problems presented by the vegetative organs of petrified plants would not only extend our knowledge of the morphology of ancient types but would stimulate comparative study and, incidentally, relieve the dullness of pure description. It may be argued that we should first establish a more solid foundation by further observations on living plants; but even at the risk of allowing speculation too free a hand the attempt is worth making, and it may be urged that, as in phylogenetic enquiries so in other branches of botany, facts obtained from plants of other ages may serve to supply deficiencies in knowledge based only on existing forms. One of the difficulties inseparable from the study of fossil plants, namely the identification of impressions and casts with specimens exhibiting anatomical characters, is particularly well illustrated by seeds. The description of a genus based on mere external form may sometimes be extended without great risk of error to include species founded on anatomical characters, but on the other hand, there are many instances in which—despite a general resemblance in form and size between petrifactions and impressions—lack of evidence of generic identity requires the employment of distinctive names. The determination of impressions is, as Lesquereux recognised, ‘subject to a great deal of uncertainty,’ and many of the genera founded on external features are purely artificial and include species that have no essential features in common. Moreover in the case of petrified specimens the apparent absence of an external fleshy layer is often due to destruction before preservation: as Solms-Laubach[801] points out, it is obviously impossible to be certain as to the number of integumental layers in seeds that are not well preserved in all their parts. Goeppert founded a new genus, _Acanthocarpus_, on a Permian seed described as _A. xanthioides_[802], because of the occurrence of spinous processes attached to an obcordate kernel: these apparent spines are in all probability the remains of a very imperfectly preserved sarcotesta. The preservation of the central portion of a seed, that is the seed-cavity with the enclosing shell, conveniently called the nucule, has often led to an unnecessary multiplication of generic terms. Other examples of confusion resulting from different states of preservation are quoted in the accounts of some of the selected types.

Williamson in 1877 pointed out that we learn from the large number of different kinds of Palaeozoic seeds that ‘there were in the Carboniferous forests many gymnospermous stems clothed with foliage of which we have not yet discovered any traces, probably because these Gymnosperms did not flourish upon the low swampy grounds which were the homes of the great mass of the coal-producing plants[803].’ Prof. Zeiller[804] has also drawn attention to the numerical excess of seeds over vegetative organs. This discrepancy has to a large extent been explained by the discovery that many of the supposed Ferns were seed-bearing plants, and a further explanation is suggested by the superiority of seeds over stems and leaves in their adaptation to dispersal by water.

In 1874 Brongniart[805] described several petrified seeds from material discovered by Grand’Eury in the St Étienne coal-field, and seven years later his descriptions were republished[806], with the addition of several beautifully executed drawings, as a posthumous volume edited by his distinguished pupil Renault. Williamson’s researches supplied much additional information, and in recent years the more detailed study of French and English seeds by Bertrand and particularly by Oliver and his pupils has further emphasised the interest and importance of this field of work. Brongniart proposed a two-fold classification of French seeds: (i) bilaterally symmetrical seeds, more or less flattened in section, which he believed to be Cordaitean; (ii) radially symmetrical seeds, circular in transverse section: the latter group he considered to be less closely allied to recent types. The employment of the terms Platyspermeae and Radiospermeae, proposed by Oliver[807] for Brongniart’s divisions, serves a useful purpose if due regard is paid to the adequacy of the evidence as to symmetry and if it is recognised that this classification cannot be rigidly employed in all cases. It was pointed out by Brongniart that the occasional occurrence of tricarinate seeds of _Ginkgo_ (fig. 631) and _Taxus_ is an exception to the general rule of bilateral symmetry: seeds of _Cycas_ are normally bilateral, but radially symmetrical forms also occur[808]. The genus _Conostoma_ (fig. 494, B) represents an intermediate type which, though almost radially symmetrical, exhibits a slight tendency towards platyspermy. Evidence recently brought forward by Nathorst[809] renders probable a connexion of a presumably radiospermic seed _Lagenospermum Arberi_[810] with the Lower Carboniferous fronds _Adiantites bellidulus_ Heer, and this furnishes an interesting illustration of the association of both platyspermic and radiospermic seeds with the same generic type of foliage. While retaining Radiosperm and Platysperm as convenient descriptive terms, I have not adopted them as group-designations on the ground that they do not in themselves serve as trustworthy criteria of relationship. Attention is called by Salisbury[811] to the occurrence of bilaterally and radially symmetrical fruits among British Carices and to a similar mixture in the family Polygonaceae.

The acquisition of more detailed and accurate knowledge of Palaeozoic seeds led to an extension of the two-fold division of Brongniart and Oliver which is based on such characters as the position of the vascular tissue in relation to the integument and nucellus, the form of the pollen-chamber, and other features. The division Lagenostomales has been instituted for _Lagenostoma_ and some other Radiosperms connected by certain important characters: these seeds may be referred to the Pteridospermeae though it is only in the case of _Lagenostoma_, and to a less extent _Sphaerostoma_, that a correlation between vegetative organs and seeds has been rendered sufficiently probable to justify an assumption of generic identity. The name Trigonocarpeae[812] has recently been used for a section of Radiosperms represented by _Trigonocarpus_, _Stephanospermum_, and other genera. Although the genus _Stephanospermum_, as Oliver[813] says, may be regarded as the type-genus of a group of seeds, it is more fitting, as the same author[814] insists, to adopt a divisional term based on the generic name of the much more widely spread and more familiar _Trigonocarpus_. For the sake of uniformity in nomenclature it is proposed to adopt the name Trigonocarpales instead of Trigonocarpeae to rank with Lagenostomales.

The Platyspermeae comprise such seeds as _Cardiocarpus_, _Mitrospermum_, and _Rhabdospermum_, genera characterised by well-marked anatomical features and probably Cordaitean; it has, however, been shown that typical Platysperms were also borne on leaves of Pteridosperms and, as Mrs Arber[815] says, the notion that every member of the Platyspermeae was necessarily a Cordaitean seed has been discredited by the discovery of the seeds of _Aneimites_ (_Wardia_) and _Pecopteris Pluckeneti_[816]. For general purposes it is hardly necessary to adopt the subdivisions of the Lagenostomales used by Oliver and Salisbury[817], though as facts accumulate we shall no doubt be able to make further advances towards a natural system of classification. The following three divisions of Permo-Carboniferous seeds include genera founded on petrified specimens and thus afford valuable morphological data. The groups Lagenostomales and Trigonocarpales include types belonging to closely related plants, a relationship clearly expressed in the seed-characters.

I. +Lagenostomales.+

The seeds included in this group are for the most part Radiosperms, but in its slightly developed bilateral symmetry _Conostoma oblongum_ is a type transitional between Radiosperms and Platysperms. The testa may be ribbed and the ribs vary in number. The nucellus (megasporangium) is united to the integument not only at the base but laterally as far as the shoulder of the seed up to a level corresponding to the base of the pollen-chamber (lagenostome) as in all recent Cycads and in the majority of Conifers. The seeds proper apart from the cupule are supplied with a single set of vascular bundles; there is no vascular tissue in the nucellus, a feature no doubt correlated with the fusion of nucellus and integument[818]. The free portion of the integument is more or less deeply lobed or, in _Lagenostoma_, it forms a pyramidal canopy of fused lobes enclosing the lagenostome. The presence of a tapetal zone surrounding the megaspore is believed to be a feature characteristic of the group[819]. The testa, wholly or partially ribbed, is relatively thinner than in the Trigonocarpales and Cardiocarpales, and in its differentiation agrees less closely with the testa of recent Cycadean seeds. In _Lagenostoma_ and possibly in other genera a loose sheath or cupule surrounded the ovule, while in _Gnetopsis_ a similar envelope enclosed two to four seeds.

The microspores are multicellular and smaller than those of Trigonocarpales, the average dimensions (_Conostoma_, _Physostoma_, _Lagenostoma_) being 67μ × 52μ.

Genera. _Physostoma_; _Conostoma_; _Sphaerostoma_; _Lagenostoma_; _Gnetopsis_.

_Lagenostoma_ may safely be referred to _Lyginopteris_, and _Sphaerostoma_ with but little risk of error to _Heterangium_: the parent-plants of the other genera are unknown, but all may be regarded as the seeds of Pteridosperms and probably of genera more nearly allied to the Lyginopterideae than to the Medulloseae. The genus _Lagenospermum_, recently instituted by Nathorst[820], is dealt with in Chapter +xxxi.+

=Physostoma.= Williamson.

_Physostoma elegans_ Williamson.

The generic name _Physostoma_[821] was applied by Williamson[822] to a seed from the Lower Coal Measures of Lancashire which he named _P. elegans_; he afterwards described it as _Lagenostoma physoides_, but the original name has been revived by Oliver[823] to whom our knowledge of this type is chiefly due. The specimens figured by Williamson[824] as _Sporocarpon ornatum_ also belong to _Physostoma elegans_. The seeds are circular in section, approximately 6 mm. long with a maximum diameter of 2 mm. The testa has about 10 longitudinal ribs passing in the apical region into a ring of free lobes or tentacles surrounding and considerably overtopping the nucellar apex: these tentacles take the place of a micropylar tube (fig. 494, I; fig. 493, D) and are a feature ‘in which this seed differs from all other known seeds, fossil or recent[825].’ A single vascular strand passes through the chalazal region and divides into 10 bundles, one to each rib and tentacle. The single integument consists of a few layers of cells, those of the epidermis being prolonged into clavate mucilaginous hairs, fig. 494, I, _h_, that may reach a length of ·5 mm. and in the living seed almost covered the whole surface of the testa, being especially long on the ribs and tentacles. There is no special development of sclerous tissue, the vascular bundles, _v_, being embedded in parenchyma in the inner portion of the integument. The nucellus is represented by a zone rich in secretory cells, and internal to this is a tapetum. Integument and nucellus are coalescent up to the apical region where the former splits into 10 tentacles. The nucellar apex has the form of a tall dome surrounded by a bell-shaped pollen-chamber (fig. 494, I, _pc_; fig. 493, C, D, _c_) into which it projects like the base of a wine-bottle. The circular opening of the pollen-chamber overtops the roof of the dome formed of the secretory tissue of the nucellus and the carbonised remains of the tapetum: this dark band surrounds the large megaspore-cavity (fig. 494, I). _Physostoma_ is the only member of the Lagenostomales in which the megaspore projects into the free nucellar apex: in other genera intercalary growth has produced a more or less prominent plinth, the name given to the free portion of the nucellus between the megaspore and the pollen-chamber. Williamson[826] described the mammillated apex of the nucellus as pushed up into the base of the lagenostome which ‘looks like a bladder half full of fluid resting upon and overhanging the end of a soda-water bottle’: it was this appearance that suggested the name _Physostoma_. The section reproduced in fig. 493, D, shows in the centre the limiting tissue of the nucellus surrounded by the pollen-chamber, _c_, and external to this are the tentacles with their groups of long hairs: the vascular bundles are represented by spaces in the more internal small-celled tissue (see also fig. 494, I). A characteristic feature is the presence of a tapetum or megaspore-jacket[827] in the nucellus: immediately internal to the vascular bundles stretching from the chalaza to the apex of the megaspore-cavity is a layer of delicate cells with secretory sacs, and this is succeeded by a broad black layer of rather larger cells, a tissue which was probably in full activity in a younger stage of development.

[Illustration: Fig. 493. A, _Lagenostoma ovoides_, longitudinal section of the lagenostome. B, _Lagenostoma Lomaxi_, transverse section of the lagenostome and canopy; _c_, pollen-chamber; _d_, space between the nucellus and integument. (A, after Miss Prankerd; B, from a section in the Manchester Museum, R 1048.) C, _Physostoma elegans_, longitudinal section showing microspores in the pollen-chamber, _c_. (From a section in the University College Collection.) D, _Physostoma elegans_, transverse section of the apical region showing the tentacles, pollen-chamber, _c_, etc. E, _Telangium Scotti_, transverse section of sporangia (Manchester Coll. R 1047). F, G, _Stephanospermum akenioides_. F, section of sclerotesta. G, tracheids from the nucellus. (University College Coll.)]

A comparable tapetal layer is described by Lang[828] in the ovule of _Stangeria_: the majority of the sporogenous cells surrounding the megaspore become disintegrated and are absorbed, but the outermost zone forms a more definite tapetal layer: as already suggested, this tissue in _Physostoma_ may be a group-character. No archegonia have been found, but in a few cases some of the delicate prothallus-tissue occurs in the interior of the seed. Microspores are often abundant in the pollen-chamber (fig. 493, C, _c_); in one seed 80 are recorded. The occurrence of so many microspores suggested to Oliver that insect-agency may have been responsible for the precision in pollination that is greater than one would expect in anemophilous plants. The spores are smaller than those of _Lagenostoma_ (55μ × 45μ) and in several of them the remains of a cellular tissue are preserved (fig. 494, N), also some sub-reniform bodies (fig. 494, M) similar to those described as spermatozoids by Dr Benson in _Lagenostoma_ (fig. 408, D).

The most interesting features of _Physostoma_ are: the absence of a continuous micropylar tube and its replacement by a circle of integumental lobes; the apical prolongation of the nucellar apex into the pollen-chamber, and the presence of long mucilaginous hairs on the integument. The large pollen-chamber is a character which distinguishes _Physostoma_ from _Conostoma_ and its form is very different from that in _Lagenostoma_.

The tentacles of the integument and the form of the nucellar apex are features consistent with Oliver’s view that _Physostoma_ is the most primitive of Palaeozoic seeds though, as Burlingame[829] says, the elaborate form of the encasing envelope marks a considerable advance beyond the earliest type of megasporangium integument.

A new type of _Physostoma_ has been briefly described by Gordon[830], without a specific designation, from the Lower Carboniferous beds of Pettycur (Fife): it was found in association with _Heterangium_ and _Sphaerostoma ovale_.

We have no knowledge of the plant to which _Physostoma_ belonged, but the general plan of organisation of the seed points to a near relationship to _Lagenostoma_ and presumably, as regards the parent-plant, to a genus related to _Lyginopteris_.

=Conostoma.= Williamson.

This name[831], suggested by the funnel-like pollen-chamber or lagenostome, was applied by Williamson[832] to some seeds from the Lower Coal Measures of Lancashire and Yorkshire and from the Lower Carboniferous beds of Burntisland. The Burntisland seeds, referred by Williamson to two species, have recently been united and described by Miss Benson as _Sphaerostoma ovale_[833]. The English species has been thoroughly investigated by Oliver and Salisbury[834] who have also described a second species, _C. anglo-germanicum_, from the Coal Measures of Lancashire and Germany.

_Conostoma oblongum_ Williamson.

This rare type is represented by approximately cylindrical seeds with an average length of 5 mm. and a maximum breadth of 2·3 mm. borne on a relatively stout stalk and tapering to a blunt apex characterised by a canopy of six short lobes (fig. 494, B, C) in marked contrast to the long tentacles of _Physostoma_. In the basal region the integument has six prominent ribs which soon die out when traced upwards: a transverse section through the body of the seed is hexagonal (fig. 494, D), the angles corresponding to the basal ribs, and there is a slight tendency to platyspermy. The testa has an epidermal mucilaginous layer which becomes exfoliated through the lifting-up of the cuticle by the underlying mucilage: below this, at the apex of the seed, is a cap of fleshy tissue (fig. 494, B, _sa_) which, it is suggested, may have had a secretory function in connexion with a drop-mechanism in pollination like that in recent Conifers. No microspores have been found in the pollen-chamber. The epidermis, called by Oliver and Salisbury the blow-off layer (fig. 494, B, _m_), together with the cap of soft tissue constitute a feebly developed sarcotesta. A sclerotesta consisting of a palisade-layer and a fibrous hypoderm extends over the main body of the seed below the epidermis; it forms the basal ribs and increases considerably in breadth at the apical region to form a sclerous cone penetrated by six strands of parenchyma enclosing vascular bundles (fig. 494, D) which pass up from the conducting tissue immediately external to the nucellus. The nucellus is coalescent with the integument, as in _Physostoma_, as far as the level of the domical free apex of the nucellus where the tapetal tissue that lines the seed-cavity passes across the almost flat top of the central region originally occupied by the megaspore. In some sections prothallus-tissue was found with an apical ‘tent-pole’ protuberance. A striking feature of _Conostoma_ is the mechanism for the reception of the microspores. The free part of the nucellus consists of the plinth, a broad tapering region originally filled with parenchyma but in most cases represented only by its epidermis: the plinth, _p_, is seen in fig. 494, B, to be two-storeyed, the upper and narrower storey being a space formerly filled by a pad of tissue suspended from the floor of the superposed lagenostome (pollen-chamber)[835]. The greater development of the domical plinth is a feature in which _Conostoma_ differs from _Physostoma_. At the apex of the plinth and resting on a slight depression is a small lagenostome, bowl-shaped in section, and like the pollen-chamber of _Lagenostoma_, formed as the result of enzyme-action on the apical papilla of the nucellus (fig. 494, B, B′, _l′_). The mouth of the lagenostome engages with the micropylar tube by a projecting flange (fig. 494, B′, _f_) of tissue lining the micropylar canal and by a second flange (_f′_) at the base of the lagenostome where the roof of the plinth (fig. 494, B′, _p_) bends downwards and inwards. The walls of the lagenostome are formed by strong cells with thickening bands giving them the appearance of tracheids (_l′_), but the floor of the lagenostome is made of thinner cells which become disorganised, allowing the microspores to fall into the large plinth-cavity below (_p_, fig. 494, B), an arrangement comparable with the two-storeyed pollen-chamber of _Bowenia_[836] and, to a less extent, with the micropyle of the Conifer _Tsuga_. The microspores are multicellular and ellipsoidal measuring 75μ × 65μ.

[Illustration: Fig. 494. A, _Stephanospermum akenioides_; _s_, space between integument and nucellus; _n_, nucellus; _e_, inner part of testa; _sc_, sclerotesta; _t_, _t_, tracheidal mantle; _pc_, pollen-chamber; _a_, archegonia; _td_, tracheidal pad. B–D, _Conostoma oblongum_; _a_, level of the transverse section shown in fig. C; _b_, level of transverse section D; _m_, lobes of tissue surrounding the micropyle; _sa_, sarcotesta; _v_, vascular bundles; _p_, plinth. B′, section of lagenostome and part of the integument; _f_, _f′_, flanges; _l′_, wall of micropyle-funnel. E, F, G, _Gnetopsis elliptica_; _t_, tent-pole. F, section of the plumes. G, section through the lagenostome. H, _Gnetopsis_ from Barnsley, Yorkshire, slightly enlarged. I, _Physostoma elegans_; _h_, hairs. K, L, _Mitrospermum compressum_; _cf_, inner flesh; _n_, nucellus and megaspore. L, transverse section. M, microspores and (?) male gametes of _Physostoma elegans_. N, microspore of _Physostoma_ showing part of exine and the internal cells. (A–I, N, M, after Oliver; H, Kidston Coll. No. 1255; K, L, after Mrs Arber.)]

The species _Conostoma anglo-germanicum_ agrees closely with _C. oblongum_ in general form and organisation, but it has eight ribs, four more prominent than the others, and differs also in other minor characters from the rather shorter seeds of the type-species. _Conostoma_ differs from _Lagenostoma_ in the absence of the tubular prolongation of the lagenostome, the micropyle being like that in recent Gymnosperms. In _Conostoma_ the tracheid-like elements of the lateral wall of the lagenostome are a characteristic feature, and no evidence has been found of the existence of a central core of tissue such as occupies the centre of the seed-apex in _Lagenostoma_. The long hairs of _Physostoma_ are represented in _Conostoma_ by the much smaller mucilaginous cells of the epidermis and in _Lagenostoma_ by the less closely united mucilage-cells of the superficial layer of the testa.

=Sphaerostoma.= Benson.

As already pointed out in Chapter +xxix.+ where this genus is described as probably the seed of _Heterangium_, there is a fairly close general resemblance between _Sphaerostoma_ and _Lagenostoma_. In the presence of free apical lobes the former genus resembles _Conostoma_, and while agreeing with _Lagenostoma_ in its annular pollen-chamber it is peculiar in the retention of an epidermis over the roof of the pollen-chamber: as in _Lagenostoma_ the seed is enclosed by an outer integument or cupule.

=Lagenostoma.= Williamson.

An account of this type of seed is included in the description of _Lyginopteris_[837]. The more striking peculiarities are exhibited by the pollen-chamber and the free region of the integument: the annular pollen-chamber (fig. 493, D, _c_; fig. 409) surrounds a central nucellar cone and is prolonged upwards as a tube engaging with the micropyle in contrast to the form of the pollen-chamber and the absence of a tubular prolongation in _Conostoma_. The tentacles of _Physostoma_ and the short apical lobes of _Conostoma_ are replaced by an apical cone formed by the coalescence of the integument containing nine cavities originally filled with parenchyma (figs. 409; 493, B). The presence of a cupule is a characteristic feature of young seeds, but from negative evidence in the case of most other seeds it is unsafe to assume that the cupule of _Lagenostoma_ is an exceptional feature. The nucellus and testa are united as far as the shoulders of the seed as in the seeds of recent Cycads and in contrast to their lateral independence in _Trigonocarpus_, _Stephanospermum_, and other genera.

=Gnetopsis.= Renault.

This generic name was given by Renault[838] to some small petrified seeds from the Stephanian of Grand’ Croix and to impressions from the Commentry coal-field which he believed to belong to some Gnetaceous plant. Saporta and Marion[839] and other authors have accepted these seeds as evidence of the existence of Palaeozoic Gnetales: it has, however, been shown[840] that _Gnetopsis_ has no claim to such relationship and is a type of seed closely allied to _Conostoma_. Renault described three species, afterwards adding three from another locality[841]; the genus is recorded also from Commentry[842] and Gard[843]. More recently Depape and Carpentier[844] have described examples from the Westphalian of Valenciennes which they place in the Pteridosperms in accordance with the conclusion of Oliver and Salisbury. _Gnetopsis_ has also been discovered by Mr Hemingway in the Middle Coal Measures of England[845] (fig. 494, H).

_Gnetopsis elliptica_ Renault.

The seeds of this species, slightly oval in section, occur in groups of 2–4 in a cupular investment (fig. 506, E, p. 359) described by Renault as an ovary but correlated by Oliver and Salisbury with the cupule which surrounds the single ovule of _Lagenostoma_. The cupule is lined with hairs similar to those on the wall of the cupule of _Lagenostoma_. A characteristic feature of the French seeds is the presence of three or four long plumes of hairs at the apex (fig. 494, E, F). As seen in fig. 494, E, a small lagenostome (pollen-chamber) rests on the roof of a broad plinth precisely as in _Conostoma_, and four vascular bundles, corresponding to the six bundles in _Conostoma_, pass into the apical cap of sclerous tissue enclosed by a sarcotesta, _sa_ (fig. 494, E, G). A ‘tent-pole’ prolongation (fig. 494, E, _t_) occurs at the apex of the prothallus. Renault described a portion of the integument as consisting of lacunar tissue which Oliver and Salisbury homologise with the superficial mucilaginous layer of _Conostoma_: this is seen above the sclerotesta in the apical region of fig. 494, E, _sa_.

_Gnetopsis anglica_ Kidston +MS.+

This species (fig. 494, H) is represented by seeds from the Middle Coal Measures near Barnsley, Yorkshire, 4 mm. long with apical appendages at least 2·2 cm. in length and probably four in number. The appendages do not show the hairs which form a prominent feature in the French specimens, but this is probably the result of imperfect preservation: there are indications of hairs on other specimens in Dr Kidston’s Collection. The type-specimen, in Dr Kidston’s Collection, was generously lent to me for examination.

_Gnetopsis_, while agreeing with _Conostoma_ in the more important features, is distinguished by the apical plumes, the very slight development of a tent-pole prolongation of the nucellar apex (fig. 494, E, _t_), the smaller number of vascular bundles, and by the presence of an enclosing cupule (fig. 506, E). In its slight departure from radial symmetry _Gnetopsis_ forms a transition between the Radiospermeae and the Platyspermeae. It is undoubtedly the seed of a Pteridosperm, but nothing is known as to the nature of the vegetative organs of the parent-plant.

II. +Trigonocarpales.+

In this group are included radially symmetrical seeds for the most part belonging to members of the Medulloseae. The peripheral zone of the nucellus is supplied with vascular tissue and the nucellus is free within the integument except at the base; it is superior and not semi-inferior[846] as in recent Cycads and in Lagenostomales. The ovule of _Myrica Gale_, in which the nucellus stands free within the single integument, affords an interesting parallel to seeds of this class in contrast to the usual Angiospermous type with a laterally coalescent nucellus. In _Myrica Gale_[847] the vascular supply is confined to the integument. There is a comparatively broad pollen-chamber and in some types the lateral tissue of the nucellus is prolonged as a tube within the micropyle. The usually ribbed testa is differentiated into an outer flesh (sarcotesta), a sclerotesta, and probably in most cases an endotesta or inner flesh. The ribs of the sclerotesta are in the majority of genera in multiples of three and in position correspond to the outer ring of vascular bundles. The presence of lacunar tissue in the sarcotesta of several genera may be correlated with dispersal by water. The apical region of the integument is not lobed but extends as a longer or shorter micropylar tube above the summit of the nucellus. In the differentiation of the testa, the form of the pollen-chamber, and in some other features, the seeds of this group present a general agreement with those of recent Cycads.

The microspores are multicellular and larger than those of the Lagenostomales: in _Stephanospermum akenioides_ they measure 160μ × 100μ[848] while in _Aetheotesta_[849] they reach 360μ × 290μ.

Genera. _Trigonocarpus_; _Tripterospermum_; _Ptychotesta_; _Hexapterospermum_; _Polypterospermum_; _Pachytesta_; _Stephanospermum_; _Polylophospermum_; _Codonospermum_; _Aetheotesta_; _Eriotesta_; _Gaudrya_.

=Trigonocarpus.= Brongniart.

A description of the morphological features of _Trigonocarpus Parkinsoni_ and _T. shorensis_ is given in the chapter on _Medullosa_ (p. 117), as there is good evidence that they are the seeds of that genus. There is considerable difference in size and to some extent in the form of casts included in _Trigonocarpus_ and, in the absence of anatomical data, it is hardly possible to determine the actual systematic position of many of the specimens so named. Dr Arber[850] has recently proposed a new generic name _Schizospermum_ for casts very like those of _T. Parkinsoni_, but distinguished by the splitting of the shell into three valves, a character which leads him to conclude that it is the external surface which is preserved and not a mere cast of the seed-cavity. It is, however, more probable that the specimens are casts of a split sclerotesta. In _Trigonocarpus pusillus_[851] the shell is divided into three valves, the dividing lines being marked by greatly reduced ribs, and in _T. schizocarpoides_ Grand’Eury[852], a species that may not be a true _Trigonocarpus_, there is also evidence of splitting. Arber points out that the species _Rhabdocarpus Boschianus_ Berg. is founded on a _Trigonocarpus_ from which the outer flesh has disappeared leaving the shell as the external covering. _Trigonocarpus_ seeds are widely distributed in Carboniferous and Permian rocks in Europe and North America: from the latter continent Newberry[853] has described several different forms that afford good examples of the abundance and variety of the genus. Some of the specimens included by Newberry[854] in _Trigonocarpus_ are probably distinct generic types: his species _T. multicarinatus_ may be identical with the ribbed cast shown in fig. 506, A. The casts described by Lindley and Hutton and by other authors as _T. Dawesi_[855] are nearly 5 cm. long, and if these are correctly included in the genus they point to the occurrence of seeds much larger than the type-species. The French species _Trigonocarpus pusillus_[856] Brongn., one of the smallest Palaeozoic seeds, from 6·5 to 15 mm. long, differs from _Trigonocarpus Parkinsoni_ and _T. shorensis_ in the absence of prominent ribs and in the much feebler development of the sarcotesta. Specimens of the German type _T. sporites_ Weiss, believed by some authors to be megaspores, were described by Zeiller[857] from Valenciennes as seeds: these are from 2·5 to 3·5 mm. long and have three small ribs. Zeiller quotes the presence of cell-outlines on the surface as evidence of their seed-nature, but it may be that this feature represents a sculpturing of the exine of a spore. Typical _Trigonocarpus_ seeds agree in several morphological characters with those of recent Cycads. They differ in the lack of a lateral union between nucellus and integument; the presence of nucellar tracheids, though a feature shared with _Bowenia_, distinguishes them from the majority of recent Cycadean seeds. In the comparatively long and fleshy micropylar tube a seed of _Encephalartos Lehmanni_ presents a fairly close resemblance to a _Trigonocarpus_. Salisbury has pointed out that the three species _T. Parkinsoni_, _T. shorensis_, _T. pusillus_ form a consecutive series illustrating the gradual disappearance of the secondary ribs that form a prominent feature in _T. Parkinsoni_; but for a comparison of these with other types of fossil and recent seeds the reader is referred to Salisbury’s summary[858].

=Tripterospermum.= Brongniart.

The seed on which this genus was founded by Brongniart[859] is clearly very closely related to _Trigonocarpus_ and, as Oliver[860] says, the distinguishing character described by the author of the genus is unimportant. Brongniart describes the type species, _T. rostratum_, as characterised by the presence of three prominent wings composed of a testa differentiated into an inner hard tissue and an outer lacunar tissue. It is, however, hardly possible to say whether the outer soft tissue was originally flattened in the form of ‘wings’ or pressed down on to the harder shell. Renault[861] notes the association of seeds that he refers to this genus with the leaves of _Dorycordaites_, but apart from the improbability of any connexion between _Tripterospermum_ and _Cordaites_, Renault’s seeds are too imperfect to demonstrate their identity with Brongniart’s genus. Kidston[862] has described an impression of a three-winged seed from the coal-field of Staffordshire as _Tripterospermum ellipticum_, a form described on page 357 as _Polypterocarpus anglicus_ (fig. 496, B).

=Ptychotesta.= Brongniart.

The type-species of this genus[863], _Ptychotesta tenuis_[864], about 3 cm. long, is characterised by six very prominent flanges or wings formed by the fissured or folded sclerotesta (fig. 495, B). There is no information as to the vascular supply or other anatomical details. It is not at all improbable that there is no real distinction between this genus and Brongniart’s genus _Hexapterospermum_.

=Hexapterospermum.= Brongniart.

In this genus Brongniart[865] included two species, _Hexapterospermum stenopterum_ and _H. pachypterum_: the shell is hexagonal in transverse section, each angle being prolonged as a narrow flange. In one of the sections figured by Brongniart (fig. 495, E) the ribs are not fissured: this is said to be a feature distinguishing _Hexapterospermum_ from _Ptychotesta_, but the occurrence of a fissured rib in another section suggests that in the structure of the ribs there is no essential difference between the two genera. In _Ptychotesta pachypterum_ the testa is prolonged at the chalazal end as in _Polylophospermum_, and it is possible that there is no generic difference. Williamson described a cast from the Coal Measures of Lancashire as _Hexapterospermum_ [= _Hexagonocarpus_] _Noegerrathi_[866] (fig. 506, H), but in the absence of anatomical characters it is preferable to avoid the use of Brongniart’s term and to assign them to Renault’s genus _Hexagonocarpus_[867]. Similarly the seeds referred by Dr P. Bertrand[868] to _Hexapterospermum_ may appropriately be included in the genus _Hexagonocarpus_.

[Illustration: Fig. 495. A, _Polypterospermum Renaultii_. B, _Ptychotesta tenuis_. C, D, _Polylophospermum_ _stephanense_; _sa_, sarcotesta; _s_, spaces; _v_, vascular bundles. E, _Hexapterospermum stenopterum_. F, _Diplotesta avellana_. G, _Taxospermum_ _Grüneri_. H, _Diplotesta Grand’Euryana_. (After Brongniart; C, after Oliver.)]

=Polypterospermum.= Brongniart.

This generic name was proposed for an ovoid seed from St Étienne described as _Polypterospermum Renaultii_[869]; hexagonal in section with six deep and sharp flanges at the angles alternating with six secondary ridges distinguished by their blunter edges and slightly smaller depth (fig. 495, A). Without further anatomical details it is hardly possible to say whether or not the species represents a well-defined generic type, but it is not improbable that a fuller knowledge would confirm Brongniart’s institution of a new generic designation. The seeds described by Arber[870] and Kidston[871] respectively as _Radiospermum ornatum_ and _Polypterospermum ornatum_ are transferred to _Polygonocarpus_ on the ground that they afford no evidence of anatomical characters of the _Polypterospermum_ type.

=Pachytesta.= Brongniart.

Brongniart[872] established this genus for some unusually large seeds (fig. 496, A) from the Coal Measures of France reaching a length of 11–12 cm. and resembling in form and size a hen’s egg. Petrified examples have been described by Brongniart and Renault[873], and Oliver[874] has given a lucid statement of the more interesting features of this Permo-Carboniferous genus.

_Pachytesta gigantea_ (Grand’Eury).

In the separation of nucellus and integument _Pachytesta_ agrees with _Stephanospermum_ and _Trigonocarpus_. There is a double series of vascular bundles in the outer part of the testa or exotesta (fig. 497, _ex_) consisting of a spongy tissue bounded externally and internally by sclerotic layers: the exotesta is succeeded by a broader endotesta of spongy consistency which Renault suggested may have served as a floating mechanism, and this is intimately associated with the nucellus by means of grooves engaging with short ridges (fig. 497, _nr_) on the surface of the nucellar tissue. The exotesta is divided into three valves by radial extensions of the sclerotic tissue (_b_, _c_, fig. 497; at _a_ the exotesta is shown in an unsplit condition). In connexion with each radial plate are two curved plates of fibrous tissue which extend towards the grooves in the endotesta (fig. 497, _t_): as Oliver suggests, these plates may have served a mechanical purpose for the support of the bulky nucellus. The peripheral region of the nucellus is supplied by several vascular bundles (fig. 497, _n_) from the tracheal chalazal disc. The exotesta is regarded by Oliver as corresponding to the sarcotesta and sclerotesta of other seeds, and the grooves in the endotesta interlocking with the nucellar ridges are compared with the interlocking of nucellus and canopy in _Lagenostoma_, as also with the ruminated endosperm of _Torreya_.

[Illustration: Fig. 496. A, _Pachytesta incrassata_. B, _Polypterocarpus anglicus_. (A, after Renault; B, after Kidston.)]

[Illustration: Fig. 497. _Pachytesta_. Transverse section of a seed showing the exotesta, _ex_, with sclerous layers and vascular strands (black), the endotesta (dotted), the nucellus with its ring of vascular bundles, _n_, and the prothallus, _p_. The three radial plates in the exotesta are seen at _a_, _b_, _c_, with the trabeculae, _t_, and internal to these are the grooves in the endotesta engaging with ribs on the nucellus, _nr_. (After Oliver.)]

We have no definite information as to the plants which bore _Pachytesta_, but it is probable that they were members of the Medulloseae. Grand’Eury[875] believes _Pachytesta_ to be the seed of _Alethopteris Grandini_, though this view requires confirmation. This author figures several _Pachytesta_ seeds attached in two oblique rows to a comparatively slender axis which may be the rachis of a large compound frond[876]. Renault and Zeiller[877] have figured specimens of _P. gigantea_ and _P. incrassata_ from Commentry which afford a good idea of the form of these large seeds; the genus is recorded also from Gard, Blanzy[878], and other localities. Kidston[879] described a large oval seed, blunt at each end, from the Middle Coal Measures of Lancashire as _Carpolithus Wildii_ which he thinks may be allied to _Pachytesta_, but adds, ‘I do not think we are justified in placing mere impressions of plants in genera, whose distinctive characters are derived from their internal organisation, unless there is conclusive evidence to show their identity.’ On this specimen Arber[880] has founded a new genus _Megalospermum_, but as the type-specimen shows no distinctive features other than large dimensions it would seem preferable to retain the more general designation _Carpolithus_[881].

Another example of a seed that may be generically identical with _Pachytesta_ is that described by Lesquereux from North American Coal Measures as _Rhabdocarpus Mansfieldi_ and more recently recorded by White from Missouri as _Rhabdocarpus (Pachytesta) Mansfieldi_[882]. There is little doubt that this and other species of seeds preserved as impressions are examples of _Pachytesta_ but, as Kidston says, it is preferable to reserve the name for specimens showing anatomical features.

=Stephanospermum.= Brongniart.

The genus _Stephanospermum_, founded by Brongniart[883] on petrified specimens from French Stephanian beds, affords a good example of a radiospermic seed without ribs differing in certain well-marked characters from such seeds as _Lagenostoma_, _Physostoma_, and _Conostoma_, notably in the complete separation of the nucellus from the integument and in the possession of a nucellar vascular system. The descriptions by Brongniart and Renault[884] have been extended by the thorough investigations of Oliver[885].

_Stephanospermum akenioides_ Brongniart.

This species is represented by ellipsoidal seeds, 10 mm. long with a maximum breadth of 4–4·5 mm., circular in transverse section. The integument consists of a thick sclerotesta lined internally by a thinner soft layer, the endotesta (fig. 494, A, _sc_, _e_), and in all probability covered by an outer flesh or sarcotesta as in _Trigonocarpus_, though this tissue is not preserved and is omitted from the section shown in fig. 494, A. The sclerotesta is composed of a broad outer band of thick-walled palisade cells—the contracted contents of which are a striking feature in the silicified seeds (fig. 493, F); these are succeeded by an inner zone of longitudinal fibres. In the structure of the palisade-layer _Stephanospermum_ bears a close resemblance to the sporocarp wall of _Pilularia_[886]. In the apical region the shell forms a circular ridge surrounding a peri-micropylar trough, a character expressed by the term ‘crown-seed[887]’ employed by Grand’Eury: the trough sometimes contains partially destroyed tissue that may be a remnant of a sarcotesta. A _Stephanospermum_ seed, with its apical core and surrounding depression with remnants of some partially destroyed tissue, may be compared with a fruit of _Attolia speciosa_ (Palmae) in which a tuft of fibrous tissue picked out by decay from the mesocarp surmounts the conically pointed apex of the harder interior of the fruit-wall. The base of the sclerotesta is perforated by a vascular strand which expands into a tracheal disc, _td_, fig. 494, A, at the base of the megaspore from which a continuous mantle 2–3 cells broad, of short spiral and scalariform tracheids (fig. 493, G), spreads over the whole of the nucellus immediately below the nucellar epidermis as far as the lower part of the sides of the pollen-chamber: this mantle forms the floor of the large domical pollen-chamber excavated out of the nucellar cone (fig. 494, A, _pc_). The fact that in sections of older seeds the tracheal floor shows signs of splitting and disorganisation led Oliver to conclude that in the living seed the tracheids underwent a gradual disintegration prior to fertilisation, thus allowing the passage of the antherozoids to the egg-cells (fig. 494, A, _a_). The presence of a continuous tracheal sheath instead of separate vascular bundles is a special feature in which _Stephanospermum_ differs from _Trigonocarpus_ and other Palaeozoic seeds as well as from those of recent Cycads; as Oliver suggests, ‘the apparent perfection of the vascular mantle in _Stephanospermum_ may have proved an obstacle to further development[888]’ and was not retained by the more successful types. In its tracheal sheath _Stephanospermum_ resembles the seeds of _Ginkgo_. The nucellar cone is prolonged as a beak into the micropyle formed by the tubular integument. The megaspore occupies the central portion of the seed and in the course of its development it compressed the megasporangium (nucellus) to such an extent that little more than the epidermis remains: there is a definite megaspore-membrane surrounding the prothallus-tissue in which there were probably two archegonia (fig. 494, A, _a_)[889]. The nucellus stands free within the integument from which it is separated by a space (_s_, fig. 494, A). Microspores are frequently met with in the pollen-chamber in which they matured after their introduction through the micropyle, probably with the aid of a drop of mucilage: they are large oval bodies with an average size of 160μ × 100μ, some reaching over 200μ, in length and, as Renault was the first to point out, within a finely granulate exospore there are several thin-walled cells: this statement was not accepted by some authors but Prof. Oliver’s researches have amply confirmed it, and an examination of the original preparations convinced me that Renault had correctly described the structural features. Oliver shows that there are about 20 cells within each microspore regularly arranged as rows of five wedge-shaped elements with their pointed ends directed towards the centre, and he thinks that these cells may have undergone further division to produce sperm-mother-cells, though there is not such good evidence of this as in the differently constituted microspores of _Stephanospermum caryoides_. No trace of pollen-tubes was found and it is probable that the antherozoids were liberated by the rupture of the delicate prothallus-tissue. The tracheal sheath afforded an adequate means of water-supply to the pollen-chamber and this enabled the motile antherozoids to reach the archegonia.

_Stephanospermum caryoides_ Oliver[890].

This species, founded on a specimen from ., is a larger seed (15·5 × 12·5 mm.) and more globular than _S. akenioides_; there is a shorter micropylar beak and a less developed peri-micropylar trough. As in _S. akenioides_ the nucellus is free from the base and the two species conform to the same general type of construction. The microspores are distinguished by the presence of only two internal cells which do not occupy the whole of the spore-cavity but are surrounded by a large peripheral cell comparable with the tube-cell in recent microspores, though there is no proof that a tube was formed: in the case of _S. akenioides_ Oliver considers that fertilisation was not assisted by the production of a pollen-tube. The two cells by further division gave rise to a secondary cell-complex consisting of at least eight antherozoid-mother-cells. No antherozoids have been found in the microspores but it is possible that two small bodies, 17μ × 15μ, met with in a pollen-chamber may represent the nuclei of sperms. Their small size differentiates them from the much larger male gametes of Cycads and from the supposed sperms of _Physostoma_ and _Lagenostoma_.

We have no proof as to the nature of the plant which bore seeds of the _Stephanospermum_ type, but it is significant that the specimens occur in close association with fragments of _Alethopteris_ and _Myeloxylon_, a circumstance that favours the view, based on the resemblance of these seeds to _Trigonocarpus_, that _Stephanospermum_ is the seed of a member of the Medulloseae.

=Polylophospermum[891].= Brongniart.

The type-species _Polylophospermum stephanense_ Brongn.[892], founded on partially preserved material from Grand’ Croix, is a narrow hexagonal seed 15 mm. long. Additional facts as to the structural features have been contributed by Oliver[893]. The testa is differentiated into an inner shell (sclerotesta) and an outer flesh (_sa_, fig. 495, C, D): the sclerotesta has six prominent, fissured, ribs, one at each angle, and between these are six solid and less prominent secondary ribs. Oliver states that there is an outer series of vascular bundles in the sarcotesta, one bundle immediately external to each secondary rib (fig. 495, C, _v_). In the presence of two kinds of ribs and in the relation of ribs to tracheal strands _Polylophospermum_ agrees with _Trigonocarpus Parkinsoni_. Strands of short tracheids supply the peripheral region of the nucellus and, as in _Stephanospermum_, reticulate elements extend as far as the floor of the large pollen-chamber. There was probably no lateral union between nucellus and integument. A striking feature is the prolongation of the testa at each end of the seed to form an open chamber surrounding the micropylar beak and the seed-base (fig. 495, D, _s_, _s_): the apical chamber, though relatively more prominent, is comparable with that in _Stephanospermum_, while the basal chamber recalls that in the seed named by Scott and Maslen _Trigonocarpus Oliveri_[894] but subsequently removed by Salisbury[895] from that genus. There is no evidence as to the parent-plant of _Polylophospermum_, but it may be assumed to have been a Pteridosperm, probably one of the Medulloseae.

[Illustration: Fig. 498. A, B, _Codonospermum_. A, _C. olivaeforme_, longitudinal section showing the seed-proper and the air-chamber, _s_; _pc_, pollen-chamber. B, Diagrammatic sketch of a _Codonospermum_ showing the vascular supply, _v_, and the fibrous arcs, _f_, in the wall of the lower part of the seed. C, _Aetheotesta elliptica_, _s_, sarcotesta with lacunae; _v_, vascular supply. (After Renault.)]

=Codonospermum[896].= Brongniart.

Our knowledge of this peculiar genus is derived from Brongniart’s description of the type-species _Codonospermum anomalum_[897] (fig. 506, B, C) from St Étienne and from Renault’s account of _C. olivaeforme_[898]. The seeds are globular or ellipsoidal and reach a length of 2 cm.: the testa is differentiated into an outer flesh and a sclerotesta and has usually eight ribs. The most striking feature is the division of the seed into two regions, an upper portion containing the nucellus and megaspore and a lower portion in the form of an empty chamber that probably served as a float (fig. 498, A, _s_). Externally the upper half or seed-proper forms a depressed cupola with eight slightly developed ribs, separated by a circular transverse constriction from the basal chamber (fig. 498, B). There is a fairly large pollen-chamber, _pc_, in which Renault found multicellular microspores in _C. olivaeforme_. In _C. anomalum_ 16 vascular bundles (fig. 498, B, _v_) surround the central region of the seed probably in the peripheral tissue of the nucellus; these bundles unite in the chalazal region to form a strand that passes up the centre of the empty chamber. In _C. anomalum_ the testa of the lower half has eight ribs and corresponding with each rib is a strand of fibrous tissue (fig. 498, B, _f_). In _C. olivaeforme_ the testa is thicker than in _C. anomalum_ and the lower region of the seed is smooth and circular in section.

Impressions of _Codonospermum_ are described[899] from Commentry, from the Loire—the Gard district[900], and by Zeiller[901] from Blanzy. There is no decisive evidence as to the parent-plant, but some French authors[902] regard the frequent association of _Codonospermum_ with leaves of _Dolerophyllum_ as significant.

=Aetheotesta.= Brongniart.

_Aetheotesta elliptica_ Renault.

Brongniart instituted the genus _Aetheotesta_[903] for an incomplete seed (_A. subglobosa_) from Grand’ Croix, and Renault[904] subsequently founded the species _A. elliptica_ on much better material. The testa of the elliptical seed, 3 cm. long, consists of an outer region very thin on the flanks but highly developed at the apex and base which forms a sarcotesta characterised in the apical region by large radially disposed spaces, fig. 498, C, _s_; the sclerotesta, composed of harder tissue, is prolonged as an apical beak. There is a large pollen-chamber, _pc_, in which Renault found multicellular microspores (320μ–400μ). In the chalazal region the main vascular supply forms a cup-like investment, _v_, in the basal portion of the nucellus, and at a higher level this breaks up into several nucellar strands. Renault thinks that _Aetheotesta_ is the seed of _Dolerophyllum_, but there is no proof of any connexion. The presence of large spaces in the sarcotesta may be interpreted, as Renault suggests, as evidence of adaptation to dispersal by water.

=Eriotesta.= Brongniart.

Brongniart[905] instituted this genus[906] on an incomplete transverse section of a seed from Grand’ Croix, 8 mm. in diameter, which he called _Eriotesta velutina_, characterised by a ribbed and probably octagonal testa bearing numerous elongated hairs over the whole surface. The material is, however, too meagre to throw any light on the important features of the seed.

=Gaudrya.= Grand’Eury.

_Gaudrya trivalvis_ Grand’Eury.

The generic name _Gaudrya_[907] was proposed for two petrified seeds from the Gard coal-field briefly described as _G. trivalvis_, the type-species, and _G. lagenaria_. The testa of the former species shows signs of splitting along three equidistant lines; it consists of a sarcotesta enclosing an endotesta with six longitudinal ribs: Grand’Eury speaks of lacunae between the endotesta and nucellus which he regards as spaces in a tissue which made the seed buoyant and facilitated dispersal by water. It is not clear whether the nucellus and integument were originally connected or laterally free: the lacunae may be the remains of an inner flesh and not spaces in a spongy tissue. The genus is radiospermic and characterised by a long micropyle and a broad pollen-chamber. On the available evidence it is difficult to assign the specimens to their systematic position, but the genus is probably a member of the Trigonocarpales.

III. +Cardiocarpales.+

Platyspermic seeds for the most part belonging to Cordaitalean plants and agreeing in their plan of organisation, even more closely than the seeds of the Trigonocarpales, with those of recent Cycads. The nucellus is free laterally from the integument and there is a series of vascular bundles in the nucellus. The testa consists of an outer flesh, the sarcotesta, which may reach a considerable thickness, a shell or sclerotesta, and in some genera an endotesta. In seeds preserved as impressions the sarcotesta often gives them a winged appearance (_Samaropsis_). The pollen-chamber is relatively small, resembling in this respect the chamber in recent Cycads. A central prolongation of the prothallus-tissue in the form of a blunt column or ‘tent-pole’ is a characteristic feature; the same feature occurs in seeds of the Lagenostomales, but in the Cordaitales it resembles more nearly the ‘tent-pole,’ as it was called by Hirase, in recent Cycadean seeds and in _Ginkgo_. The presence of two vascular strands in the sarcotesta in the principal plane of the seeds is a character shared with the seeds of _Cycas_: the position and course of these bundles are useful characters for distinguishing different types within the group. The microspores are multicellular.

Genera. _Cardiocarpus_; _Cyclospermum_; _Cycadinocarpus_; _Rhabdospermum_; _Mitrospermum_; _Diplotesta_; _Leptocaryon_; _Taxospermum_; _Compsotesta_. These generic names are all used in the following pages for seeds known to possess certain anatomical features; there are also included in the Cardiocarpales the genera _Samaropsis_, _Cordaicarpus_, and _Rhabdocarpus_, but it is proposed to limit their use to specimens which furnish no anatomical data and cannot therefore be assigned with equal confidence to a section of seeds based on definite morphological characters. It is certain that some at least of the seeds described under these names would, if preserved as petrified specimens, be included in one or other of the genera named above.

There is ample proof that some of these seeds were borne on Cordaitean plants and that the group as a whole represents the seeds of the Cordaitales[908]. It is, however, certain that some Platyspermic seeds were produced by Pteridosperms. No little confusion has been caused by the employment of the same generic names for petrified seeds and for casts and impressions affording no evidence as to similarity in anatomical characters. With a view to avoid the risks necessarily entailed by following this practice it is suggested that a clearer distinction should be drawn between genera based primarily on structural features and form-genera. The following notes on the genera _Cardiocarpus_, _Cordaicarpus_, _Cyclocarpon_, _Cycadinocarpus_, _Jordania_, and _Samaropsis_, may serve to illustrate some of the difficulties connected with the terminology of Palaeozoic seeds.

_Cardiocarpus_. Brongniart[909] proposed the name _Cardiocarpon_ in 1828 for Upper Carboniferous seeds described as compressed lenticular, cordiform or reniform ‘fruits’ with an acute apex: in his later work[910] he recognised their true morphological nature and gave an account of some exceptionally well-preserved examples from Grand’ Croix. Brongniart in common with other authors believed _Cardiocarpus_ seeds to belong to Cordaitean plants, a view that in several cases is based on conclusive evidence. The specimen represented in fig. 501, D, illustrates the characteristic form of a cast of a _Cardiocarpus_ seed, and the sections shown in fig. 501, A and B, are from an identical or a very closely allied species. The generic characters are: (i) the presence of a narrow flattened border or wing surrounding a platyspermic nucule, (ii) the cordiform base and more or less pointed apex, (iii) the differentiation of the testa into a sarcotesta and sclerotesta free from the nucellus except at the base, (iv) the ‘tent-pole’ prolongation of the prothallus (fig. 510, A, _b_) and the presence of a fairly large pollen-chamber, _pc_, (v) the occurrence of two sets of vascular bundles, an inner nucellar series and two double vascular strands (fig. 500, A, _v_, B) which are given off from the main supply before it reaches the sclerotesta. The term _Cardiocarpus_ as used by Brongniart signifies a type of seed possessing both certain anatomical and external characters. The proposal is to restrict the generic appellation to seeds exhibiting definite structural features agreeing in essentials with _Cardiocarpus sclerotesta_ and _C. drupaceus_.

_Cordaicarpus_. This name was first employed by Geinitz[911] in the form _Cordaicarpon_, the type-species being _C. Cordai_ (fig. 502, C) from the Coal Measures of Germany, a seed referred by the author of the genus to _Cordaites principalis_ but, as Kidston has pointed out, there is evidence that this correlation may be incorrect: there is, however, no doubt as to its Cordaitean parentage. Specimens included in this genus agree closely with species of _Cardiocarpus_, but they are usually described as being distinguished by the absence of a flat border and by a more rounded and less cordate base. In the example of _Cordaicarpus Cordai_ shown in fig. 502, C, and in other species assigned by authors to this genus there is a narrow border and the form of the base is an inconstant character. As Kidston[912] and other authors point out, there are no definite and constant characters by which to distinguish _Cardiocarpus_ from _Cordaicarpus_ as regards the form of the seeds preserved as casts or impressions. A further account of _Cordaicarpus_ is given on a later page.

_Cordaispermum_. This designation was formerly adopted by Renault[913] for seeds having the form and anatomical features of _Cardiocarpus_ which there is good reason for attributing to _Cordaites_ or to some allied genus. In view of the fact that the majority of the seeds under consideration are undoubtedly Cordaitean there is no need to employ this additional generic name.

_Cyclocarpon_. Fiedler[914] instituted this term for seeds previously described by Berger as _Cardiocarpon emarginatum_ (fig. 502, B, now included in _Samaropsis_) and added a new species _Cyclocarpon nummularium_. Brongniart[915] subsequently described the structure of two species, _C. tenue_ and _C. nummulare_, which he referred to Fiedler’s genus, and Bertrand[916] has shown that these differ from the genus _Cardiocarpus_ in the recurrent course of the bundles given off from the chalazal strand as in the genus _Rhabdospermum_ (cf. fig. 501, E). There are no good grounds for retaining the designation _Cyclocarpon_ for casts and impressions, as the specimens so named are indistinguishable from impressions referred to _Cordaicarpus_. The generic name _Cyclocarpus_, retained by Bertrand only for _C. tenuis_ and _C. nummularis_ simply for anatomical reasons, is now altered to _Cyclospermum_ on the ground that the designation _Cyclocarpon_ has been used for impressions affording no information with regard to anatomical features.

_Cycadinocarpus_. Renault[917] transferred to this genus Brongniart’s species _Cardiocarpus augustodunensis_ on the ground that the vascular system exhibits in a greater degree than the other types included by Brongniart in _Cardiocarpus_ a resemblance to that in recent cycadean seeds. Bertrand[918] confirms Renault’s account and retains _Cycadinocarpus augustodunensis_ as a species worthy of generic distinction. A short account of this seed is given on a later page.

_Jordania_. This name was given by Fiedler[919] to compressed ovate-cordate seeds characterised by a broad membranous border bearing a superficial resemblance to the seeds of _Bignonia_. The type-species, _J. bignoniodes_ (fig. 502, I), from the Coal Measures of Saarbrücken has the form usually associated with the designation _Samaropsis_, and as the latter term is generally adopted there are no adequate reasons for the retention of _Jordania_. The name _Jordania_ has also been applied, by Schenk[920], to fossil Dicotyledonous wood and was previously used by Boissier[921].

[Illustration: Fig. 499. _Samaropsis emarginata_, from the Westphalian series, Yorkshire. (Kidston Coll., 4227; × 2.)]

_Samaropsis_. Goeppert[922] defined _Samaropsis_ as ‘Fructus samaroideus membranaceus, compressus, margine alatus, monospermus.’ The type-species, _S. ulmiformis_, from the Permian of Brenau, is a small seed with a broad wing or border, but a better example of _Samaropsis_ is figured by Goeppert as the wing of an insect[923]. Examples of the genus are shown in figs. 502, A–H; 503; 504. Seeds included in _Samaropsis_ differ from those referred to _Cordaicarpus_ in the presence of a broader and more clearly defined border which in some cases, as in the genus _Mitrospermum_ (fig. 494, K, L), undoubtedly represents a lateral wing-like extension of the sarcotesta. In some instances the wing may be a tangentially expanded integument comparable with the perianth of _Welwitschia_, and in some Jurassic seeds referred by Heer[924] to _Samaropsis_ the lateral appendages are probably true wings. It is advisable to restrict the designation _Samaropsis_ to Palaeozoic seeds. Nucules deprived of the broad border would be referred to _Cordaicarpus_ as usually employed for impressions. The generic name _Samaropsis_ serves a useful purpose as a distinctive term for platyspermic seeds preserved as casts or impressions characterised by the possession of a wide border or wing broader than in typical examples of _Cordaicarpus_. The specimen represented in fig. 499 affords a good illustration of the difference between _Samaropsis_ and _Cordaicarpus_. In this specimen the border clearly consists of two portions, an inner narrower border (black in the drawing) and an outer more delicate portion; the former is the impression of the sclerotesta and the outer represents the fleshy sarcotesta which in the living seed may have formed a wing. If, as often happens, the seed were preserved with the narrow border only it would be assigned to _Cordaicarpus_, many species of which are undoubtedly incomplete _Samaropsis_ seeds.

The seeds described by Lindley and Hutton as _Cardiocarpon acutum_ (fig. 444, p. 171) have been made by Arber[925] the type of a new genus _Cornucarpus_, the distinguishing feature being the triangular form and the apical horns of the wing. The seeds figured by Arber[926] from the Kent coalfield as _Cornucarpus acutus_ are, however, not identical with the type of Lindley and Hutton, which has the characters of _Samaropsis_. _Samaropsis_ is widely distributed in Permo-Carboniferous rocks in Europe and North America and is recorded also from India[927] (fig. 504), China[928], South Africa[929] (fig. 503), South America[930] (fig. 502, F, G) and Australia[931]. Some seeds of this form were certainly borne on Cordaitean plants (cf. fig. 480, A), but seeds of similar type have been found in organic connexion with the foliage of Pteridosperms (figs. 442, 445, pp. 167, 172). The Permian ‘cone-scales’ bearing seeds described by Geinitz as _Cardiocarpon triangulare_[932], represented by well preserved impressions in the Dresden Museum, appear to be of the _Samaropsis_ type: the same author referred some _Samaropsis_ seeds to the Conifer _Walchia_, but Weiss[933] dissents from this correlation as the seeds often occur in beds in which _Walchia_ is not represented. Renault states that the seeds of the fertile shoot described by him as _Cycadospadix Milleryensis_[934] from Autun [= _Strobilites Milleryensis_ (Ren.)] closely resemble _Samaropsis fluitans_ Daws. as figured by Weiss. The suggestion by White[935] that _Samaropsis_ seeds were borne on fertile leaves of _Gangamopteris_ adds a further difficulty to the use of the generic characters of _Samaropsis_ as criteria of systematic position. The Permian seeds figured by Goeppert[936] as _Oreodoxites Martianus_ are possibly specimens of _Samaropsis_.

The designation _Samaropsis_, though usually restricted to Palaeozoic species, is applied by some authors to ‘winged’ seeds from Mesozoic strata; but as some of the Jurassic seeds[937] so named appear to have true wings like those of some recent Conifers it is advisable to adhere to the more limited use of the name.

It is safe to assert that many _Samaropsis_ seeds agreed generally in structure with Cordaitean seeds such as the petrified examples described by Brongniart as _Cardiocarpus_. The species _Mitrospermum compressum_[938] is an example of a petrified seed having the external features of _Samaropsis_.

=Cardiocarpus.=

This generic title I propose to restrict to petrified seeds exhibiting the characters described by Brongniart[939] and more recently by Bertrand[940] in _C. sclerotesta_ and _C. drupaceus_. In general organisation seeds of this generic type agree with those of recent Cycads and with the seed of _Ginkgo biloba_, but there are certain distinguishing features. An important character is afforded by the course and place of origin of the lower vascular strands from the main supply at the base of the seed. The outer vascular system consists of two bundles given off from the main strand, before it reaches the sclerotesta, which pass up the sarcotesta (fig. 500, B). In _Rhabdospermum_, on the other hand, the corresponding bundles arise at a higher level and form recurrent strands which penetrate the sclerotesta before passing up the fleshy part of the integument (cf. fig. 501, E).

[Illustration: Fig. 500. _Cardiocarpus drupaceus_ var. _expansus_. A, transverse section of the testa approximately in the middle of the seed; _v_, vascular bundles. B, diagrammatic sketch, adapted from Brongniart’s drawing of the base of the seed showing the sarcotesta, _sa_, the sclerotesta, _sc_, and the pad of vascular tissue, _d_, at the base of the nucellus. (After Brongniart.)]

_Cardiocarpus sclerotesta_ Brongniart. The testa is differentiated into an inner shell and an outer sarcotesta (fig. 501, A; the sclerotesta is shown in black); there is a well developed pollen-chamber (_pc_) and below this the prothallus-tissue is prolonged as a blunt and short tent-pole, _b_, as in _Gingko_ and in several fossil seeds. On each side of the apical tent-pole the slightly shrunken prothallus shows two small archegonia, _a_, which in the relatively small size and spherical form of the egg-cells resemble those of _Ginkgo_. In transverse section (fig. 501, B) the seed is bi-convex and at each end of the major axis the sclerotesta forms a small keel. There are two sets of vascular bundles concerned in the supply of material to the ovule; a lower pair of bundles given off from the central strand in the sarcotesta (fig. 500, B) which pass to the apical region in the inner tissues of the sarcotesta in the principal plane (fig. 500, A, _v_), and an inner set of bundles that pass up the peripheral tissue of the nucellus.

The species described by Brongniart as _Cardiocarpus (Cyclocarpus) tenuis_ and _C. nummularis_ have been removed by Bertrand[941] from _Cardiocarpus_ on the ground that the integumental bundles pursue a course like that in _Rhabdospermum_; it is now referred to the new genus _Cyclospermum_[942].

[Illustration: Fig. 501. A, _Cardiocarpus sclerotesta_ in longitudinal section; _pc_, pollen-chamber, _b_, tent-pole, _a_, archegonia. B, transverse section; _a_, archegonia. D, cast of the same or a similar type of seed. C, E, _Rhabdospermum cyclocaryon_, longitudinal section and a diagrammatic sketch of the seed-base; _m_, micropyle; _b_, tent-pole; _a_, archegonium. (A, B, C, after Brongniart; D, after Renault and Zeiller; E, adapted from Brongniart.)]

=Cyclospermum.= Gen. nov. (= _Cyclocarpus_ Bertrand).

As stated on a previous page Bertrand[943] re-establishes the generic name _Cyclocarpus_, founded on impressions without reference to anatomical characters, for two petrified seeds from St Étienne described by Brongniart as _Cardiocarpus tenuis_ and _Cyclocarpus nummularis_[944]. These types differ from _Rhabdospermum_ in the absence of an apical snout but agree with that genus in the steeply descending course of the vascular strands in the basal region of the seed. As stated on a previous page, the name _Cyclospermum_ is proposed as a substitute for _Cyclocarpus_ because of the employment of the latter term for impressions.

=Cycadinocarpus.= Schimper[945].

_Cycadinocarpus augustudunensis_ (Brongniart[946]). In the possession of two sets of vascular bundles this type agrees with _Cardiocarpus_, but the more internal strands pass up on the inner face of the sclerotesta without penetrating into the nucellus, a feature in which _Cycadinocarpus_ agrees with the majority of recent cycadean seeds: the outer bundles are given off from the main supply after it has entered the sclerotesta and not before as in _Cardiocarpus_; they follow an oblique course in the sclerotesta and emerge into the sarcotesta at the shoulders of the basal curve of the seed. As in _Cardiocarpus_ and _Rhabdospermum_ the two outer bundles lie in the principal plane of the seed. There is a pollen-chamber at the apex of the nucellus and the latter tissue is prolonged as a tent-pole which engages with the micropyle. In the absence of data as to the course of the vascular bundles in the chalazal region it would not be possible to distinguish between this genus and _Cardiocarpus_.

=Rhabdocarpus= Berger and =Rhabdospermum= gen. nov.

It is proposed to restrict the name _Rhabdocarpus_[947] to impressions and casts of seeds of the type represented by _R. tunicatus_ as figured by Berger[948] and reproduced in fig. 506, K, the term _Rhabdospermum_ being applied to seeds of similar form in which are shown certain distinguishing anatomical features. _Rhabdocarpus tunicatus_ Berger is a species founded on a specimen from the Coal Measures of Silesia characterised by an outer carbonised testa prolonged apically as a blunt snout and, as seen in fig. 506, K, covering an apparently ribbed nucule, but the ‘ribs’ are due to the presence of fibrous strands and are not ribs in the ordinary sense. The outer tissue shows numerous longitudinal striations due presumably to the presence of fibrous elements in the sarcotesta like those shown in the petrified seed represented in fig. 501, C. The genus is defined by Berger as follows: ‘Semina ovata vel elliptico-oblonga secundum longitudinem parallele nervosa vel tenuissime striata, putamine (interdum deficiente) instructa.’ As used by Berger and many other authors _Rhabdocarpus_ includes a miscellaneous collection of seeds often differing widely from the type-species. Many of the examples correctly referred to Berger’s genus are platyspermic though a bilateral symmetry is by no means always clear. Renault and Zeiller[949] in their definition of _Rhabdocarpus_ include bilateral symmetry as a characteristic feature and speak of the seeds as oblong or oval with a pointed or truncate apex and a rounded base. Impressions of _Rhabdocarpus_ differ from those of _Cardiocarpus_ or _Cordaicarpus_ in their more elongate form, always longer than broad, and in the absence of a basal sinus. The seeds found attached to _Neuropteris_ pinnae and, in external features, agreeing with many specimens included in Berger’s genus, have been transferred by P. Bertrand[950] and Arber[951] to a new genus _Neurospermum_[952]. Arber[953] in his recent revision of British seeds proposes to restrict the name _Rhabdocarpus_ to platyspermic seeds having a ‘large unsymmetrical nucule enclosed in a large unsymmetrical wing or sarcotesta,’ that is to forms symmetrical in one plane. In this category he includes _Rhabdocarpus tunicatus_ Berg. (fig. 506, K) and _R. subtunicatus_[954] Grand’Eury, but it is not clear on what grounds Berger’s species is spoken of as symmetrical in only one plane: in the species _R. Lilleanus_ Arb.[955] the symmetry is hardly sufficiently well defined to rank as a generic character. In the case of the _Neuropteris_ seeds the apical snout is slightly curved, thus giving them an unsymmetrical appearance (cf. fig. 422, p. 114). The Carboniferous and Permian seeds usually referred to _Rhabdocarpus_ are transferred by Arber to his genus _Platyspermum_[956], a designation for which it is proposed to substitute Nathorst’s genus _Holcospermum_[957]. In seeds preserved as more or less flattened impressions it is practically impossible in many cases accurately to determine the symmetry: as fig. 506, A, shows, casts indistinguishable from some examples of _Platyspermum_ are radially symmetrical. Brongniart[958] extended the original definition of _Rhabdocarpus_ to include certain anatomical characters, and these have been more fully defined by Bertrand[959]. It is for seeds showing these anatomical features that the name _Rhabdospermum_ is now proposed. This course is followed on the ground that it is advisable to avoid confusion between petrified specimens and impressions which in spite of superficial resemblance may not be closely related. In some cases it is practically certain that an impression of the _Rhabdocarpus_ type is generically identical with a seed of similar form showing the anatomical structure of _Rhabdospermum_, but unless identity is established a distinct terminology is preferable. The use of the generic name _Rhabdospermum_ carries with it an implication of platyspermy, but under _Rhabdocarpus_ may be included seeds which are radiospermic and platyspermic. Some seeds agreeing with _Rhabdospermum_ are referred by Grand’Eury[960] to _Poroxylon_, and it is probable that _Rhabdospermum_ like _Cardiocarpus_ is a Cordaitean seed. On the other hand _Rhabdocarpus_ may well include species, apart from those transferred to _Neurospermum_, that belong to Pteridosperms. A species, _Rhabdocarpus Oliveri_, recently described by Kidston[961] from the Staffordshire coalfield is an example of a radiospermic seed which may be assigned to a Pteridosperm. The type-specimen is an ovate seed 4 cm. long and 2 cm. broad agreeing in form and surface-features with _Rhabdocarpus_ as already defined, but the evidence it affords of internal structure is insufficient to determine its position with regard to genera founded on anatomical characters. Other examples of _Rhabdocarpus_ are described by Lesquereux[962] and White[963] from American Coal Measures, by Grand’Eury[964] from the Loire, by Renault[965] from Autun, and by many other authors.

=Rhabdospermum.= Gen. nov.

The platyspermic seeds included in this genus agree in size and form with impressions assigned to _Rhabdocarpus_ and as regards the main features conform anatomically to _Cardiocarpus_; they were probably borne on Cordaitean plants. Fig. 501, C, represents a longitudinal section of the species _Rhabdospermum cyclocaryon_ described by Brongniart as _Rhabdocarpus cyclocaryon_: the sarcotesta is particularly well developed in the apical region; at the apex a portion of the micropyle is seen at _m_ and near the nucellus are pieces of the sclerotesta shown in black. The presence of anastomosing fibres near the surface is a characteristic feature: these, as Bertrand points out, do not form a hypodermal tissue in the strict sense as they may be separated by some of the thin-walled parenchyma of the sarcotesta from the epidermis. The sclerotesta is only partially preserved but the inner portion forms a dark line enclosing the nucellus, the superficial tissue of which is separated from the shrunken prothallus represented by the almost spherical dotted region: the remains of an archegonium are seen at _a_ (fig. 501, C) and the characteristic tent-pole apex of the prothallus is shown at _b_. While in shape and in the general plan of organisation _Rhabdospermum_ agrees with _Cardiocarpus_, the vascular system in the chalazal region constitutes a distinguishing feature. In _Rhabdospermum_ the main vascular strand passes through the sclerotesta, _e_, _e′_, fig. 501, E, before giving off two bundles which bend back (‘faisceaux récurrents’), traverse the shell, and then pass up the sarcotesta in correspondence with the feebly developed lateral keels as far as the micropyle, while in _Cardiocarpus_ (fig. 500, B) the bundles are given off before the main strand reaches the sclerotesta. Similar recurrent bundles occur also in _Mitrospermum_ (fig. 494, K)[967].

=Mitrospermum.= A. Arber.

_Mitrospermum compressum_ (Williamson). Mrs Arber[968] proposed the name _Mitrospermum_[969], suggested by the peculiar form of the seed-base, as a substitute for _Cardiocarpon_ for Williamson’s species _C. compressum_[970] from the Lower Coal Measures of Lancashire. The seed is platyspermic and there is some evidence that it split into two valves along the principal plane (the longer axis of the section, fig. 494, L). The diagrammatic and partially restored longitudinal section reproduced in fig. 494, K, shows the main features: a sarcotesta, _sa_, covers the surface of the testa as a thin layer except at the edges of the flattened sides where it forms a wing-like border; preserved as an impression the seed would be assigned to _Samaropsis_. The sclerotesta, _sc_, has a pointed apex which surrounds the lower third of the micropyle and a broad base perforated by the chalazal vascular strand. There was probably a narrow inner flesh as in _Trigonocarpus_ and recent Cycadean seeds (fig. 494, K, _cf_). The nucellus was free from the integument except at the base, as in _Trigonocarpus_ and _Stephanospermum_ (fig. 494, K, _n_): internal to the shrivelled remains of the inner flesh there was a nucellar tapetum surrounding the megaspore. Details as to the pollen-chamber are lacking though there are indications that it resembled that of some species of Cordaitean seeds. The main vascular supply passes through the sclerotesta and then forms a low cushion of short reticulate elements below the base of the nucellus from which two bundles are given off (fig. 494, K, _v_) in the principal plane. The course of the bundles which pierce the sclerotesta led Mrs Arber to remove this seed from _Cardiocarpus_, as recently defined by Bertrand[971], since in that genus the integumental bundles have their origin below the sclerotesta. In the course of the vascular bundles _Mitrospermum_ is intermediate between _Rhabdospermum_ and _Taxospermum_.

This genus is founded on a detached seed, but its resemblance to undoubted Cordaitean species favours its attribution to that group though, as Mrs Arber points out, some markedly platyspermic seeds are known to have been borne on fern-like fronds and _Mitrospermum_ may belong to some genus of Pteridosperms.

=Diplotesta.= Brongniart.

_Diplotesta Grand’Euryana_ Brongniart. The generic name _Diplotesta_, suggested by Grand’Eury, was given by Brongniart[972] to a Grand’ Croix seed which he compared with that of the Conifer _Cephalotaxus_. The type-specimen is elliptical and platyspermic (fig. 495, H, p. 322), and differs from _Taxospermum_ in the cordate form of the seed-cavity, also in the more restricted union of nucellus and testa. The testa is differentiated into a sarcotesta (_sa_) and sclerotesta, and the latter forms two feebly developed keels in the principal plane: a characteristic feature shared with _Mitrospermum_ is the splitting of the shell into two equal valves (fig. 495, F). Fig. 495, H, shows the contracted cylindrical nucellus and the pollen-chamber: the sarcotesta (_sa_) is only partially preserved. _Diplotesta_ differs from _Cardiocarpus_ in the course of the integumental vascular bundles which are of the type illustrated by _Rhabdospermum_, _Taxospermum_, and _Cyclocarpus_, but from these genera it is distinguished by the dehiscence of the shell, also by its form and certain anatomical features as described by Brongniart and Bertrand[973].

In this genus Bertrand includes Brongniart’s species _Sarcotaxus avellana_ (fig. 495, F), a correlation suggested by the latter author.

=Leptocaryon.= Brongniart.

Brongniart[974] founded this genus for a single species, _Leptocaryon avellana_, represented by a Grand’ Croix specimen 12 × 10 mm., which he believed to be related to _Taxus_. _Leptocaryon_ differs from _Taxospermum_ in the structure of the testa, but resembles it in external features. Bertrand[975] in his revision and extension of Brongniart’s account says that the sections throw no light on the nature of the vascular supply, and it is therefore impossible to form a satisfactory opinion as to the relationships of the genus. Renault[976] referred this genus to the Cordaitales, but we have no evidence as to the parent-plant.

=Taxospermum.= Brongniart.

Brongniart[977] gave this name to a small elliptical seed, 15×9 mm., recalling in external features the seed of _Taxus_, a genus to which he believed the Grand’ Croix species to be related. The type-species _Taxospermum Grüneri_ (fig. 495, G) has a comparatively thin testa characterised by the absence of a sub-chalazal pad formed by the swelling of the sclerotesta. The nucellus is attached by a broad base to the testa and the two regions appear to be connected for a short distance on the flanks[978]; in this feature the seed is comparable with that of the Conifer _Torreya_ and differs from the other platysperms, _Diplotesta_, _Rhabdospermum_, etc. Bertrand[979] states that the main vascular strand extends from the hilum to the chalaza before giving off the two opposite bundles which ‘follow the floor of the shell-cavity, and on reaching the flanks traverse the shell obliquely from below upwards[980].’ In the course and position of the integumental bundles _Taxospermum_ differs from _Cardiocarpus_, _Cycadinocarpus_, and _Rhabdospermum_. In this genus Bertrand includes _Sarcotaxus angulosus_ Brongn. and _S. olivaeformis_ Brongn.[981]

=Compsotesta.= Bertrand ex Brongniart +MS.+

_Compsotesta Brongniarti_ Bertrand. The generic name _Compsotesta_[982], though adopted by Brongniart for some incomplete specimens from Grand’ Croix, was not published either by him or Renault: it has recently been revived by Bertrand[983] in his account of the anatomical details of Brongniart’s sections. This seed appears to be closely allied to the polypterous forms _Ptychotesta_ and _Hexapterospermum_: the testa consists of a sarcotesta differentiated into two zones the outer of which contains vascular bundles in correspondence with the ribs, and an inner shell. There is a nucellar vascular supply and the nucellus is free on the flanks as in _Stephanospermum_ and _Trigonocarpus_.

=Samaropsis.= Goeppert.

The characters of this Permian and Carboniferous genus have already been described: the name has reference only to superficial features especially the samara-like ‘wing,’ and connotes no special anatomical features.

_Samaropsis fluitans_ (Dawson). The species described by Dawson[984] as _Cardiocarpum fluitans_ from Carboniferous strata in Nova Scotia, is represented by oval seeds with a fairly broad border usually showing an apical notch. Fig. 502, A, is a copy of Dawson’s figure: the apparent absence of an apical sinus in the ‘wing’ is probably due either to an error in interpretation or to some imperfection in the specimen. As fig. 502, A, shows, the type-specimens are far from satisfactory, and it may be that they are not specifically identical with the more complete specimens from European strata referred to Dawson’s species. Zeiller[985] points out that seeds of this species vary considerably in size, but there is always in well-preserved examples a bifid beak at the apex. Seeds of similar form though not specifically identical are described from the Coal Measures of Missouri as _Cardiocarpon (Samaropsis) Branneri_ Fairch. and White[986]. Good examples of _S. fluitans_ are figured by Weiss[987] from the Coal Measures of Saarbrücken and the species is widely distributed in Upper Carboniferous beds generally.

_Samaropsis bicaudata_ Kidston. This species (fig. 502, E) originally described[988] from Lower Carboniferous rocks in Scotland as _Cardiocarpus bicaudata_ and subsequently assigned to _Samaropsis_, is characterised by a greater development of the flat wing-like border which is divided into two long tapering basal lobes. Seeds of similar form are figured by Lesquereux[989] from Pennsylvania as _Cardiocarpus_ (_Ptilocarpus_) _bicornutus_.

_Samaropsis_ (_Samarospermum_) _moravica_ (Helmhacher). This type[990] (fig. 502, H) is characterised by the great length of the wing-like border and on that account it was transferred by Arber to a new genus. It was originally described by Helmhacher from the Permian of Moravia as _Jordania moravica_ and the type-specimen has been refigured by Zeiller[991] who records the species from Upper Carboniferous and Permian rocks in France: it is recorded also from several localities in Germany[992]. Seeds figured by Potonié[993] from the Permian of Thuringia as _Samaropsis Crampii_ (Hartt) are undoubtedly examples of _S. moravica_: the true _S. Crampii_ has recently been well illustrated by Dr Stopes[994] from the Westphalian of New Brunswick. The species is recorded by Arber[995] from the Kent coalfield.

_Samaropsis emarginata_ (Goeppert and Barger).

The seed represented in fig. 502, B, from the Lower Coal Measures of Kilmarnock, Scotland[996], affords a good example of the genus: the species was originally described by Berger as _Cardiocarpon emarginatum_ and it was on this type that Fiedler founded the genus _Cyclocarpon_[997]. It has been referred by many authors to _Cardiocarpon_ and might be regarded as a type intermediate between _Cordaicarpus_, as used in this chapter, and _Samaropsis_, though the breadth of the border is more in keeping with the latter designation. The figured specimen is 1·6 cm. long and 1·4 cm. broad; the nucule has a slightly cordate base and shows several faint converging ribs which are too inconspicuous to be represented in a natural-size drawing. A narrow median groove in the apical region shows the position of a vascular strand. The species is recorded from several countries: similar though specifically distinct seeds, described by Dawson as _Cardiocarpon cornutum_, have recently been re-described by Dr Stopes[998] from the Westphalian of New Brunswick where they occur in association with the leaves of _Cordaites Robbii_ Daws.

[Illustration: Fig. 502. A, _Samaropsis fluitans_. B, _Samaropsis emarginata_. C, D, _Cordaicarpus Cordai_. E, _Samaropsis bicaudata_. F, _Samaropsis Seixasi_. G, _Samaropsis barcellosa_. H, _Samaropsis moravica_. I, _Samaropsis bignonioides_. K, _Samaropsis Newberryi_. (A, after Dawson; B, R. K. 1576; C, R. K. 1899; D, R. K. 4647; E, after Kidston; F, G, after White; H, after Zeiller; I, after Fiedler; K, R. K. 2313. All nat. size.)]

_Samaropsis Newberryi_ Andrews.

This species was originally described by Andrews[999] from the Coal Measures of Ohio: the specimen shown in fig. 502, K, was sent to Dr Kidston by Mr Claypole. The whole seed is 5 cm. wide and 3·5 cm. in depth; it is characterised by a short and relatively broad nucule surrounded by a very broad and flat border showing faintly marked radially disposed lines and in places some irregularly distributed pits. The apex is emarginate and there is a broad and deep sinus in the sarcotesta in the chalazal region. The seed resembles _Samaropsis alata_ Kidst.[1000] and _S. Baileyi_ (Daws.)[1001] but it differs from these in the greater breadth of the ‘wing’ and in the form of the nucule.

The seed described by Fiedler[1002] as _Jordania bignonioides_ (fig. 502, I) agrees closely with _S. alata_ Kidst. but is probably specifically distinct.

_Samaropsis barcellosa_ (White).

White[1003] described this species (fig. 502, G) from Permo-Carboniferous rocks in Brazil (Rio Grande do Sul) as _Cardiocarpon barcellosum_. The nucule is said to be cordate but, as seen in the figure, there is no clear indication of a basal sinus: the presence of a relatively broad ‘wing,’ as White says, entitles the seed ‘to a place in the _Samaropsis_ section of the genus.’ The author of the species compares it with seeds described from Westphalian rocks in Ohio[1004] and Pennsylvania[1005]. It is interesting to find a type which is common in both Europe and North America in the western portion of Gondwana Land. Seeds of similar form are recorded also from India, South Africa, and Australia[1006].

_Samaropsis Seixasi_ (White).

This Brazilian species from the same beds is described by White as _Gangamopteris_ (_Samaropsis_) _Seixasi_[1007]: it is characterised by a small ovate nucule 8–10 mm. long and 5 mm. broad, in some specimens surrounded by a complete ‘wing’ extending above the apex and below the base, giving the seed an appearance similar to that of _Samaropsis_ (_Samarospermum_) _moravica_ while sometimes, as in the example shown in fig. 502, F, the broad border is preserved only at the sides. These seeds are abundant in the Santa Catharina beds, where they were discovered by Dr Esdras do Prado Seixas, in association with leaves of _Gangamopteris_, and White thinks that they were borne on the fertile leaves of that genus which he has named _Arberia_[1008]. Although there is as yet no proof of a connexion between _Gangamopteris_ and seeds of this or any other type it is almost certain that it was a seed-producing plant.

_Samaropsis Leslii_ sp. nov.

[Illustration: Fig. 503. _Samaropsis Leslii_. From Vereeniging, S. Africa. (Mr Leslie’s Collection; nat. size.)]

The seed on which this species is founded was discovered by Mr T. N. Leslie in the Ecca beds (Permo-Carboniferous) of Vereeniging, South Africa, a locality from which the same geologist has obtained leaves of _Cordaites_, _Psygmophyllum_, _Glossopteris_ and other genera[1009]. In the slightly cordate base and tapered apex (fig. 503) the nucule agrees closely with those of European examples, but the Vereeniging type is distinguished by its larger dimensions and by the wider border indicating a thick sarcotesta continued basally into a stout stalk. The apex is emarginate and a median rib marks the position of a vascular strand. There is no evidence as to the nature of the parent-plant.

_Samaropsis indica_ (Zeiller).

Prof. Zeiller[1010] described this species as _Cardiocarpus indicus_ from the Karharbari (Lower Gondwana) beds of India. An examination of the type-specimens enables me to confirm the accuracy of the original account. The platyspermic seed is 5·5 cm. long and 4·5 cm. broad; a cordate nucule is enclosed by a flat border similar to that of _Samaropsis Leslii_ but narrower especially on the sides of the nucule. At the apex there is a deep sinus extending to the nucule, and at the base a fairly broad band of carbonaceous matter shows the position of the chalazal vascular strand. The seed is characterised by its large size and by its almost orbicular form: it occurs as a detached specimen in beds containing _Cordaites_ (_Noeggerathiopsis_) and _Glossopteris_.

A species from Arkansas described by Lesquereux[1011] as _Cardiocarpus ingens_ [= _Cordaicarpus ingens_ (Lesq.)] affords another example of a large seed similar to _C. indicus_ and, except in its more orbicular form, to _Samaropsis Leslii_.

_Samaropsis Milleri_ (Feistmantel).

[Illustration: Fig. 504. _Samaropsis Milleri._ (Indian Geological Survey, Calcutta; nat. size.)]

An examination of the type-specimen from the Calcutta Museum enables me to amplify the original description in an important particular. The species was found in Lower Gondwana (Katharbari) beds in India and referred by Feistmantel[1012] to the genus _Carpolithes_. Arber[1013], who tentatively employed the generic name _Cardiocarpus_ in place of _Carpolithes_, suggests that the seeds may be radiospermic, as Feistmantel’s drawings show only a very narrow border to the nucule. The specimen reproduced in fig. 504 was figured by Feistmantel without any indication of a definite sarcotesta or wing, but as seen in the drawing the ovate cordate sclerotesta is surrounded on one side and at the base by an outer envelope: this is clearly seen at the apex where it shows a rounded termination sloping downwards towards the micropyle precisely as in _Samaropsis indica_ (Zeill.). The border is narrow at the sides and broader at the base as in _S. Leslii_. The seed is 4·5 cm. long and 2·3 cm. broad, differing from _S. indica_ in its rather smaller size and in the slightly narrower nucule. Though there is no decisive evidence as to the parent-plant the occurrence of a specimen of this species partially covered by a scale-leaf of a type[1014] very similar to that which is generally recognised as belonging to _Glossopteris_ suggests the possibility that the seeds may belong to that genus. Scale-leaves of _Glossopteris_ are described in Volume +ii.+, but it may be added here that leaves similar in form to those from India, Australia, and elsewhere are figured by Geinitz[1015] from the Altai Mountains as _Trigonocarpus_? _actaeonelloides_: the specimens are represented with the basal scar at the apex.

=Cordaicarpus.= Geinitz.

In view of the fact that the generic names _Cardiocarpus_, _Cordaicarpus_, and other designations have been applied to casts and impressions which cannot be distinguished by any constant or important feature it is proposed to adopt the name _Cordaicarpus_ for platyspermic seeds, preserved as casts or impressions, having a comparatively narrow border enclosing an ovate or cordate-ovate nucule; the base is either rounded or cordate. The choice between _Cordaicarpus_ and _Samaropsis_ depends on the breadth of the border. _Cordaicarpus_, though more suggestive of a Cordaitean alliance, may in some cases be a Pteridosperm seed.

_Cordaicarpus Cordai_ (Geinitz). Lenticular seeds more or less orbicular or broadly ovate (fig. 502, C, D), often slightly cordate at the base of the nucule and with a broadly acute apex: the border is narrow or sometimes hardly represented as in the seeds described by Berger as _Rhabdocarpus ovoides_, a species similar to but more oval than _Cordaicarpus Cordai_. The latter species[1016] occurs in several coalfields in Britain, France, Germany, and elsewhere. Fig. 502, C, shows a good example from the Middle Coal Measures of Yorkshire, 9 mm. long by 8 mm. broad; on the very slightly cordate nucule are several faint ribs converging towards the base and apex and between them fine striations, characters too indistinct to be reproduced in the natural-size drawing. The flat border represents the sclerotesta. The seed shown in fig. 502, D, from the Westphalian series of Warwickshire belongs to the same species or is a closely allied type: the faint suggestion of reticulation on its surface might be regarded as a reason for referring it to _C. areolatus_ Boul.[1017], a form characterised by a reticulation on the testa, described by Zeiller[1018] and other authors. This reticulation is, however, in some cases at least, formed by crumpling and splitting of the superficial carbonised film into more or less regular meshes: the figured specimen occurs with several other seeds of the same type, most of which have a smooth surface. Dr Kidston tells me that a recent critical examination of seeds in his collection leads him to regard some specimens (_e.g._ fig. 502, D) previously referred by him to _C. Cordai_ as identical with _Carpolithes membranaceus_ Goepp.[1019]

[Illustration: Fig. 505. _Cordaicarpus Cordai._ Cuticle of the testa. (Kidston Collection.)]

The photograph reproduced in fig. 505 shows the result of chemically treating the carbonised cuticle of a seed of _Cordaicarpus Cordai_, a method little used as yet in the case of Palaeozoic plants but which may be useful in distinguishing seeds which cannot be satisfactorily separated by microscopic features. The superficial cells have very thick walls and present an appearance similar to that of the sclerous cells in the testas of some petrified specimens.

The larger seed shown in fig. 501, D, from Commentry, described by Renault and Zeiller[1020] as _Cardiocarpus sclerotesta_, is another example of _Cordaicarpus_ as now defined.

IV. +Miscellaneous Seeds.+

In this section are included genera based on characters of comparatively little morphological importance; their claim to treatment under a common heading is that we know little or nothing of their anatomical features or of the parent-plants.

i. _Ribbed seeds._

The seeds represented by the following genera possess ribs or flanges usually in multiples of three; the transverse section may be triangular, hexagonal, or polygonal. The symmetry is generally radial, but it is difficult to determine whether a slight departure from the radiospermic form is an original feature or the result of pressure. Some of the ribbed seeds with which we are now concerned are no doubt closely allied to _Trigonocarpus_ and should be included in the Trigonocarpales, but others may well be distinct forms.

Genera: _Hexagonocarpus_, _Decagonocarpus_, _Polypterocarpus_, _Rhynchogonium_, _Boroviczia_, _Diplopterotesta_, _Musocarpus_, _Holcospermum_.

=Hexagonocarpus.= Renault.

This generic name is adopted by Renault[1021] for casts from Commentry which, while probably identical with Brongniart’s _Hexapterospermum_, afford no confirmatory anatomical evidence. The species _Hexagonocarpus crassus_[1022] is represented by casts or nucules characterised by six well-defined ribs and grooves with a length of nearly 3 cm. In this genus may be included the cast from the Coal Measures of Lancashire described by Williamson[1023] as _Hexapterospermum Noeggerathi_ (fig. 506, H), and another British example is afforded by _Hexagonocarpus Hookeri_ Kidst.[1024], a rare fossil in the South Staffordshire coalfield. It is impossible confidently to assign these ribbed casts to genera founded on petrified specimens as they almost certainly belong to different types, but the employment of the name _Hexagonocarpus_ may conveniently be used for casts or impressions of seeds with six longitudinal ribs differing in their relatively broader and less prominent form from the wing-like flanges of _Polypterocarpus_ (cf. fig. 496, B). The seeds described by Dr P. Bertrand[1025] as _Hexapterospermum modestae_, and believed by him to belong to a plant with _Neuropteris_ fronds, should be included in _Hexagonocarpus_ as we have no definite knowledge of their anatomical features.

=Decagonocarpus.= Renault.

This name, proposed by Renault[1026], is applied to seeds without petrified tissues characterised by ten ribs; an example is afforded by _Decagonocarpus olivaeformis_ from the Commentry coalfield, an elliptical seed which bears a close resemblance to _Holcospermum sulcatum_ (fig. 506, A) except in the smaller number of ribs.

=Polypterocarpus.= Grand’Eury.

Grand’Eury[1027] adopted this generic name for seeds from St Étienne characterised by the presence of three, six, or more deep wings or flanges. The term _Pterospermum_ has recently been proposed by Arber[1028] for a seed from the Coal Measures of Staffordshire, which he names _P. anglicum_: the type-species of the genus has three deep wings, one from each angle. For the same seed Kidston[1029] proposed the name _Tripterospermum ellipticum_, but as Arber’s account was published first his specific name has priority. _Pterospermum_ had, however, been previously used for a genus of Sterculiaceae, and partly on this account but mainly because Grand’Eury’s genus _Polypterocarpus_ is available the latter designation is adopted. In _Polypterocarpus anglicus_ (fig. 496, B) the flanges project slightly beyond the apex of the seed and there is a small notch at the base; the nucule is 5 cm. long and 1 cm. broad. While it is not improbable that this seed is generically identical with Brongniart’s _Tripterospermum_[1030], it is safer, in the absence of structural details, to employ the less committal term. There is no information with regard to the nature of the parent-plants of species of _Polypterocarpus_. The English seed from the Middle Coal Measures of Derbyshire and the Staffordshire coalfield described by Arber[1031] as _Radiospermum ornatum_ and by Kidston[1032] as _Polypterospermum ornatum_ affords another example of _Polypterocarpus_ as the generic name is here employed.

=Rhynchogonium.= Heer.

Heer[1033] proposed this generic name for some globose, ovate, or oblong ‘fruits’ from Lower Carboniferous strata in Spitzbergen, including also fragments of ‘leaves’ which without satisfactory evidence he believed to belong to the same plant. The supposed fruits are clearly seeds, and Nathorst regards the ‘leaves’ as portions of a Fern rachis. Heer described four species, but these have since been reduced to two, and indeed it is probable that only one type, _Rhynchogonium costatum_, is represented. Nathorst[1034] compares Heer’s seeds with a Lower Carboniferous species described by Young[1035] as _Trigonocarpum gloagianum_, the resemblance of which to the Spitzbergen seeds was pointed out by Kidston. A seed of _Rhynchogonium costatum_ is about the size of a hazel-nut and may reach a length of 21 mm.; it is ovate, with a broad rounded base, and in the upper third is tapered and conical, the sides of the characteristic snout being distinguished from the smooth surface of the rest of the seed by the presence of eight ribs converging towards the apex (506, G). Zalessky[1036] recorded closely allied seeds from Lower Carboniferous beds in Northern Russia, assigning them to a new genus _Boroviczia_, the type-species being _B. Karpinskii_; he adduced arguments in favour of Heer’s interpretation of the fossils as fruits but, according to Nathorst, this view has been abandoned. In his recent memoir on the Culm flora of Spitzbergen Nathorst[1037] discusses the morphological nature of _Rhynchogonium_ seeds and describes additional specimens. Without the aid of petrified examples it is hardly possible to determine the true nature of the fossils.

[Illustration: Fig. 506. A, _Holcospermum sulcatum_. B, C, _Codonospermum anomalum_. D, _Diplopterotesta spitzbergensis_ (× 3). E, _Gnetopsis elliptica_ (cupule). F, _Thysanotesta sagittula_. G, _Rhynchogonium costatum_. H, _Hexagonocarpus Noeggerathi_. I, _Boroviczia Karpinskii_. K, _Rhabdospermum tunicatum_. (A, Kidston Collection; B, after Grand’Eury; C, after Renault and Zeiller; D, F, G, after Nathorst; E, after Renault; H, after Williamson; I, after Zalessky; K, after Berger.)]

Some specimens of _Rhynchogonium sulcatum_ in Dr Kidston’s collection show the original surface-features: the carbonised integument is divided in the upper region into linear lobes separated from one another at their origin by fairly wide sinuses, a type of integument suggesting comparison with _Physostoma_. This species was originally described by Lindley and Hutton as _Carpolithes sulcata[1038]_ from Lower Carboniferous rocks at Newhaven in Scotland and has recently been figured by Zalessky[1039] from specimens in the Kidston collection from the Lower Calciferous sandstone of Midlothian. Nothing definite can be said as to the parent-plant, but it is significant that in Midlothian _Rhynchogonium sulcatum_ occurs in a bed full of isolated pinnules of a _Cardiopteris_ almost to the exclusion of any other fossils[1040].

The seed figured by Nathorst[1041] from the Culm of Spitzbergen as _Lagenospermum? glandiforme_ agrees closely with _Rhynchogonium_ and _Boroviczia_, and it is impossible to decide whether the lobed appearance is due to the presence of a cupule or to the divisions of an integument.

=Boroviczia.= Zalessky.

The specimens from Lower Carboniferous beds in Russia[1042] on which this genus was founded are perhaps unnecessarily separated from _Rhynchogonium_; they are ovate and beaked, 10 mm. long with a maximum breadth of 6 mm. The type-species, _Boroviczia Karpinskii_ (fig. 506, I), is represented both by specimens in which the cast is complete and by others in which the apical snout of the integument is split into separate lobes identical in form and apparently in number with those of _Rhynchogonium_. Nathorst[1043], who retains the generic name and describes two additional species from Spitzbergen, points out that in Boroviczia the tapered apex is more sharply differentiated from the broader basal portion, a difference hardly worthy of generic recognition. If _Boroviczia_ is retained as a genus distinct from _Rhynchogonium_ it should include the seeds described by Young as _Trigonocarpum gloagianum_.

=Diplopterotesta.= Nathorst.

_Diplopterotesta spitzbergensis_ (Heer). Heer[1044] included in _Samaropsis_ some seeds, described as _Samaropsis spitzbergensis_, from Lower Carboniferous strata in Spitzbergen which differ considerably from typical examples of the genus. These have recently been made the type of a new genus _Diplopterotesta_[1045]: they are platyspermic seeds nearly twice as long as broad, 6–9 × 3·5–5 mm., characterised by a thin sclerotesta expanded throughout the length of the seed into two prominent transversely striated wing-like ribs or flanges (fig. 506, D). At the apex, which is broad in contrast to the pointed basal end of the seed, the sclerotesta forms a crown of eight lobes about the flat summit of the seed-body. As Nathorst points out, this species closely resembles the Carboniferous seeds figured by Grand’Eury[1046] as _Polypterocarpus_, but in the latter type there are more than two wings. Nathorst’s figures, one of which is reproduced in fig. 506, D, suggest a difficulty in determining the number of the flanges, which would seem to be more than two, but this appearance is regarded by Nathorst as misleading and he believes that except at the apex there are only two wings.

=Musocarpus.= Brongniart.

Brongniart[1047] proposed this name for two species, _M. prismaticus_ and _M. difformis_, from French Coal Measures, but gave no description of the specimens. The type-species, described from the Loire by Grand’Eury as _Musocarpus prismaticus_[1048], is an ovate seed nearly 3 cm. long with six longitudinal ribs, three being more prominent than the others, and characterised by a ribbed prolongation of the base of uniform diameter spoken of as a carpophore, which was apparently articulated to the lower part of the seed-proper from which it was easily detached by a natural absciss-layer. Nothing is known of the structure or of the affinity of the genus. Some specimens in Dr Kidston’s collection from Westphalian beds in Lanarkshire are closely allied to or perhaps identical with _M. prismaticus_. In the presence of a distinct basal region _Musocarpus_ resembles _Codonospermum_[1049], though without any knowledge of the anatomy of the former type it is impossible to say whether or not this resemblance has any morphological importance.

=Holcospermum.= Nathorst.

_Holcospermum sulcatum_ (Sternberg). The cast reproduced in fig. 506, A, affords a good example of a form of seed recorded under several generic names and not uncommon in Upper Palaeozoic rocks in Europe and North America, which in most cases cannot be assigned to a genus implying the possession of certain anatomical characters. This type was figured by Sternberg[1050] from the Coal Measures of Radnitz as _Carpolites sulcatus_. Some ‘fruits’ collected on the beach near Newhaven, Midlothian, from the Calciferous Sandstone series, are figured by Lindley and Hutton[1051] as _Carpolithes sulcata_, but as already stated these have been transferred[1052] to the genus _Rhynchogonium_. Several authors have referred specimens of the type now included in _Holcospermum_ to _Rhabdocarpus_, but that genus, as stated on a previous page, is restricted to seeds agreeing in form with _R. tunicatus_. Renault[1053] instituted the genus _Colpospermum_ for a partially petrified seed from Commentry agreeing externally with _Carpolites sulcatus_ Sternb. and regarded by him as specifically identical, characterised by longitudinal ribs which represent folds of the testa, the intervening grooves being occupied by an irregular reticulum formed by occasionally anastomosing smaller ribs. The generic name _Colpospermum_ should therefore be reserved for ribbed seeds showing the anatomical features described by Renault and Zeiller: its application to Sternberg’s species is inadvisable on the ground that we have no information with regard to the morphological nature of the ribbing. More recently Arber[1054] has proposed the name _Platyspermum_, a name previously applied to a Cruciferous plant, for Stephanian and Permian seeds formerly assigned to Berger’s genus _Rhabdocarpus_, which are symmetrical in two planes. In this genus he includes _Platyspermum sulcatum_ and among other species _P. Kidstoni_ founded on a seed originally identified by Kidston[1055] as _Rhabdocarpus multistriatus_ Sternb. which, though probably a distinct species, is of the same general type as _C. sulcatus_ Sternb.

The cast represented in fig. 506, A, from the Middle Coal Measures of Yorkshire is 3·5 cm. long and has 18 regular longitudinal ribs: it is radially symmetrical and does not conform in this respect to Arber’s definition of _Platyspermum_. In many cases, _e.g._ the flattened seed figured by Kidston as _Rhabdocarpus multistriatus_, it is impossible to determine the symmetry of the seed. The old generic name _Carpolites_, _Carpolithes_ or _Carpolithus_ has been used in a very wide sense and does not connote any well-defined features; it should be reserved, in the form _Carpolithus_ as used by Linnaeus, for seeds that cannot be assigned to a systematic position or which do not exhibit any distinctive characters worthy of emphasis by the institution of a special name. The generic term _Holcospermum_, recently proposed by Nathorst[1056], is a suitable name for this type of seed; as defined by the author, it includes seeds that may be radiospermic or platyspermic. Specimens of the type-species, _H. dubium_[1057],from the Lower Carboniferous beds of Spitzbergen, agree closely in size and shape with some examples of _Rhynchogonium_ and _Boroviczia_, but they are distinguished by prominent ribs extending the whole length of the cast. In some seeds similar to _H. sulcatum_ the ribs form sharp ridges, but the difference between rounded and sharp ridges is often determined by the method of preservation: a specimen deprived of its outer flesh would form a cast more sharply ribbed than a seed in which the sarcotesta had been moulded on to the ribbed sclerotesta. On the other hand some ribs are formed by blunt sclerotestal folds as in _Colpospermum_: both types of ribbing are included in the genus _Holcospermum_.

ii. _Other Genera._

=Malacotesta.= Williamson.

This generic name was instituted by Williamson[1058] for a petrified seed from the Coal Measures of Lancashire which he named _Malacotesta oblonga_: the type-specimen is a small seed 6–7 mm. long characterised by a thick fleshy integument. The structure is however insufficiently known to admit of a satisfactory comparison of the imperfectly preserved specimen figured by Williamson with other types.

=Thysanotesta.= Nathorst.

Nathorst[1059] founded this genus on a single seed from the Lower Carboniferous plant-beds of Spitzbergen which exhibits features sufficiently distinctive to justify its recognition as the type of a separate genus.

_Thysanotesta sagittula_ Nathorst. The type-specimen is a long and narrow seed, 20 mm. long, ovate below and prolonged apically into a slender beak, 8 mm. in length, which probably represents a micropylar tube; the beak bears numerous stiff hairs (fig. 506, F). The seed closely resembles a carpel of _Erodium_ without the horizontal part of the awn. There is no evidence as to the nature of the parent-plant but the species affords an interesting example of a Palaeozoic seed apparently adapted for wind-dispersal, or possibly the bristles may have served the same purpose as in the fruits of _Erodium_. Attention is called elsewhere[1060] to the frequent resemblance of Palaeozoic seeds such as _Thysanotesta_ to fruits of recent flowering plants.

=Carpolithus.= Linnaeus.

Nathorst[1061] has pointed out that this generic name was employed by Linnaeus for fossil fruits: in the form _Carpolites_ (Sternberg) or _Carpolithes_[1062] it has been widely used and differently defined by authors, usually in a comprehensive sense including fossil seeds from both Palaeozoic and Mesozoic strata which cannot be assigned to a definite position in the plant-kingdom. It is desirable to adopt some designation for seeds from strata of different ages which do not exhibit any features sufficiently distinctive to justify the creation of a special genus. By employing such a name as _Carpolithus_, without attempting to define its characters within prescribed limits, for casts or impressions of seeds which are not distinguished by any striking characters and cannot be allocated to any particular section of seed-bearing plants the unnecessary multiplication of generic titles is avoided: when any additional data are obtained differentiating particular types from other forms of _Carpolithus_ the provisional term should be superseded by some distinctive generic name. Among seeds from the Coal Measures there are several examples of both large and small types without any regular ribs or lacking such features as serve to distinguish the genera already described, which are appropriately included in _Carpolithus_. The two species _Carpolithus Wildii_ Kidst.[1063] and _C. bivalvis_ Goepp.[1064] are two examples of seeds from Upper Carboniferous rocks which it is desirable to refer to this comprehensive genus pending further discoveries as to their morphological features. The Jurassic species _C. conicus_ Lind. and Hutt.[1065] is another type which it has been customary to include in _Carpolithus_.

(_Microspermum._ Arber.)

The generic name _Microspermum_ has recently been proposed by Arber[1066] for some Westphalian seeds described by Carpentier[1067] from the North of France as _Carpolithes? samaroides_ and for similar fossils from the Middle Coal Measures of Nottingham. The supposed seeds are small, ovate or pyriform bodies, 5–12 mm. long and 2·5–6 mm. broad, with one extremity broadly rounded and the other acute: one side is keeled, the other grooved, and a large foramen occurs near the broader end and on the grooved surface. Arber emphasises the fact that the specimens are symmetrical in one plane only. As the author of the genus points out the morphological nature of the fossils is not thoroughly established, nor is the parent-plant known. In view of the ill-defined characters of the specimens so far discovered it is hardly desirable to institute a new generic name implying their seed-nature; moreover _Microspermum_ has previously been employed for a genus of Compositae. An examination of specimens leads me to doubt their seed-nature and it would seem more likely that they are foliar organs, possibly bracts which originally bore seeds or sporangia.

* * * * *

An examination of some of the specimens of supposed Permian seeds for which Geinitz[1068] proposed the generic name _Guilelmites_ convinced me that they are inorganic structures, probably nodules in shale smoothed and rounded by slickensiding.