A third example from the same locality (fig. 557) is described by Feistmantel[1241] as part of one of Williamson’s ‘carpellary discs[1242],’ a comparison that is fully justified. The accompanying drawing has been carefully made from the actual specimen: portions of 10 very hairy bracts radiate in a horizontal plane from a continuous lamina with a wrinkled and ridged surface bent sharply back at right-angles to the bracts and forming a double curve as seen in the sectional view (fig. 557, C). The form assumed by the vertical part of the disc is, I believe, the result of compression. Wieland[1243] regards this fossil as a whorl of microsporophylls originally attached to the lower portion of the receptacle of a bisexual flower. Close to the edge of one of the bracts is an imperfectly preserved structure (fig. 557, B, s) which may represent two alternately arranged rows of synangia belonging to one of the hairy bracts; but we have no evidence as to the position of the microsporophylls on the flower-axis. The central space enclosed by the crushed concrescent portion of the disc is large enough to have embraced a receptacle but, on the other hand, the portion preserved may have broken off from a proximal cup like that of W. spectabilis[1244], which, as Nathorst’s specimens show, is sometimes broken across near the upper edge of the basal funnel. This specimen is spoken of by Feistmantel as Williamsonia gigas. It is impossible to say whether these Rajmahal specimens belong to one species, and they are therefore provisionally designated Williamsonia sp. and Williamsonia sp. cf. W. setosa.
This species is based on a compressed ovate strobilus surrounded by linear bracts and a portion of the cylindrical axis[1245]. It is possible that this smaller, bud-like, specimen may be a younger example of the species referred by Feistmantel to Williamsonia gigas.
Founded on a small strobilus enclosed by linear bracts, from the Jurassic rocks of Cutch[1246], very similar to Williamsonia pecten; as seen in fig. 558, drawn from the original specimen, the flower is associated with a Ptilophyllum frond indistinguishable from some examples of Ptilophyllum pecten.
This name is proposed for some imperfect specimens described by Feistmantel from the Godaveri district and named by him Williamsonia sp. cf. Williamsonia gigas[1247]. They differ from Williamsonia gigas in the larger size of the bracts which reach a length of 13 cm. and may be compared with those of a large specimen recorded from Mexico as Williamsonia Cuauhtemoc[1248].
It is almost certain that some at least of the Indian flowers were borne on stems with the foliage known as Ptilophyllum acutifolium, an inference based not only on the almost constant association of flowers and fronds but also on the juxtaposition of both kinds of organs with stems precisely similar to those described from England and Mexico. Though none of the specimens are sufficiently well preserved to afford much information as to structural features, Miss Bancroft[1249] has shown that the bracts of one of the examples assigned by Feistmantel to W. gigas are similar anatomically to those of Williamsonia scotica and are clothed with simple hairs. The important point is the very close correspondence between the Indian and English types of Williamsonia, as regards flowers, fronds, and stems.
This species was founded on several specimens from Wealden beds on the Sussex coast none of which afford information as to anatomical structure[1250]. It is not improbable that more than one species is represented. The ovulate cone, 6 cm. long, is surrounded by several linear bracts (fig. 559, A) and in shape resembles Bennettites Morierei Lign.; the bracts are broken across near the base, as is frequently the case in Williamsonia gigas, exposing an annular zone formed by persistent interseminal scales. From the inner face of some of the bracts project slender radiating plates (fig. 559, A′) which no doubt mark the boundary of the superficial and relatively large interseminal scales, like those forming the so-called pericarp in Bennettites Gibsonianus. The receptacle appears to have been conical, a feature recalling Bennettites rather than Williamsonia. The saucer-like impression shown in fig. 559, B, is practically identical with the corresponding portion of Williamsonia Leckenbyi: the centre is occupied by a raised area, the basal part of the receptacle, on which a series of peripheral prominences represents the vascular strands; the sides of the saucer show very clearly the reticulum formed by the distal ends of interseminal scales. One reason for assigning this species to Williamsonia rather than to Bennettites (or Cycadeoidea) is the occurrence in the same bed of a peduncle 12 cm. long and 3 cm. broad which probably belonged to the parent-plant of the cone. The surface of the peduncle shows spirally disposed scars of bracts crowded at the distal end and more widely separated in the lower portion.
In 1837 Buckland[1251] gave an account of a ‘unique and beautiful fossil fruit’ from Inferior Oolite beds at Charmouth in Dorsetshire and stated that the type-specimen was in the Oxford Museum. Professor Sollas kindly searched for the specimen some years ago but without success. Buckland considered that the fruit was related to the Pandanaceae and described it as follows: ‘The size of this fruit is that of a large orange, its surface is occupied by a stellated covering or epicarpium, composed of hexagonal tubercles, forming the summits of cells, which occupy the entire circumference of the fruit. Within each cell is contained a single seed, resembling a small grain of rice more or less compressed, and usually hexagonal. When the epicarpium is removed, the points of the seeds are seen, thickly studded over the surface of the fruit. The bases of the cells are separated from the receptacle by a congeries of foot-stalks formed of a dense mass of fibres, resembling the fibres beneath the base of the seeds of the modern Pandanus.’ At the suggestion of Robert Brown he called the ‘fruit’ Podocarya, the specific name Bucklandi being afterwards given by Unger[1252]. Brongniart[1253] called attention to the resemblance of Buckland’s specimen to Williamsonia, and that name has been adopted by Saporta[1254], Nathorst, and other authors[1255]. Sowerby’s drawings illustrating the original description, one of which is reproduced in fig. 560, show that this unusually fine specimen is an ovulate Bennettitean strobilus very similar in its thick conical receptacle to some of Wieland’s species of Cycadeoidea, e.g. C. dacotensis (fig. 528): the armour of scales and megasporophylls agrees exactly with that of some species of Williamsonia from Yorkshire and with the flowers of Cycadeoidea. Though included in the genus Williamsonia it would not be out of place in Cycadeoidea.
The type-specimen was found by Hugh Miller near Cromarty (N.E. Scotland) and figured as a cone of peculiar form[1256]; it was obtained from a limestone nodule probably derived from Upper Jurassic rocks. The fossil is 11 cm. long and has a maximum breadth of 6 cm. (fig. 561): numerous linear bracts cover the surface and in the lower portion many of them are broken. A noteworthy feature is the absence of any clean-cut base, a fact pointing to fracture rather than a natural abscission of the fertile axis. The following description may serve to give a general idea of the salient characters. Flowering shoot ovoid, covered with linear bracts some of which are prolonged above the conical apex as slender tapered organs and two of them bear a few short lateral appendages (fig. 561, l), probably reduced leaflets, near their distal ends. The cylindrical axis, completely hidden by bracts, 1·5 cm. in its widest part, bears in the lower or sterile region bracts and long hairs and in the upper part interseminal scales and immature megasporophylls which together form a narrow band (fig. 562, S) 2 mm. broad extending over the incompletely preserved and conical apex, as in some of the American examples of Cycadeoidea. The strobilus was probably borne at the apex of a lateral branch given off from a stem covered with persistent petiole-bases: there is no evidence that this was the case, but the appearance of the ovoid cone suggests comparison with those of Williamsonia gigas which were terminal on fairly long branches and not partially hidden among the bases of fronds as in Cycadeoidea. It is, however, possible that the cone of Williamsonia scotica is a lateral structure: this suggestion is based on the occurrence of a small branch or bud, which may be the apex of the whole fertile shoot, given off from the cone-axis but only revealed in transverse sections. The interseminal scales, 2 mm. long and 0·23 mm. broad at the truncate distal end (fig. 563), are polygonal in section and arranged as rosettes of 5–6 around each megasporophyll (fig. 564, a section tangential to the peripheral layer of scales and sporophylls). The megasporophylls, equal in length to the scales, consist of a cylindrical axis bearing a terminal megasporangium, an undifferentiated nucellus, enclosed in a single integument prolonged as a micropylar tube above the conical end of the nucellus (fig. 563, B, C, a). Fig. 562 represents a transverse section through the cone showing the cylindrical axis with its compact covering layer (fig. 563, A, s) of sterile and fertile appendages, and beyond this sections of the enveloping bracts embedded in a dense felt of long hairs. The tissue of the axis, though very imperfectly preserved, shows occasional groups of secretory sacs and a few patches of scalariform tracheids: there is evidence of the occurrence of peripheral conducting tissue in the lower portion of the axis such as occurs in the peduncles of American species of Cycadeoidea described by Wieland. The bracts nearer the axis are more shrivelled than those farther away, the result of the feebler development of hypodermal stereome in the more internal bracts. Sunken stomata occur on the lower surface of some of the bracts: several collateral bundles are present in each and large secretory ducts are abundant. The numerous hairs on the bracts and the sterile region of the cone are outgrowths of epidermal cells; most of them consist of a short basal cell and a very long thick-walled tubular hair reaching a length of several centimetres. In some cases the basal cell bears a group of short cells each of which is the starting-point of a long hair: this is worthy of notice from the point of view of comparison with the ramenta of other Bennettitalean flowers. The short proximal cell of a hair is surrounded by a cuticular ring like a rounded base-moulding where it rests on the epidermis: this has been aptly compared to the dark rings that form a striking feature of the cuticular membrane of Ptilophyllum leaflets[1257].
In the examination of the type-specimen the first section cut was transverse to the axis (fig. 562), and this happened to traverse the lowest part of the fertile region of the receptacle, as was shown by the fact that in the next lower section the axis bore only bracts and hairs. It is clear that the sterile portion of the receptacle passed abruptly upwards into the fertile region, and it is extremely unlikely that any microsporophylls were borne at the base of that portion of the cone-axis which produced the scales and megasporophylls. The cone was, in all probability, unisexual. On the analogy of the cones shown in figs. 513, 514, one would expect to find between the sterile and fertile regions either a verticil of microsporophylls or the remains of an annular disc from which the effete sporophylls had been detached. There is no trace of any such disc, and the fact of the immaturity of the megasporophylls renders it unlikely that were the cone bisexual the microsporophylls would have been detached. As previous records show, there is nothing improbable in the occurrence of a unisexual Bennettitean flower. These remarks are made in view of an opinion expressed by Dr Wieland that the bracts with lateral appendages (fig. 561, l), to which allusion has been made, are microsporophylls and that if the cone had been sliced longitudinally the presence of a microsporophyll-disc would have been discovered. The latter possibility has already been considered, and as regards the former there is nothing in the structure of the small lateral appendages of the longest bracts to indicate that they were connected with spore-production. It is not unlikely that the bracts with small outgrowths (fig. 561, l) correspond to the more leaf-like bracts of Wielandiella and Williamsoniella. The two sets of organs spoken of as interseminal scales and megasporophylls are probably homologous, foliar, structures; in the one case leaves transformed into cylindrical organs bearing terminal integumented and undifferentiated megasporangia and, in the other, sterile or sterilised sporophylls. The polygonal truncate distal end of an interseminal scale is flat or slightly concave and covered by a thick epidermis, and on the sides of the scale many of the surface-cells are strongly papillose (figs. 563, 564). The rest of the interseminal scale consists mainly of elongated cells, which in the lower portion of the axis of the scale assume a tubular form, presumably immature conducting elements: in one scale only was any tracheal tissue found and that was represented by 2–3 scalariform tracheids. The scales appear to arise from the axis like the bracts as superficial outgrowths, and probably in a later stage of development the centre of each scale would be occupied by a vascular strand. The megasporophylls bear a close resemblance to the scales, but in transverse section they appear as smaller and circular organs each the centre of a group of polygonal interseminal scales precisely as in other Bennettitean flowers (fig. 564; cf. fig. 515). The proximal part of a megasporophyll consists of a column of parenchyma (fig. 563, A, s) extending through half of the length; from this column is detached a narrow cylinder of small crushed cells which most likely represents the remains of tissue that originally occupied the space surrounding the axial column. At a higher level the axial column becomes broader and its short cells more elongated and slightly divergent towards the sloping sides of the conical nucellus. The loose cylinder of tissue is attached to the nucellar cone and prolonged beyond its apex as a broad integument enclosing a very small micropyle (fig. 563, C). The apex of the integument has the form of a shallow funnel: its epidermal cells are papillose (fig. 563, B, C) and the presence of short transversely elongated cells is a characteristic feature of the tissue lining the micropylar canal. The bracts agree generally with those of Cycadeoidea Gibsoniana, Cycadeoidea Morierei, and the American species. The ground tissue is composed of sclerenchyma comparable with the scalariform elements in the bracts of Cycadeoidea Gibsoniana (cf. fig. 520). It is in the possession of long hairs like those on the leaves of Dioon and other recent Cycads that Williamsonia scotica differs from previously described flowers in all of which the fern-like ramental scales are a conspicuous feature. It is interesting to find that similar hairs are substituted for scales in some Indian stems described by Miss Bancroft[1258]. Lignier[1259] mentions the occurrence of long unicellular hairs on Cycadeoidea micromyela (p. 415), a Jurassic French species, but the ramenta are in part multicellular lamellae and the presence of transitional forms suggests a possible derivation of hairs from scales both in fossil species and in recent Cycads. The megasporophylls and interseminal scales are much shorter than in Cycadeoidea Gibsoniana and other species in which the axis of the cone forms a depressed receptacle (cf. fig. 521, A, C), but they correspond closely with those of several American species. In Cycadeoidea Gibsoniana and C. Morierei the distal ends of the interseminal scales are much broader and their diameter greatly exceeds that of the micropylar tubes, 2·8 mm. as compared with 0·25 mm., whereas in Williamsonia scotica the scales are 0·23 mm. broad and the micropylar tubes 0·15 mm. The mummified micropylar tubes of W. pecten bear a striking resemblance in form and in the papillose epidermal cells to those of the Scotch species.
There are two additional points suggested by the structure of the fertile region, namely the possibility that the megasporophylls are arrested rather than immature organs and, secondly, the method of pollination. In regard to the first there would seem to be no adequate reason for doubting the correctness of the view that the sporophylls are potentially perfect ovules which were petrified at a comparatively early stage in development. The dense woolly covering investing the surface of the scales and megasporophylls recalls an inflorescence of Aesculus hippocastanum in its winter-fur and hardly suggests a collection of ovules accessible to microspores. In all probability at a later stage the protecting bracts with their felt of hairs would bend outwards leaving exposed the receptive micropyles.
In view of the association of microsporophylls and ovulate strobili in the flowers of Cycadeoidea described by Wieland, most of which are bisexual, it is reasonable to expect a similar association in the flowers of Williamsonia which agree closely in the essential features of both micro- and mega-sporophylls with those of Cycadeoidea. It is, therefore, surprising that in no single case have the microsporophylls attributed to Williamsonia been found in actual connexion with a receptacle bearing interseminal scales and megasporophylls. The same statement holds good with regard to the Williamsonias discovered in Mexico. Nathorst believes that the microsporophylls on which he has founded several species are unisexual flowers with the possible exception of W. pyramidalis[1260]. This species, found by Dr Halle at Cloughton Wyke, is represented by a small ovulate strobilus characterised by a conical receptacle with a blunt mucronate apex: with it is associated a microsporophyll bearing synangia. The orientation of the two specimens is such as to suggest an original connexion. As Nathorst says, there is, however, no proof that the two belong to one flower. Wieland[1261], though believing that the existence of bisporangiate Williamsonia flowers is ‘reasonably certain,’ agrees with Nathorst’s conclusion as to the unisexual character of W. spectabilis and W. pecten. On the other hand, he regards the microsporophyll-verticil which was first described by Williamson as a carpellary disc, then named by Nathorst[1262] W. bituberculata and afterwards identified as a microsporophyll-disc closely allied to W. whitbiensis, as the staminate part of W. gigas. This view is, in my opinion, impossible to reconcile with the nature of the specimen. If, as Wieland suggests, it is the staminal collar split off from the base of a large ovulate cone like that of Williamsonia gigas, one would expect to find a central space in the middle of the cupular base large enough to embrace the receptacle. Neither in this specimen nor in several other forms of microsporophyll-verticils is there such a central space. It is clear that the discs described as W. spectabilis, W. whitbiensis, and other species were not borne as concrescent collars on a stout axis as are the corresponding organs in Cycadeoidea. The Indian specimen reproduced in fig. 557 and the very closely allied type W. setosa are incomplete at the centre and may possibly have been borne at the base of an ovulate bisporangiate strobilus, but there is no definite evidence that this was the case. Moreover, in W. spectabilis the lower part of the cup (fig. 551) easily splits away from the rest of the staminate disc, and this may explain the central space in the specimen shown in fig. 549, A. If W. spectabilis, to take one example, is a complete flower there are certain difficulties which are not easily explained: as Nathorst has shown, in this type there is a short stalk, but in W. whitbiensis the base of the funnel has no stalk and there is no interruption of the stout lamina at the centre or any indication of a scar. Nathorst compares the funnel-like region of W. whitbiensis to a kind of cupule which became detached after flowering[1263]. But a cupule is supported on an axis and, though no scar is apparent on some of the more complete specimens, it is obvious that the verticil must have been supplied with vascular tissue from some axial organ. This brings us to the consideration of a morphological point which cannot be definitely answered. Nathorst has previously raised the question—is it possible that the microsporophylls were attached to the upper part of an ovulate strobilus; were the flowers bisexual and epigynous? He believes the answer to be in the negative. Reference has already been made to the probable occurrence at the apex of the receptacle of Williamsonia gigas of a funnel-like organ of the type described by Williamson as ‘carpellary disc[1264],’ a term under which Nathorst[1265] believes that Williamson included two different things,—staminate discs borne on separate, unisexual, flowers, and sterile organs called by Lignier[1266] the infundibuliform apparatus. The latter, it is believed, were attached to the apex of an ovulate strobilus as shown in fig. 548, comparable in position with the leaves at the summit of an inflorescence of Ananas. A comparison of the fossils regarded as infundibuliform appendages with some of the microsporophyll-verticils shows that they are identical in form, the only difference being that on the former there are no synangia. This fact can hardly be regarded as negative evidence fatal to the morphological identity of these sterile and fertile organs. The available evidence, though far from complete, is favourable to the view that in some Williamsonia flowers, e.g. W. gigas, the microsporophylls were produced at the apex of the axis in the position shown in fig. 548. To cite a rough analogy,—in Cycadeoidea the flower was hypogynous as in Erica; in Williamsonia epigynous as in Vaccinium. This view does not exclude the possibility of the existence of unisexual flowers in some species, but the evidence in favour of a separation of the sexes is by no means decisive. Another difficulty is the absence of protective bracts in the case of the microsporophylls, a striking contrast to the bract-enclosed ovulate strobili of Williamsonia or to the bract-covered flowers of Cycadeoidea. Were the microsporophylls borne on a separate axis general considerations would lead us to expect the association of bracts with the essential organs. The fact that no such association has been recorded is a fact favourable to the view that the flowers were bisporangiate.
The student is warned that the opinion expressed with regard to the position of the microsporophylls is contrary to that which is held by several palaeobotanists[1267].
This species is one of the numerous types discovered by Wieland in Mexico[1268]. It is distinguished by the deep campanulate concrescent portion of the whorl of microsporophylls and by the ten short free lobes which bear two rows of lateral synangia (fig. 565, C). Among the British forms Williamsonia whitbiensis would appear to be the most closely allied type. W. mexicana occurs in association with Otozamites fronds, as is the case with an Italian specimen figured by Zigno[1269], but in the Mexican species there is no conclusive evidence of organic connexion.
A Jurassic species from Afghanistan[1270] founded on a single specimen of a flattened broadly-oval flower similar to W. gigas but smaller and enclosed by narrow linear bracts. An interesting feature is the occurrence of a funnel-shaped depression at the apex showing radiating lines on its surface as in the Williamsonia represented in fig. 546, A. In all probability these lines denote the persistence of a collar of interseminal scales on the upper part of the elongated conical receptacle. An examination of the carbonaceous material revealed the presence of some short rods agreeing in size and shape with the interseminal scales of other species.
Zigno[1271] instituted the generic name Blastolepis for some specimens from the Jurassic rocks of north Italy which he referred to three species, B. falcata, B. acuminata, and B. Otozamitis. These are true Williamsonias closely allied to W. gigas. An examination of the type-specimen of B. Otozamitis in the Padua collection led me to the conclusion that Zigno was correct in representing the ovulate strobilus, which shows an annular area like that in specimens of W. gigas, as being in organic connexion with an Otozamites frond. Wieland has recently discovered species of Williamsonia in Mexico that he correlates with Otozamites leaves.
Though by no means satisfactory as a trustworthy record of the genus, the specimens so designated from the Kimmeridgian of Scotland[1272] are probably imperfect examples of the genus.
The type-specimen from Jurassic rocks in Oregon[1273] consists of a stout axis bearing several contiguous linear bracts forming a more or less spherical cluster 5 cm. broad. The form suggests a Williamsonia, but the evidence is hardly decisive.
A Bornholm species[1274] from Lower Jurassic strata represented by an annular zone surrounding a central area, probably a portion of an ovulate strobilus.
This Upper Jurassic species from Portugal[1275] consists of a number of small bracts or leaves attached to a central axis and is of no value as evidence of the occurrence of Williamsonia.
This species was first described from the Amboy clays as Palaeanthus (Williamsonia) problematica[1276] and compared to a Composite inflorescence with long ray-florets. More recently Hollick[1277] has referred to this species specimens from the Cretaceous of Long Island and Martha’s Vineyard, but none of them afford conclusive evidence of Williamsonian affinity. Some of the examples, as suggested by Hollick, agree fairly closely with Williamsonia cretacea Heer.
The curious urn-like bodies from the Amboy clays[1278] on which this species is based are not unlike the capsules of a Papaver but bear no real resemblance to a Williamsonia.
A species from the Cretaceous rocks of Martha’s Vineyard[1279] referred to Williamsonia on rather slender grounds and represented by imperfect material.
This fossil from the Cretaceous of the Black Hills, Dakota[1280], is much too imperfect to be accepted as a record of Williamsonia.
This species, founded on part of a conical receptacle with some appendages superficially resembling those of Williamsonia[1281], is too incomplete to be determined with any confidence. The type-specimen is from Cenomanian beds in Kansas.
This is undoubtedly a true member of the genus; it is characterised by narrow ovate bracts bearing conspicuous hairs surrounding the base of a receptacle which shows a portion of an annular zone of interseminal scales[1282].
This species from the Potomac beds[1283] is of no value as a record of Williamsonia.
The imperfectly preserved specimen from the Potomac series so named by Ward[1284] is probably part of an Abietineous cone as stated by Berry[1285] who includes Ward’s type with other specimens referred to Abietites macrocarpus Font.
Two specimens, figured by Heer[1286] from the Lower Cretaceous of Greenland, on which this species is founded, consist of an axis covered with small scale-leaves and at the broad apex bearing numerous narrow linear bracts forming a more or less spherical cluster 3·8 cm. in breadth. The species resembles W. problematica (Newb.).
The specimens from the Middle Cretaceous of Canada figured by Dawson[1287] under this name are very imperfect and of no botanical value.
The specimens on which this genus is founded were originally described by Nathorst from Höör in Scania as Williamsonia angustifolia[1288]: in a second paper[1289] a restoration of the plant was published (fig. 566). The examination of additional specimens from the Rhaetic of Bjuf and of cuticular preparations led to the establishment of a new genus Wielandia[1290] for which Wielandiella[1291] was afterwards substituted, Wielandia having been previously employed for an existing plant. Wielandiella agrees in the general morphology of its bisporangiate flowers with Cycadeoidea, but differs widely from nearly all other members of the Bennettitales in the repeatedly forked slender stem which is in marked contrast to the vegetative axis of any recent Cycad. Among recent Cycads with terminal strobili an indication of a primitive dichasium is afforded by the occurrence of an aborted bud in a stem of Dioon edule described by South and Compton[1292]; but in the habit of the stem Wielandiella is far removed from any recent Cycadean type. The microsporophylls are smaller and simpler than in Williamsonia or Cycadeoidea and the foliage-leaves are of the Anomozamites type (cf. fig. 615).