The microsporophylls (figs. 389, A; 392, E) are in all genera aggregated into strobili which often bear a close resemblance to seed-cones (fig. 393, A). On a single sporophyll of Cycas circinalis there may be as many as 700 sporangia while in Zamia floridana there are only two microsporangia. The spore-output is large and in extreme cases, e.g. in Dioon spinulosum, the average number of spores in a sporangium is said to be 30,000[61].

Seeds. In the great majority of recent species the seeds may be described as large and afford a striking contrast to the small seeds of the Mesozoic Bennettitales. A feature of interest from the point of view of comparison with Palaeozoic seeds is the absence of a resting stage, germination in some cases following seed-fall without an interval. As Warming pointed out, the embryo is often undeveloped when the seeds are shed. An interesting fact is recorded by Capt. Dorrien-Smith[62] with regard to seed-dispersal: he describes the heavy pebble-like seeds of a Macrozamia as being hurled from the ripe cones a distance of 12 ft. The seeds of Cycas are platyspermic; the woody shell exposed on removal of the outer flesh is slightly flattened and has two prominent angles, but three-angled seeds may occur as in Ginkgo biloba (fig. 631, C). In other genera the seeds are radiospermic. The seed of Encephalartos Altensteinii[63] (fig. 396, D) has a square-cut distal end with a small papilla at the summit of the unusually long micropylar canal (17 mm.). The stone of this seed (fig. 396, C) shows parallel curved ridges which mark the position of vascular strands in the inner region of the outer flesh. The large ovules of Cycas circinalis[64] have an integument 1 cm. thick consisting of an outer and inner flesh and an intervening stony layer which reaches its greatest development at the base and apex. Three vascular strands enter the base of the seed, the concentric strand breaks up in the broad inner flesh into a group of bundles which embrace but do not penetrate the lower end of the nucellus. Each of the two lateral strands branches in the outer flesh near its entrance into the seed; the outer and larger collateral and mesarch bundle passes up close to the surface of the shell to the seed-apex, while the inner branch penetrates the shell and, occasionally branching, passes up the inner region of the inner flesh as far as the micropyle. In other seeds the tracheal supply of the outer flesh consists of several bundles and not two as in Cycas. The inner flesh abuts on the nucellus and is connected with it except at the apex (fig. 396, B). In ripe seeds the nucellus is reduced to a thin membrane enclosing the large megaspore at the upper end of which is a depression (fig. 396, B′) or sometimes two depressions (fig. 396, I) in the prothallus containing the archegonia. In the seed of Dioon edule[65] (fig. 396, A) the position of the absciss-layer (s) is indicated by a slight transverse constriction. In the seeds of Bowenia, constructed on the same plan, the inner series of vascular strands appears to be nucellar in position, thus differing from the strands in Dioon, Cycas, and other genera which are confined to the integument. Miss Kershaw[66] in describing Bowenia speaks of an upper and a lower pollen-chamber; the former serves as a storage-place for the microspores prior to their further development in the lower chamber. Dr Stopes[67] regards the integument as double in origin, a view suggested by Griffith[68] in 1835, and as homologous with the single integument plus the cupule of Lagenostoma. This view is supported by Mrs Thoday[69]: on the other hand Miss Kershaw’s investigation of Bowenia seeds leads her to regard the integument as single. Although there would seem to be a prima facie case in favour of the dual nature of the integument, the arguments on the other side have greater weight[70].

Recent observations point to the probability that insects play a part in the pollination of cycadean ovules. Kraus[71] drew attention to the strong smell emitted by the microstrobili of Dioon edule and noticed that small bees were attracted to the ripe strobili of Macrozamia, while odourless cones of a neighbouring Ceratozamia received no attention. Pearson[72] and Rattray[73] have obtained evidence that beetles and weevils act as pollinators to species of Encephalartos.

Anatomical features. Allusion has already been made to some of the more striking anatomical features; the large pith, the occasional occurrence of medullary vascular bundles, the presence of one or more cambiums, the large size of the medullary rays, etc. It is worthy of remark that the occurrence of an anastomosing system of medullary bundles is not a constant feature within a genus; in Macrozamia Fraseri such a system is present, but absent in M. Denisoni[74]. In the pith of stems with no medullary bundles cylinders of collateral bundles may occur in connexion with a fertile shoot. These bundles arise from the inner face of the main cylinder and pass upwards as a domical system into the base of the terminal strobilus which is eventually pushed to one side by the growth of a lateral bud[75]. The secondary xylem tracheids are usually provided with several rows of bordered pits on the radial walls and resemble those of the Araucarieae[76], but in Cycads the pits are often not contiguous and less compact in their distribution. The wood of Stangeria is peculiar in consisting of scalariform tracheids[77] (fig. 397). Chamberlain describes growth-rings in the wood of Dioon; but this is exceptional. In tangential sections of the stele leaf-trace bundles are constantly seen passing horizontally through the broad and deep medullary rays. The pith-cast of a cycadean stem reproduced in fig. 398 shows the wide meshes in the reticulum of tracheal tissue originally occupied by parenchyma, which on decay left lenticular depressions represented on the cast by tapered convex areas occasionally bearing the impress of an outgoing trace in the form of a narrow groove. The secondary phloem often rivals the xylem in breadth and is not always easily distinguishable from it; it consists of sieve-tubes, parenchyma, and fibres. The secondary cambial cylinders characteristic of Cycas, Encephalartos, Macrozamia, and Bowenia, to which reference was made in the summary of anatomical features, arise in the pericycle, and a few layers of pericyclic parenchyma occur between adjacent extrafascicular cylinders of xylem and phloem. In a stem of Cycas media 35 cm. in diameter examined by Worsdell there were 12 concentric cylinders. Matte[78] and Miss Dorety[79] have described partially flattened arcs of extrafascicular xylem and phloem in the hypocotyl of Ceratozamia mexicana. Worsdell[80] first drew attention to the occasional occurrence of short tracheids on the inner edge of the secondary wood and to the spasmodic development of cambial arcs in the tissue between the extrafascicular cylinders forming strands of inversely orientated xylem and phloem. More recent work by Matte gives support to Worsdell’s comparison between Medullosan stems and those of recent Cycads with inversely orientated arcs or concentric vascular cylinders. The French author draws attention to the close resemblance between the seedling stems of such species as Encephalartos Barteri (fig. 396, K) and Cycas siamensis (fig. 396, L) with their polystelic type of structure and the adult stems of Medullosa[81]. In the stems of Dioon, Microcycas, Stangeria, and Zamia no extrafascicular cylinders are recorded. Two main vascular bundles enter the cortex from each leaf-base and in most stems these diverge right and left and more or less completely encircle the stele before passing through the medullary rays and joining the inner portion of the xylem of the stele either as double or single bundles. These girdle-bundles (fig. 396, H) first described by Karsten and Mettenius form a very characteristic cycadean feature[82]. Adjacent girdles are joined by connecting cortical bundles and, in addition, there are cauline collateral bundles in the cortex which form an anastomosing system. In some cases, e.g. species of Macrozamia and occasionally in Stangeria, the female peduncle of a Ceratozamia, and in seedlings of Bowenia and Cycas revoluta[83], the leaf-traces pursue a direct course from petiole to stele as in stems of Bennettitales. It is noteworthy that in seedlings of Microcycas[84], a genus characterised by a large number of male gametes—presumably a primitive feature—the leaf-traces are of the girdle-type. The two bundles at the base of a petiole by repeated subdivision give rise to the numerous collateral strands of the rachis. A leaf-trace in its passage to the leaf is like that of a Conifer in having the protoxylem on its inner edge, whereas in the petiole and elsewhere in the frond it is characterised by an arrangement of the xylem that has usually been described as mesarch. A typical vascular bundle from a cycadean frond is seen in fig. 399, C; by far the greater part of the xylem is centripetal, the centrifugal xylem being confined to an arc of scattered tracheids or a small strand separated by a few parenchymatous cells from the protoxylem.

As considerable stress has been laid on the anatomical features of the cycadean foliar bundles in discussions on the affinities and phylogeny of certain Palaeozoic genera, it is important to consider the facts more closely[85]. French anatomists described the cycadean bundle as diploxylic on the ground that the centripetal and centrifugal xylems are distinctly different things, the centripetal xylem being primary—a relic of a former organisation—and the centrifugal xylem secondary and homologous with the normal wood of the cauline bundle. The term mesarch has in recent years been applied to the cycadean type of bundle. A mesarch bundle is, however, one in which centripetal and centrifugal xylem are alike in origin, both being primary structures derived from a desmogen strand. Typical mesarch bundles occur in several recent ferns; in the stele of the Osmundaceae, Gleichenia, and other genera; but in these plants the xylem is all produced directly from one primary desmogen region and there is no question of ‘primary’ and ‘secondary’ as in the two portions of the xylem of a cycadean bundle. Recent researches into the development of cycadean foliar bundles show that they do not conform to the mesarch type as generally understood. A leaf-trace at the base of a petiole (fig. 399, A) comprises centrifugal xylem only, and this consists of regular rows of tracheids separated by medullary rays: in the lower part of the petiole the structure is gradually modified, the centrifugal xylem is reduced and the formation of centripetal xylem is initiated. At a higher level (fig. 399, B) the centripetal xylem is in excess of the centrifugal and the latter, for a time connected with the former, eventually becomes separated by a few parenchymatous cells from the protoxylem and persists as a small strand or arc of tracheids. Fig. 399 illustrates stages in the transformation of a typical collateral bundle, at the base of a Stangeria petiole, into one in which the xylem is almost wholly centripetal at a higher level in the axis of the frond. A cambium is present in all: in B the centrifugal xylem is more or less clearly differentiated into two portions, loosely arranged tracheids near the phloem, and the more compact groups abutting on the centripetal xylem: figs. C–E show a further reduction in the centrifugal tracheids. The conclusion drawn from developmental study is that the two xylem portions of the bundle are independent in origin[86]. Marsh has, however, shown that in Stangeria bundles near the base of the petiole the centrifugal xylem consists of rows of secondary tracheids and an inner portion not in rows which connects the centrifugal with the centripetal elements; this connecting portion, he adds, is ‘probably primary and connects up the Cycadean foliar bundle with the truly mesarch bundle of the Cycadofilices.’

In the xylem portion of the bundle from the midrib of a forked pinna of Cycas Micholitzii shown in fig. 400 the centrifugal xylem elements are unusually numerous: the space between the two xylems is occupied by parenchyma and the whole strand is enclosed by a sheath of crystal-containing cells, s, with thick inner walls. Fig. 400, 1–4, illustrates the gradual change in the form of the bundle in the region of dichotomy[87]. The ground-tissue of the petiole is abundantly supplied with secretory canals and in the hypodermal region is a cylinder of stereome. In some petioles, e.g. Macrozamia heteromera[88], the ground-tissue cells are lignified and reticulately pitted, a feature met with in some Mesozoic cycadean leaves[89]. In Cycas media Worsdell noticed a tendency of the leaf-trace bundles towards a concentric arrangement and similar vascular strands are recorded in the peduncle of Dioon edule, in various sporophylls[90] and in other cases. It is possible, as Worsdell believes, that the fairly frequent occurrence of concentric bundles in plants characterised by collateral bundles may have a phylogenetic significance.

The pinnae are dorsiventral and the veins exarch or pseudomesarch: secretory canals occur between (Encephalartos), above, or below the veins. The mesophyll of Cycas is characterised by the presence of isolated xylem-elements passing from the midrib to the edge of the lamina and, as Lignier[91] suggests, these may be regarded as a reduced system of lateral conducting strands.

The epidermal cells of the leaflets have straight or slightly curved walls except in Stangeria where they are undulate and fern-like[92]. The stomata, with few exceptions confined to the lower epidermis, are larger than in other gymnosperms (on the average ·075 × ·034 mm.) and are more or less depressed below the surface; the guard-cells are usually surrounded by 4–6 subsidiary cells.

The roots exhibit no feature to which attention need be called: the pericycle is several cells broad and as in the stem there may be extrafascicular cylinders of xylem and phloem.