On the basis of the leaf forms Grand 'Eury pus. On the basis of wood structure classifica tion is likewise provisional, since studies of entire stems are seldom or never possible to carry out. Three more or less overlapping divisions are also distinguished by Scott. (1) divided the group into the genera Cordaites, Dory- and Poa-Cordaites. The first of these divisions (cf. Fig. 3) includes mainly very large spatulate and blunt-end leaves, the second large lanceolate -and sharply pointed leaf types, and the third grass-like forms. The genus Cordaites, as first used, may include any of these. Isolated stems have usually been described as Cordai oxylon, or often as Arancarioxylon, because of the structural similarity to the wood of modern Araucarias. Branches can be called Cordai cladus, and the large isolated pith casts, look ing like piles of coins, were long only known as Artisia or Sternbcrgia. Staminate fruits are named Cordaiantkus, and the seeds Cordaicar The Poroxylme were far the most fern-like. The stems were slender, only a few centimeters in diameter. They had long internodes and a double leaf trace bundle as in Lyginopteris, from which they mainly differ in their lack of centrifugal primary xylem. The petioles are known, and the bilateral seeds called Rhabdo carpus appear to belong here. (2) The Pityea include Pitys (with three species from the lower Carboniferous of southern Scotland), Arcistropitys of the Waverly-shale of Kentucky, and Callisylon. A very fine stem of the latter from the Indiana Devonian is called Callixylon Oweni, and a related species is Russian. This group is only known from the pith and wood. The trees reached large size, but still retained mesarch strands of primary xylem traversing the medulla next the secondary wood. They are thus of great interest as forms intermediate between the fernace Poroxylese and typical Cordaiteans. (3) The Cordaitez are the most advanced type and the greatest forest makers. Some species reached 150 feet in height and three feet in diameter. The stems still more closely resemble Araucaria in structure, the elements of the wood all being radially ar ranged. No distinction between primary and secondary wood appears in the transverse sec tion. The strands of 4old cryptogamic wood," characteristic of the preceding divisions, are lost. In the radial section narrow spiral elements of the protoxylem next the broad discoid pith are succeeded by wider spiral tracheides, many rows of scalariform eletnents, and lastly the pitted tracheides forming the bulk of the wood. The pits of this wood are generally in two or more densely crowded rows. Their alternation tends to produce hexagonal outlines, though isolated pits retain their circular form. The pore is usually an oblique slit, and the crossed slits of appressed facing pits are highly charac teristic. The medullary rays are thin, mostly one cell thick. The secondary wood is thus quite indistinguishable from that of Araucana. The phloem is radial, the cortex parenchyma tous and provided with secretory sacs. The outgoing leaf traces are double, as in Araucaria, Ginkgo and various primitive gymnospermous forms. There was a thick bark, recalling that of Araucaria. As in that modern type there are diarch roots with spiral protoxylem and the scalariform tracheides, followed by the secon dary pitted wood. Triarch and tetrarch roots of similar structure are known. The very large cycad-like pith of all these stems appears to owe its discoid structure to tension during growth; but this feature may be seen in vanous modez-n plants such as the hickory and the walnut Growth rings of the discontinuous or Aran carian type may occur. Amongst forms struc turally antecedent to the three foregoing groups, and more fern and cycad-like may be mentioned the stems classed as the Calamopityee with petioles lcnown as Kalymma. Well-conserved forms of these occur in the Waverly of Ken tucky at the close of Devonian time. The leaves of the Cordaitaleans are conspicuously parallel veined with repeated dichotomy in all but the narrowest grass-hke forms. Anatomi cally they are very similar to the single pinnules of either the cycads or the cycadeoids. As in the latter the structure is well lcnown from silicified specimens. The bundles are usually described as of the mesarch type, but the cen trifugal xylem may fail. Each is surrounded
by a strong sheath, usually connected with heavily developed hypodermal sclerenchyma. Intervening strands of the latter tissue may give to the leaf secondary venation.
Fructification, The knowledge of the floral morphology of the Cordaitales now con stitutes one of the most interesting chapters of paleobotany. It largely rests on the labors of Grand 'Eury and Renault, the latter having been so fortunate as to find silicified strobili of both sexes of certain species preserved in astonishing detail in the siliceous fragments or nodules of Grande Croix. Both male and female flowers, as well as seeds, have often been found directly attached to the leafy stems or branches (Figs. 4, 5, 7). The numerous flowers are borne in more or less catkin-like male and female inflorescences. Such recall those of Gnetaleans or even angiosperms. The floral order is usually spiral but may be distichous. No amphisporanr-iate flowers are known, al though one form, CordiatIthus Penjoni, sug gests that such have occurred. The usual condition indicates either moncecism or dioecism. The staminate flower is about a centimeter long and consists of a thick axis covered by spirally arranged bracts with interspersed fertile fila ments bearing erect tufts of sporangia (Fig. 5).
Or, as Renault says, “In the midst of the sterile bracts arose one or more fertile brac,ts, the fila ments of which scarcely modified bo're at their summits three or four sacs containing the pol len.) There is, therefore, a noteworthy analogy to the ovulate cones of the Cycadeoids. In these the stems bearing single erect seeds ace likewise interpolated atnongst sterile organs or scales.
The pollen grains which have been found both in place and actually in the pollen cham bers of the macrospores are large and contain a highly interesting group of prothallial cells. This group is theoretically a transition stage between Pteridophytes and Spermatophytes. An extraordinary instance is recorded. In a -seed of Cordianthus Grand 'euryi the outer integu ment is partly broken away, the nucellus being left standing free. In its upper end the pollen chamber contains the pollen grains as in Ginkgo and the cycads. An actual stage in pollination is thtts seen. It is of fundamental' importance that the pollen grains found in the nucellar canal are larger than those in the sporangia, and their group of internal cells more devel oped; while grains found free in the matrix exhibit an intermediate condition. Evidently the pollen grains continued to grow after their discharge from the anther, and especially after entrance into the pollen chamber. Such a con dition is one of the most readily thinkable later stages in the evolution of heterospory and seeds from asexual spore-producing diceecious free prothalli. The presence of free swimming spermatozoids is surmised.
The seed-bearing spikes are of varied form; in some, solitary ovules accompanied by a few bracts are inserted directly on the axis (Fig. 6) ; in others this arrangement pears to be compounded, the lateral catkins each containing several ovules and in their younger stages much resem bling the staminate forms. The ovule was thus borne on a short axillary shoot as in Taxus or Ginkgo. In such forms basal stamens may earlier have been present, forming a Tumboa-like fruit (Fig. 5). While homology with other gymnosperms is not very close, it is evident that this open type of inflorescence was capable of undergoing change in the direction of modern forms of compaction and reduction. Undoubted Cordaitean seeds (those found attached) are mostly cordate, and flat or platyspermic. The winged seeds about a centi metre long called Samaropsis (Fig. 7), were attributed by Grand 'Eury to the genus Dorycordaites, seen in the restoration Fig. 1. Others reach five or more centimetres in length, and owing to their size, heavy angles, and thick testa, are also fossils of strik ing conservation. But there is no gap either of structure or of form separating such seeds from the larger radiosym metric types abundant in the Palwozoic. In general the Cor daitean testa is more or less distinctly triple-layered, if the sclerotestal tissue separating the outer fleshy sarcotesta from the inner flesh be called a layer. The chalazal bundle.