with broad lateral wings for flotation in the air, or in some cases with a pulpy envelope, which flattens into a film. There seem to have been structures of both these kinds, though in the state of preservation of these curious seeds it is extremely difficult to distinguish them. In the first case they must have been intended for dissemination by the wind, like the seeds of spruces. In the latter case they may have been disseminated like the fruits of taxine trees by the agency of animals, though what these were it would be difficult to guess. These trees had very great reproductive power, since they produced numerous seeds, not singly or a few together, as in modern yews, but in long spikes or catkins bearing many seeds (Fig. 59).

It is to be observed that the Cordaites, or the Cordaitina, as they have been called, as a family,* constitute another of those intermediate groups with which we have already become familiar. On the one hand they approach closely to the broader-leaved yews like Gingko, Phyllocladus, and Podocarpus, and, on the other hand, they have affinities with Cycadaces, and even with Sigillariæ. They were beautiful and symmetrical trees, adding something to the variety of the rather monotonous Palæozoic forests. They contributed also somewhat to the accumulation of coal. I have found that some thin beds are almost entirely composed of their leaves, and the tissues of their wood are not infrequent in the mineral charcoal of the larger coal-seams. There is no evidence that their roots were of the stigmaroid type, though they evidently grew in the same swampy flats with the Sigillariæ and Calamites.

It may, perhaps, be well to say here that I believe there was a considerably wide range of organisation in the Cordaitinæ as well as in the Calamites and Sigillariæ, and that it will eventually be found that there were three lines

* Engler; Cordaitées of Renault.

of connection between the higher cryptogams and the phænogams, one leading from the lycopods by the Sigillariæ, another leading by the Cordaites, and the third leading from the Equisetums by the Calamites. Still further back the characters afterward separated in the club-mosses, mare's-tails, and ferns, were united in the Rhizocarps, or, as some now, but I think somewhat unreasonably, prefer to call them, the "heterosporous Filicinæ." In the more modern world, all the connecting links have become extinct and the phænogams stand widely separated from the higher cryptogams. I do not make these remarks in a Darwinian sense, but merely to state what appear to be the lines of natural affinity and the links wanting to give unity to the system of nature.

On

Of all the trees of the modern world, none are perhaps so widely distributed as the pines and their allies. mountain-tops and within the Arctic zone, the last trees. that can struggle against the unfavourable conditions of existence are the spruces and firs, and in the warm and moist islands of the tropics they seem equally at home with the tree-ferns and the palms. We have already seen that they are a very ancient family, and in the sandstones of the coal-formation their great trunks are frequently found, infiltrated with calcareous or silicious matter, and still retaining their structure in the greatest perfection (Fig. 60). So far as we know, the foliage of some of them which constitutes the genera Walchia and Araucarites of some authors (Figs. 60, 63) was not dissimilar from that of modern yews and spruces, though there is reason to believe that some others had broad, fern-like leaves like those of the gingko. None of them, so far as yet certainly known, were cone-bearing trees, their fruit having probably been similar to that of the yews (Fig. 61). The minute structures of their stems are nearer to those of the conifers of the islands of the southern hemisphere than to that of those in our northern climes-a cor

FIG. 60.-Coniferous wood and foliage (Carboniferous). A, Araucarites gracilis, reduced. B, Dadoxylon Acadianum (radial), 90 diams. ; B (tangential), 90 diams.; B2, cell showing areolation, 250 diams. c, Dadoxylon materiarium (radial), 90 diams.; o1 (tangential), 90 diams.; c2, cell showing areolation, 250 diams. D, Dadoxylon antiquius (radial), 90 diams.; D1 (tangential), 90 diams.; D2, cell showing areolation, 250 diams.

relation, no doubt, to the equable climate of There is not much evidence that they grew gillariæ in the true coal-swamps, though som have been found in this association. It is that they were in the main inland and upland

C

FIG. 61.-Trigonocarpum Hookeri, Dawson, from the coal-measures of Cape Breton. Probably the fruit of a Taxine tree. A, Broken specimen magnified twice natural size. B, Section magnified: a, the testa; men; c, the nucleus; d, the embryo. c, Portion of the su inner coat more highly magnified.

that in consequence they are mostly known t drifted trunks borne by river inundations into and estuaries.

A remarkable fact in connection with them, an ing also the manner in which the most durable v structures may perish by decay, is that, like the Co they had large piths with transverse partitions, a

ure which, as I have already mentioned, appears on a minute scale in the twigs of the fir-tree, and that sometimes casts of these piths in sandstone appear in a separate form, constituting what have been named Sternbergia or Artisiæ. As Renault well remarks with reference to Cordaites, the existence of this chambered form of pith implies rapid elongation of the stem, so that the Cordaites and conifers of the coal-formation were probably quickly growing trees (Fig. 62).

The same general statements may be made as to the coal-vegetation as in relation to that of the Erian. In

FIG. 62.-Sternbergia pith of Dadoxylon. A, Specimen (natural size), showing remains of wood at a, a. B, Junction of wood and pith, magnified. c, Cells of the wood of do., a, a; b, medullary ray; c, areolation.

the coal period we have found none of the higher exogens, and there are only obscure and uncertain indications of the presence of endogens, which we may reserve for a future chapter; but gymnosperms abound and are highly characteristic. On the other hand, we have no mosses or lichens, and very few Algæ, but a great number of ferns and Lycopodiaceae or club-mosses (Fig. 63). Thus, the coal-formation period is botanically a meetingplace of the lower phænogams and the higher cryptogams, and presents many forms which, when imperfectly known, have puzzled botanists in regard to their position in one or other series. In the present world, the flora most akin