late branchlets, or on the lower part of the stem the marks of the attachment of the roots. The Calamites grew in dense clumps, budding off from one another, sometimes at different levels, as the mud or sand accumulated about their stems, and in some species there were creeping rhizomata or root-stocks (Figs. 46 to 49). But all Calamites were not alike in structure. In a recent paper * Dr. Williamson describes three distinct structural types. What he regards as typical Calamites has in its woody zone wedges of barred vessels, with thick bands of cellular tissue separating them. A second type, which "Memoirs of the Philosophical Society," Manchester, 1886-'87. he refers to Calamopitus, has woody bundles composed of reticulated or multiporous fibres, with their porous sides parallel to the medullary rays, which are better developed than in the previous form. The intervening cellular masses are composed of elongated cells. This is a decided advance in structure, and is of the type of those forms having the most woody and largest stems, FIG. 49.-Erect Calamites (C. Suckovii), showing the mode of growth of new stems (b), and different forms of the ribs (a, c). (Pictou, Nova Scotia.) Half natural size. which Brongniart named Calamodendron (Fig. 50). A third form, to which Dr. Williamson seems to prefer to assign this last name, has the tissue of the woody wedges barred, as in the first, but the medullary rays are better developed than in the second. In this third form the intermediate tissue, or primary medullary rays, is truly fibrous, and with secondary medullary rays traversing it. My own observations lead me to infer that there was a fourth type of calamitean stem, less endowed with woody matter, and having a larger fistulous or cellular cavity than any of those described by Dr. Williamson. There is every reason to believe that all these various and complicated stems belonged to higher and nobler types of mare's-tails than those of the modern world, and that their fructification was equisetaceous and of the form known as Calamostachys. We have already seen that noble tree-ferns existed in the Erian period, and these were continued, and their number and variety greatly extended, in the Carboniferous. In regard to the structure of their stems, and the method of supporting these by aërial roots, the tree-ferns of all ages have been nearly alike, and the form and structure of the leaves, except in some comparatively rare and exceptional types, has also been much the same. Any ordinary observer examining a collection of coalformation ferns recognises at once their kinship to the familiar brackens of our own time. Their fructification is, unfortunately, rarely preserved, so that we are not able, in the case of many species, to speak confidently of FIG. 50.-Stems of Calamodendron and tissues magnified (Nova Scotia). a, b, Casts of axis in sandstone, with woody envelope (reduced). c, d, Woody tissue (highly magnified). their affinities with modern forms; but the knowledge of this subject has been constantly extending, and a sufficient amount of information has been obtained to enable us to say something as to their probable relationships. (Figs. 51 to 55.). The families into which modern ferns are divided are, it must be confessed, somewhat artificial, and in the case of fossil ferns, in which the fructification is for the most part wanting, it is still more so, depending in great part on the form and venation of the divisions of the fronds. FIG. 51.-Group of coal-formation ferns. A, Odontopteris subcuneata (Bunbury). B, Neuropteris cordata (Brongniart), c, Alethopteris lonchitica (Brongniart). D, Dictyopteris obliqua (Bunbury). E, Phyllopteris antiqua (Dawson), magnified; E1, Natural size. F, Neuropteris cyclopteroides (Dawson). Of about eight families into which modern ferns are divided, seven are found in a fossil state, and of these, four at least, the Cyathacea, the Ophioglosseæ, the Hy |