824 FOSSIL PLANTS OF THE COAL-MEASURES. Passing to the second genus Heterangkim, there are many points of resemblance to the preceding type, but at the same time some striking differences in the structure of both stele and cortex. Instead of the central pith of Ly(jlnodendron. we have in Hetermu/mm an axial group of primary xylem tracheae interspersed with bands of conjunctive parenchyma. At the periphery of the stele occur the smaller elements of the wood. In most examples there is a ring of secondary xylem and phloem surrounding the central primary xylem, and beyond this a pericycle, and an inner cortex containing groups of characteristic radially elongated bands of sclerous cells. The more external part of the cortex consists of alternate plates of parenchyma and sclerenchyma as in Lyf/ino- dendron. At the peripheral part of the primary xylem region there are definite groups of tracheids similar to those of the primary xylem strands of the preceding genus, and here too the protoxylem elements are situated internally in each xylem-strand. The term mesarch has been adopted for xylem groups in which the protoxylem occupies this internal position. The course of these peripheral strands, which are continuous with the leaf-traces, differs some- what from that of the foliar bundles in Lyginodendron, but it is impossible to deal with the two genera in any detail. In one species, Heterangium tilceoides Will., the tissues of the secondary phloem have been preserved in unusual perfection, the sieve-plates being very clearly shown on the lateral walls. The leaves and petioles agree fairly closely with those of Lyginodendron, and a similar correspondence is found as regards the adventitious roots. Having thus sketched in rough outline some of the characteristic features of these two genera, we must in conclusion draw attention to their affinities with recent forms. In neither case have we as yet any satisfactory evidence as to the nature of the reproductive organs, and are thus left to discuss the question of systematic position on the evidence of vegetative structures alone. Heterangium is the more fern-like of the two forms ; the structure of the stele agrees in essentials with that of a monostelic fern such as Gleichenia. In both fossil genera the leaves and petioles are distinctly of the fern type. On the other hand, the secondary wood of the stem resembles that of a Cycad. The primary xylem bundles of each genus are practically identical with those of a recent cycadean petiole, in the position of the protoxylem elements. The roots, again, are fern- like in character, and have been compared with those of the MarattiacecB. '* We have seen how extraordinary a combination of characters belonging to various groups these genera present. In different parts of their structure they have been found to present points in common with Gldcheniacece, OsmundacecB, MarattiacecB, and Cycadece. The view of the affinities of Lyginodendron and Heter- angium which we desire to suggest is, that they are derivatives of an ancient and 'generalized' fern-stock, which already show a marked divergence in the cycadean direction." .... "We think the existence of a fossil group on the borderland of Ferns and CycadecB is now well established. The relation of these forms to those very ancient Gymnosperms, the CordaitecB, is a difficult and
