OSMUNDACEAE 541 
and leaf, and even in the segmentation of the wings of the leaf, the condition 
of Osmunda and Todea is less regular and more bulky than is habitual in 
the Leptosporangiates: and in particular, in the segmentation of some of 
their roots, where four prismatic initials take the place of the single initial of 
the Leptosporangiates, there is a near approach to the structure seen in the 
Marattiaceae: also in the apex of the leaf of Osmunda and Todea there is 
a three-sided initial cell, as against the usual two-sided type of the Lepto- 
sporangiates. When these facts are put in relation with what has been 
demonstrated for their sporangia, where there is so strange an oscillation 
between the Eusporangiate segmentation and that typical of Leptosporangiate 
Ferns, it becomes clear that the Osmundaceae hold a transitional place 
as regards their embryonic, and meristematic structure. This harmonises 
readily with their mature characters, and with their probable early origin as 
shown by palaeophytological enquiry. 
Thus an examination of the Osmundaceae, living and fossil, leads to 
the recognition of the following characters as probably existent in the stock 
from which the family sprang. It had an upright, radially constructed shoot, 
as shown both by the living species and by the related fossils ; for though the 
embryo has the prone position in living forms, this is only a temporary juvenile 
phase (see pp. 213-215). The axis was protostelic, as indicated by the 
seedling structure, as well as by that of the earlier fossils: and though the 
stele tended to be disintegrated in the more recent types there is still no 
proof that the state of typical dictyostely was ever reached. The absence of 
leaf-gaps in the early condition of the seedlings, and in the early fossils, as well 
as the fact that the leaf-trace in all consists of a single strand, indicates an 
ultimate origin from a stock in which the leaf had not attained the ascendant 
in the shoot. The young parts were protected by mucilaginous hairs, 
ramenta being absent. The disposition of the relatively bulky sporangia 
was non-soral, either uniformly on both sides or margin of the leaf, or on 
the lower surface: the individual spore-output was relatively large, and 
the opening mechanism simple. These characters all point towards the 
Botryopterideae among known early forms, and make it appear probable 
that the source of the Osmundaceae is to be found in some near relation 
to that early family of Ferns. 
