Bower. — Studies in the Phytogeny of the Fili cates. 293 
plates it as probable for them that though the original ancestry may have 
been radial, all the living members of the order may have been evolved from 
creeping forms ; some species retaining that character, and others having 
changed to the upright habit, and back again. It is necessary to keep an 
open mind on such questions, and to let opinion follow fact and comparison. 
Applying this point of view to the present case, though we may 
contemplate the ultimate origin of the Gleicheniaceae from more primitive 
Ferns with radial upright axis, we see them to-day as characteristically 
a creeping family, with dorsiventral shoot. But if the facts indicate that 
any derivatives from them became in the further course of descent upright, 
and attained to a radial axis, there seems no reason to dissent from such 
a conclusion. On grounds especially of their sori reason has been found to 
contemplate the derivation of the Cyatheaceae from a source resembling 
the Gleicheniaceae. Our point now is, that if a careful examination of 
widespread character indicates that that was so, there is no inherent reason 
to doubt that the columnar upright axis, which is the most striking feature 
of the Cyatheoids, was itself a secondary character . 1 
The question will largely turn upon the existence of intermediate 
types. There are no relatively advanced Gleichenias which have adopted 
a definitely upright, self-supporting habit. The two most advanced species, 
viz. G. linearis and G. pectinata , are both creepers. But on the other side, 
though the more advanced genera Hemitelia and Cyathea show no creeping 
species, such are seen in the more primitive Alsophila , in the species 
A. ( Metaxya ) blechnoides , and in less pronounced degree in A. ( Lophosoria ) 
pruinata. Moreover, both of these proclaim themselves as primitive by 
other characters. A detailed account of Metaxya will be held over to 
a later memoir of this series. It may suffice here to say that with its 
creeping habit go a solenostelic structure of the axis, an undivided leaf- 
trace, and hairs only as dermal appendages ; all of these being relatively 
primitive features as compared with other Cyatheaceae. The same may be 
said for Lophosoria , which, however, as regards the mature plant has an 
upright, though short axis. But the buds which originate at the bases of 
many of the leaves give rise to runners, which at first take a horizontal 
course, with their leaves coming off at intervals right and left. The creep- 
ing habit with isolated leaves is thus actually shown in the early condition 
of these shoots. Thus, as regards position of the axis, Metaxya pro- 
nouncedly, and Lophosoria in a less degree, bridge over the gap between 
1 The question here considered, of the upright habit in the Cyatheoid Ferns being a secondary 
derivative from the creeping habit, has already been stated briefly (Ann. of Bot,, xxv, pp. 567-8). 
The detailed evidence for the opinion is here stated, and it is essential that it should be, for the theory 
of medullation stated in the paper above quoted could not hold, if the vertical character were primi- 
tive. But the comparative evidence from external form, from dermal appendages, from anatomy, 
and from sorus and sporangium alike, as stated in the above and in the succeeding paragraphs, 
indicates that it was not. 
X 
