292 Bower. — Studies in the Phytogeny of the Fi lie ales. 
It will be shown that these characters indicate that Lophosoria is a more 
primitive type than the true species of Alsophila , and point with all the 
strength of parallel comparison to its position at the base of that series, 
where in point of fact it has been naturally placed by most systematists. 
On the other hand, a somewhat similar aloofness has been indicated 
within their own genus for two species of Gleichenia , viz. G> linearis and 
G.pectinata . This has been commonly recognized by writers. It has been 
shown that their peculiarities mark them off as an advance upon the genus 
Gleichenia in general. By further comparisons between this advanced 
guard of the ancient genus Gleichenia and the Lophosoria type as the most 
primitive of the Cyatheoid Ferns, it appears that the two large groups 
approach one another. In fact, on the basis of the foregoing observations, 
together with other considerations now to be advanced, the conclusion may 
be drawn that the primitive and creeping Gleicheniaceae resembled the 
ancestry of the Cyatheoid series, and that of living types they most nearly 
represent the forerunners of that great body of dendroid Ferns. 
The criteria which may be used for the purpose of this comparison are 
the following : — 
(i) The position and branching of the axis. 
(ii) The form and pinnation of the leaf. 
(iii) The dermal appendages. 
(iv) The characters of the vascular system. 
(v) The sorus, its position and constitution. 
(vi) The sporangia and spores. 
(vii) The prothallus and sexual organs. 
Each of these will now be examined with a view to an opinion on the 
phyletic relation of the Ferns named. 
i. In chapter xvi of the 4 Land Flora * I have dealt broadly with 
the question of symmetry in the sporophyte, and drawn the conclusion that 
£ the radial mode of construction was primitive for the sporophyte at large, 
and that where dorsiventrality occurs it is a secondary condition ’. If this 
were accepted in its baldest form, one might proceed to classify all radial 
types as relatively primitive, and all dorsiventral types as derivative. But 
that could only be done by making the quite gratuitous assumption that 
changes of symmetry could only occur once in any phyletic line. For this 
there is no warrant. It would seem probable rather that according to 
circumstances the change from radial to dorsiventral, and also from dorsi- 
ventral back to the radial condition of the shoot, may have occurred 
repeatedly in the course of descent. This subject has been already broached 
by Prantl in relation to the Hymenophyllaceae and the Schizaeaceae,, and 
similar suggestions have been advanced by Boodle as explaining the anatomy 
of certain Hymenophyllaceae (‘Annals of Bot.’, xiv, p. 482). He contem- 
