14 Chrysler .— The Nature of the Fertile Spike in 
between B. virginianum and Osmunda in this respect is so striking that, 
when considered in connexion with the other filicinean characters of Botry- 
chium, we must conclude either that this genus has been derived from fern 
ancestors, or that ferns have been derived from Ophioglossaceae. The latter 
alternative is adopted by Campbell (8), but is not countenanced by Bower. 
Signs of high specialization are apparent in the family, which accordingly is 
not a favourable point of departure for other groups. Without attempting 
to discuss in detail the evidence adduced by Bower for linking Ophioglos¬ 
saceae with the Lycopods, certain filicinean features of the former group may 
here be mentioned, in addition to the structure of the petiole :— 
1. The Ophioglossaceae belong distinctly to the large-leaved group. 
Although it is not impossible that large leaves may have arisen more than 
once in evolution, it is hard to believe that so great similarity of structure 
should have been attained independently. 
2. The endarch xylem (mesarch in Helminthostachys ) is in sharp dis¬ 
tinction to the exarch condition seen in most Lycopods. Since Tmesipteris 
has mesarch xylem this criterion is not so reliable as others. 
3. The central cylinder of Ophioglossaceae is hollow and is perforated 
by leaf gaps. This fact places the family in the group Pteropsida. Bower’s 
effort to show that a phyllosiphonic stele occurs also in Tmesipteris (6, 
p. 487) has been shown by Jeffrey ( 15 ) to be a misconception. Ophiogloss 7 mi 
cannot be linked with Lycopodium by means of Tmesipteris or Phyllo- 
glossum y which are both outspoken members of Lycopsida. 
4. Those who have studied the gametophytes of the three genera of 
Ophioglossaceae are agreed in considering their characters unmistakably 
filicinean, though specialized in connexion with the subterranean habit. 
The multiciliate sperm merits special mention. 
In view of the evidence in favour of the filicinean affinities of the family 
it is appropriate to inquire which of the ferns are most closely related. 
When one considers the isolation of the family as a whole, and the differ¬ 
ences between the genera, especially shown in the monotypic Helmintho¬ 
stachys, it becomes apparent that the family diverged from the main 
filicinean stock at a very remote date. The large sporangia opening 
simultaneously and lacking an annulus, the eusporangiate habit, the usually 
vertical rhizome, the gap in the leaf-trace formed at the point of exit of the 
strands which supply pinnae, indicate that the affinities of the family are 
with the group Simplices, especially with Osmundaceae and Marattiaceae. 
More than this cannot be safely inferred from the data at present available. 
The evidence brought forward in the present study supports the earlier 
view that this family shows a series of reduced forms. Starting with a fern 
whose leaves bore sporangia over the whole of the lower surface, there has 
apparently been a sterilization of certain leaflets, similar to what may be 
seen in Osmunda Claytoniana , except that in the present case it is the basal 
