Sinnott . — The Evolution of the Filicinean Leaf -trace. 175 
and T. radicans , which possess much stouter steles than does T. reniforme 
or any of the Hymenophyllums, but which resemble them in having central 
protoxylem, the leaf-trace and the petiolar bundle are distinctly triangular 
in shape and always have more than two protoxylem groups. In T. Prieurii 
there are three of them in the trace as it passes through the cortex, but 
from the division of the median group there are four in the petiole. In 
T. radicans the protoxylems are more numerous, but tend to be restricted 
to the three angles of the trace, at least while it is in the cortex. 
In the stoutest members of the family, represented by T. scandens and 
T. apUfolium , the cylinder is a solid exarch protostele, much more com- 
parable than have been any of the others to the condition found in Lygodium 
and the Gleicheniaceae. This similarity is also clearly apparent in the 
petiolar bundle, which is arched and possesses three protoxylems. The 
incurving of the ends of the arch in T. apUfolium suggests Aneimia. The 
leaf-trace in the cortex was observed by Boodle only in T. scandens , where 
he notes the important fact that here and at the node it possesses con- 
siderable peripheral or exarch protoxylem which is continuous with the 
exarch protoxylem of the stele. 
There would thus seem to be a rather complete series, both in the 
anatomy of the stele and of the leaf-trace, between the simplest and the 
most complex species of the family. As to their inter-relationships, the view 
which has the widest acceptance, and which is supported by Boodle, Bower, 
and Tansley, considers the type represented by //. dilatatum and T. reni- 
forme to be the most primitive, and the smaller forms to be derived from 
this by reduction, the larger by amplification. The fact that such species 
as T. scandens approach so closely the protostelic Simplices in the structure 
of the stele, and more especially of the leaf-trace, as well as in certain 
sporangial characters, makes it seem very probable that there must be 
a phylogenetic relationship between the two groups. If this close affinity 
is acknowledged, however, the Hymenophyllaceae are apparently either 
very primitive ferns, from which the triarch Simplices have been derived, 
or else they represent a reduction series from them, beginning with 
T. scandens as the least modified and closest to the ancestral type, and 
ending with the small collateral species of Hymenophyllum. The latter 
view seems to the writer much more worthy of acceptance, both on account 
of the filmy-leaved character of the family, which is clearly a reduced con- 
dition, and from the fact that the sorus shows a basipetal development, 
putting these ferns among the Gradatae, instead of among the Simplices, 
where they should certainly be placed were they an extremely primitive 
group. The resemblance of the family to the Botryopterideae seems, to be 
of little significance. 
The bulk of the remaining ferns in the group Gradatae are comprised 
in the Loxsoma-Dennstaedtia-Dicksonia alliance, the anatomy of many 
