448 BENEDICT: ANTROPHYUM 
however, approach this condition. A still more sharply distinctive 
character, if A. exsiforme be excluded, is the absence of paraphyses, 
which are found in all the Old World species. 
On the basis of this character together with spore-form, A. 
ensiforme Hooker with diplanate spores may be separated from the 
others which lack paraphyses and have triplanate spores. Moore's 
genus Scoliosorus was based on this species, but on the strength of 
a non-existing character inferred from Hooker's incomplete des- 
cription. The characters here given, however, and a slight but 
decided difference in soriation are sufficient to separate it at least 
as a subgenus. 
Using soriation as a basis, Desvaux in 1827 established a new 
genus, Polytaenium, with A. lineatum as type. The sporangia in 
this species are in three or four long straight deep grooves on each 
side of the costa, and parallel to it, an arrangement considerably 
different from the normal American type which consists of series 
of divergent branching lines, superficial or only slightly immersed. 
In some of the broadest forms of JA. lineatum, however, the type 
of soriation approaches the normal, and on the whole, the similar- 
ities seem greater than the differences, which are satisfied by recog- 
nition as a subgenus. 
The remainder of the species belong to the costate section of 
Euantrophyum. After we eliminate A. lanceolatum, which is easily 
distinguishable, six species remain, two of which are described 
here for the first time, and one other which has never been accepted. 
These three are based on material originally identified as 
A. subsessile Kunze (A. brasilianum (Desv.) C. Chr.), but such ee 
ference is not justified. The six species form a group of coordin- 
ate forms, no one of which is entitled to stand as representative of 
the others. If A. discoideum, for example, be refused recognition, 
then logically 4A. cayennense which has always been recognized 
must also be referred to A. drasilianum. The fact that in the most 
superficial and easily determined character, outline, A. discoideum 
resembles A. drasilianum more closely, is not a sufficient reason 
for denying its validity. In reality it is less closely related to the 
latter fern than is 4. cayennense, which shows its affinity in a Cora 
ceous texture and closed marginal areolae, and sometimes even 1? 
ee 
