62 
■\vcre a single character b}- which all fern? coxild be divided into great 
and natural groups, or if any character were of equal value in all groups. 
The fixedness of any character depends partly on grounds we understand, 
or certainly will understand with sufficient study along familiar lines — 
that is, on the relation of the character in question to the environment 
of the plant or group in question — and in part, or ground we can name, 
but do not yet know how to investigate; that is, the different heredity- 
strength among variations (including mutations), and characters already 
called hereditary. 
While the applicability of no character is universal, neither are there 
uniformly fi?ed limits to the value of any character. The margin of the 
frond is unstable in many or most genera, yet entire margins characterize 
the simple fronds of Vittarica^ and genera in various other tribes, and 
a simple and entire Lastrceaj or Atlujrhim, or MicroJepia, or Schizostege 
is unknown. Even geographical characters are useful. The whole of 
the character of not a few genera, as Prosaptla, Niphoholui;, and Acliro- 
stichum, is intelligible when, and only when, the habitat is included and 
recognized as the dominant character of all. More broadly geographical 
characters are of value too, for no plant has progeny in places inacces- 
sible to its reproductive structures. The relative antiquity of groups, 
as definable by their present characters, is important evidence in judging 
their relationships. If a species or genus is confined to one locality or 
ono part of the world, it is probably not very ancient. If it has a very 
wide and continuous distribution, its age can not be less than sufficient 
to permit such a dissemination. Our oriental Prosaptia, Acrosorus, 
Loxogramme, '' Schellolepis/' Nipliololus, Drynaria, Bnjostachyiun, 
Thayet-ia and Lecanopteris, and the American Lepicystis, Campyloneuwn 
and Phlehodium must all, be younger groups than the cosmopolitan 
Polypodium, ancestor and cousin at once of them all. A group with 
wide and discontinuous distribution must be ancient enough to have 
become widely distributed, and to have died out in the intermediate ter- 
ritory; it may not be older than a group with equally wide continuous 
distribution, but its minimum probable age is greater. 
I have tried to become so well acquainted with the San Eamon ferns 
that I might know, each species and larger group as the sum of its 
characters and not by any single character; so that I might fairly Judge 
m each of the stability and diagnostic value of each character, and 
recognize the more elusive peculiarities, as well as those which lend 
themselves readily to description. Success in such an attempt is at 
best a matter of degree, but the prol^able degree is very much greater in 
the field, especially in the field with some laboratory equipment, than it 
is in the herbarium. 
I have also tried to decide what might fairly be regarded as a primitive 
form of Polypodiaceous fern. Suclv a fern must be a very generalized 
type, not highly specialized in adaptation to any conditions; it should 
