KEYS TO THE FERNS OF BORNEO 2gi 
this family, I have attempted a natural classification, and 
in consequence am using an arrangement which demands, 
and I hope merits, explanation. 
It is a principle which must be accepted in systematic 
botany that each group we recognize, whatever its rank- 
species, genus, tribe, family or order,- is an expression of 
our ideas of real relationship, that is, of phylogeny. If 
this principle be accepted, definitions and convenience of 
definition cease to be controlling factors in determining 
what plants should be included in each group; although 
convenience of definition of course remains a dominant 
consideration in determining how large our genera should 
be made. For example, Davallia has usually been defined 
or diagnosed in such a way as to include Prosaptia. There 
was a time when this was a valid reason for including 
Prosaptia under Davallia. There is absolutely no doubt 
that Prosaptia is a group descended from Eupolypodium , 
and this is a justification for the practice, adopted by a few 
writers, of including it in Polypodium. At this point, the 
question of convenience comes up, and it seems to me 
decidedly more convenient to treat Prosaptia as a distinct 
genus closely related to Polypodium, and to define it 
completely enough to avoid possible confusion with 
Davallia. Again, we have had in Polypodium a sub-genus 
Goniophlebimn, recognized by definition, and sometimes 
treated as a genus. When so treated, it has, invariably, 
I believe, been made to include a considerable number of 
American species, along with the Oriental species to which 
the name was originally given. These American species 
represent two or more lines of descent, independent of 
that of the Oriental Goniophlebium, having a clearly 
distinct ancestry outside the group. Logically, it is clear 
that we have but two alternatives, either to include all 
of these lines of descent, back to a common source, in 
