March, 1922] BENEDICT NEW VARIETIES OF NEPHROLEPIS 
are scarcely even a matter for conjecture, the evidence dealt with here being 
entirely macroscopic. 
The study of regressive variation offers one distinct advantage over that 
of progressive variation in the fact that the former is sufficiently common to 
be frequently observed, and may thus be studied experimentally. This 
suggests the possibility of determining an external cause for reversion. So 
far no definite experiments have been carried on, but there are some sug- 
gestive facts connected with the possible relations between cultural condi- 
tions and the occurrence of some of the reversions. 
As with the first paper, the data for the present article have been obtained 
in two ways: first, by the further extension and study of the collection of 
Nephrolepis at the Brooklyn Botanic Garden. Some of the reversions here 
described have been developed at the Garden under observation. Second, 
visits to florists have been continued for the opportunity of seeing hundreds 
of thousands of plants. During 1916 these visits \^/ere made with the aid 
of a grant from the American Association for the Advancement of Science. 
A second similar grant was made in 191 7 and has served in the preparation 
of this paper. Acknowledgment is also made to the Bureau of Plant In- 
dustry, specifically to Messrs. Peter Bisset, Wilson Popenoe, and David 
Fairchild, for assistance in the collection of wild forms of Nephrolepis from 
various parts of the tropics. 
Special acknowledgment is again tendered the Brooklyn Botanic Garden 
for the very satisfactory facilities for this study which have been enjoyed. 
Description of the Regressive Varieties 
It will forward the description of the regressive variations to state here 
one of the conclusions based on the detailed observation of all the forms: 
namely, regressive variations may be expected to occur in any of the pro- 
gressively developed varieties. This statement has not been checked for 
all the one hundred varieties known to exist, nor even for the forty listed 
in the genealogical chart attached herewith, but it has been found true for 
so many of them that it is a reasonable assumption that it may be charac- 
teristic of all. 
This multiple origin of reversionary forms makes it impossible to arrange 
their descriptions in as simple a system as was used for the progressive 
varieties. With the latter it was merely a case of arranging them in a 
genealogical table, and describing for each the changes or advances in form 
it manifested. Each new kind in a series was produced by the intensification 
of some character of the parent type, or by the appearance of some other 
ot a few well marked characters. Each reversion represents similarly some 
definite divergence from its parent form, but instead of one common an- 
cestral type there are forty to one hundred possible ancestors. The rever- 
sion of any given variety is thus a distinct problem for study and descrip- 
tion, and is related to other reversions only as it may produce analogous or 
