328 BOTANICAL GAZETTE [May 
grouped with the Marsiliacee and Salviniacee, chiefly on the 
grounds of their heterospory and hydrophytic habit. DeCan- 
dolle was the first to suggest a connection with Lycopodium. In 
this view he was followed by Brogniart, Endlicher, Hofmeister, 
and the later German botanists. A summary of the various 
relationships which have been assigned to Isoetes was given in 
1888 by Vines (1),who in the same article put forth the opinion 
that its affinities are with the eusporangiate ferns, rather than 
with the Lycopodiales: More recently Farmer (1) and Camp- 
bell (4) have expressed their concurrence with this disposition 
of the genus. : 
Since this classification has been retained by Vines in his 
Text-Book of Botany, and adopted by Campbell in his Mosses and 
ferns, it will not be unprofitable to re-examine the evidence, 
with the purpose of seeing what light can be thrown upon the 
subject by the present and other recent investigations. 
In any discussion of relationships, and especially when there 
is so great diversity of opinion as in the present case, the con- 
clusion is likely to be a personal one merely, dependent on the 
kind of evidence which the examiner holds most weighty, rather 
than on its absolute nature. There are some general principles, 
however, to which everyone will probably assent, and which 
ought to govern one in estimating the relative value of the con- 
flicting evidence on which the taxonomist relies. In the first 
place, the larger the number of characters in which there !s 
agreement, the closer is the relationship, especially if the charac- 
ters are such as are known to have great taxonomic value in 
groups related to the one under consideration. Of single 
characters, those which are most constant are of most value, 
even though we are not able to detect their special utility. It es 
generally accepted, too, that those characters which appear bs 
the embryonic stages of an organism serve best to mark 1s 
wider relationships, as of class or family, while characters which 
do not display themselves till later in the individual life ae 
better adapted to distinguish the near relationships of sperits 
and genus. This principle applies not merely to the organism 
