16 MR. F. N. WILLIAMS ON THE GENUS SILENE. 
one of the sections of the genus Lychnis proposed by Fenzl in 
Endlicher’s monumental work. . 
The limits of the genus will be more conveniently circum- 
scribed if Rohrbach’s generic character of a uniloéular capsule 
septate at the base is maintained ; so that many of the North 
American species described by Sereno Watson and most of the 
species from the province of Yun-nan, described by M. Franchet,* 
in which the ovary and capsule are strictly unilocular, should 
be transferred to the genus Melandryum. It would be better 
also to exclude four species, retained by Rohrbach in his 
monograph, in which the ovary has five styles, and is, of course, 
5-septate at the base, viz., Polyschemone nivalis, Schott, Hudi- 
anthe Celi-rosa, Fenzl, EH. corsica, Fenzl, and E. leta, Fenzl. 
Husilene itself is divided into three sections according to the 
form of the inflorescence. In the same way that the principal 
primary subdivisions in the grouping of natural genera within 
the limits of the same family cannot be carried out with logical 
precision, in so far as a character in one genus is of first 
importance, and in another may be of no systematic value 
whatever, so the same thing obtains in the subdivision of a 
genus into several groups of higher or lower grade. Thus it 
would be unnatural to separate S. grisea from S. flavescens ; yet, 
strictly speaking, the first should be in Botryosilene, while 
S. flavescens is in its place in Dichasiosilene. Similarly the 
same reason holds with 8. Steberi, S. Fenzlit, and certain forms 
of S. ctalica, which should be in the section Dichastosilene, were 
it not that their special characters unmistakably show their 
relationship with S. ctalica. We must, therefore, take into 
consideration other characteristics in admitting apparently 
aberrant types into one or the other of the primary groups in 
the subdivision of the genus. Rohrbach’s greatest difficulty 
was in the section Botryosilene. The peculiarity in certain 
species for the flowers to bend downwards at the time of flower- 
ing, an observed fact utilized by Otth in his grouping of species, 
when applied to the section as a whole, leads to unnatural 
separation of closely allied forms. In the same way it is 
impossible correctly to circumscribe a natural series of forms 
by means of the various modifications of the raceme, since 
frequently one type of racemose inflorescence grades into 
another, and also species having varied modifications of the 
raceme are allied by more distinctive and important characters. , 
* Bull. Soc. Bot. France, xxxiii (1886), pp. 417-428. 
