224 
H. F. Wernhatn. 
genera of these tribes are quite distinct from the rest of the family. 
A number of the Gardenieae section of Cinchonoideae, however, 
have only two ovules in each loculus of the ovary. 
On the lines adopted in these chapters, the progressive reduction 
in the number of megasporangia represents a definite evolutionary 
tendency, and, in accordance with these lines, we expect this 
reduction to be accompanied by evidence of other changes in the 
course of descent. The broad distinction between Cinchonoideae 
and Coffeoideae to which we have just alluded provides such evidence, 
the value of which must be measured by the amount and extent of 
that distinction. In the case of parallel genera like Webern and 
Pavetta it may be that an illustration is afforded of the working of 
the reduction-tendency within the limits of a single genus; and the 
merging of such a pair of genera into one would be quite as 
justifiable as the inclusion of hypogynous and epigynous forms 
within the single genus Saxifraga. The case seems to be but a 
particular example of a general proposition which is tacitly assumed 
through all our discussions of affinities, namely, that when a 
character represents a biological tendency at work in any particular 
group, it loses its critical value in the determination of affinities 
within that group. In Rubiales the tendency to reduction in ovule- 
number, and in Rosales the “ Calycifloral” tendency (chapter II), 
represent definite biological tendencies ; hence uniovulate and 
multiovulate forms may properly be included even in the same 
genus in Rubiaceae, and hypogynous and epigynous forms in equally 
close «onnection in Rosales. A looser and somewhat misleading 
method of expressing the same proposition is familiar to most 
systematists—that the value of any character in classification 
depends upon its constancy; we shall have need to refer to this at 
the conclusion of our discussions. 
We may summarize the position by the statement that the 
tendency to reduction in ovule-number to one per loculus is realized 
to a considerable degree in Rubiaceae; over 60% of the total number 
of species have one ovule only in each ovary-chamber. The progress 
of this tendency is traceable especially in the extensive tribe 
Gardenieae of Cinchonoideae, in which the ovules tend to be 
relatively few in number, and the seeds correspondingly large. The 
occurrence of ovaries with two ovules or even one only per loculus 
in forms having obvious affinities with the Cinchonoideae should 
not, therefore, excite surprise, although the practical value 
of ovule-number as a guide to the systematist is diminished 
