I—INTRODUCTION. 1g 
sidiary use is here made as an alternative aid in the determination 
of the genera, is employed, so far as it is used at all, in an absolute 
fashion. The Se that accompany its employment in 
botanical works of the early portion of last century have been 
put aside. In these densities, as here, the primary subdivision is 
dependent on the number of stamens in the flower. In many 
obviously natural genera, however, and for that matter in not a 
few species, the number of stamens is variable. Sometimes this 
variation in number is the result of what we may term a natural 
accident, as where, among the smaller and definite numbers, the 
uniseriate stamens in the flowers of the same plant are found to 
run from 3-5, 7-10, and the like. More often the variation has an 
obvious structural explanation, as where the stamens in a species 
may be 4 or 8, or 5, 10, or 15, according to whether only one, or 
ore than one series of stamens becomes developed. This type 
of variation, occasional in species, is much more frequent within 
the limits of a genus, where, too, another type of variation, rare in 
individual species, is not uncommon. This is the type that leads 
to the number of stamens being 4 or 5, 8 or 10, and the like; one 
species and often a whole section of a genus having tetramerous 
flowers, whee another type and section may have the flowers 
pentamerou 
In works sinacs the artificial system is the only one employed— 
Roxburgh’s Flora Indica, of which those who are likely to use 
this work will probably become possessed, is an excellent example 
bility of these two objects is sufficiently obvious. No system of 
arrangement could be satisfactory that resulted in the treatment 
of the same natural genus in more places than one. A decision 
had therefore to be arrived at, in cases where a genus includes 
some species with 4 stamens and others with 5, whether the genus 
as a whole should be placed in the class Tetrandria or the class 
ge a whatever in the compromises thus called for made 
for efficiency in the arrangement of the ere correspondingly 
detracted soe the system as an instrument for the determination 
of their species. Here we are hampered na no such necessity ; 
the basis of our arrangement is derived from an independent 
source—the Genera Plantarum of Bentham and Hooker adopted 
