CL VII. GRAMINEAL (Stapf). 
3 
or dioecious). Mature spikelets falling entire from the tips of the 
pedicels, or together with a part of the pedicel or of the rhachis, or 
breaking up above the glumes into as many false fruits as there are 
fruiting florets, rarely persistent and shedding the grains. In the 
first case the glumes, in the second the valves are often decurrent 
into a callous swelling or extension (callus) at the insertion on the 
pedicel or rhachilla respectively and separate together with it. 
About 500 genera, comprising 3000 to 3500 species in all parts of the world. 
The typical structure of the spikelets is sometimes more or less obscured by 
the reduction or suppression, or by peculiar modifications of certain parts, 
generally in obvious correlation to the loss of functions, or the assumption of 
functions other than usual. The morphological character of those parts may, 
however, usually be recognised from their position in the spikelet and from 
comparison with allied, less modified species. Reduction or suppression is 
frequent in the glumes and valvules, but extremely rare in the valves, except 
in reduced florets. The lower valves are, in certain tribes, frequently without 
a flower ; but they very often enclose a rudiment of a floral branchlet, in the 
shape of a perfect or reduced valvule, thereby indicating their homology with 
typical valves. In this case they often lose some of the characteristics of the 
fertile valves, and approach either to the glumes in their general structure, or 
assume some special structure differing from that of the glumes as well as from 
that of the fertile valves. The nervation of the valve is very constant in nearly 
all the genera, and often throughout the greater part of a tribe ; but in order 
to see it clearly, it is always advisable to flatten the valve and examine it by 
transmitted light in a drop of water. Where the sexual conditions of the florets 
are of importance, it should be kept in mind, that many grasses are very dis- 
tinctly protandrous, i.e. that they shed their anthers some time before the 
stigmas expand. Such flowers have frequently been taken to be female, whilst 
they were actually hermaphrodite. To avoid this error, young spikelets should, 
if possible, be examined besides the fully developed ones, or the filaments which 
usually remain around the ovary should be sought for. 
The definition and sequence of the tribes followed here are the same as in the 
Flora Capensis. The definition of the genera of the second series ( Pooidece ) 
differs on the whole very little from that adopted in the Flora Capensis ; in the 
Panicoid series, however, a considerable change has been found necessary. It 
concerns mainly the genera Andropogon and Panicum and their allies. The 
necessity of it was foreshadowed in the Flora Capensis (vol. vii. pp. 334, 383). 
To carry out the change then was not feasible as any revision of genera has to 
take into consideration the whole of their species and those of the immediately 
allied groups, a task for which there was no time. Since the publication of the 
Graminece of South Africa much work has been done in that direction, particu- 
larly in the tribe of Panicece by American authors, whilst at the same time the 
collections at Kew have been subjected to a special scrutiny irrespective of 
floral boundaries. The absence of definite dividing lines for the genera of 
Andropogonece and Panicece has so often been demonstrated that it is unneces- 
sary to insist upon that fact. There being none, the alternatives are to unite 
the groups wherever intermediate links can be detected, or to be satisfied with 
approximately definable groups which can on the whole be easily grasped and 
remembered. The first must lead to chaos unless the species is consistently 
quoted with the name of the subordinate group (subgenus or section) to which 
it belongs, a process which is too cumbersome to be generally adopted, whilst 
the other alternative has every practical advantage, except that of introducing 
a considerable number of new combinations. These innovations are, however, 
more apparent than real, for there is no serious difference between such terms 
as “ Panicum ( Setaria ) viride ” and “ Setaria viridis ,” or “ Andropogon (Cymbo- 
pogon) Schcenanthus ” and “ Cymbopogon SchcenanthusP It has therefore been 
