33 
illustrated by the South African Flora. 
with the general study of the phytogeny and inter-relationships of the 
Angiosperms, as has been demonstrated by Small in connexion with the 
Compositae (32). Morphologists and systematists are far from agreed on 
which characters in the flower are to be considered really primitive. If 
widespread species or genera or families have given rise to types with more 
restricted distribution and this is found to hold generally, the deductions 
are obvious. The hypothesis of course requires much testing, but it may 
serve to throw light on the question of how evolution has progressed and 
whether it has been mainly from the simple to the complex or vice versa. 
A comparison of some of the examples quoted in the general survey above, 
from this standpoint, would yield interesting results. Take, for instance, 
the genus Eragrostis among the grasses. Eragrostis curvula differs from 
the numerous endemic species which can be grouped round it chiefly in 
having longer valves. A shortening of the valves would appear, therefore, 
to be one of the evolutionary tendencies in this genus. Again, if we take 
the Gramineae as a whole, it is remarkable that the south-western genera — 
the older types — belong mostly to tribes (e. g. Aveneae, Festuceae) which 
have numerous florets in the spikelet, often much exserted from the glumes, 
while the presumably more recent invasion of grasses into South Africa, as 
represented by the eastern grass-veld genera, consists of types which, as 
a rule, have only one perfect floret in the spikelet (e. g. Andropogoneae, 
Paniceae). Similar facts have been briefly touched upon in respect to other 
families, but further work is necessary from this standpoint, as from the 
others. From all standpoints, the South African flora affords excellent 
materials and opportunities for the study of plant distribution. 
Summary. 
i. Of those who have dealt with plant distribution, Darwin and Wallace 
objected to invoking geographical change as a solution of every difficulty, 
while Hooker was more inclined to postulate continental extensions to explain 
the connexions between the floras of the southern hemisphere. Schonland, 
after carefully considering the facts, finds a land connexion between South 
Africa and Australia and between West Africa and America to afford the 
simplest explanation. Thiselton-Dyer and Guppy agree on a theory of south- 
ward migration from the northern hemisphere. Guppy also elaborates a theory 
of differentiation, which he applies to families, tribes, genera, and sections of 
genera. These each tend to fall into two groups, primitive and derivative, the 
first widespread, the second restricted in area. Willis has developed the ‘ age 
and area * theory based on the fact that endemic species in any area tend to 
have a narrow range, non-endemic species a wider. The older a species is, 
the wider its range. Engler, Drude, and Clements agree that ‘ multiple 
origins * are possible, the former admitting the possibility chiefly in the case 
of larger groups, but Clements applying it also to species. The writer in 
D 
