12 Willis . — The Evolution of Species in Ceylon , 
In five cases (in this selection) the species has been observed in two 
localities a long way apart, which are joined by a wavy line. In all 
probability, as these lines go through the least thoroughly explored parts, 
the species occurs in localities between, and should really be regarded as 
R or even RR. 
The Rare species cover the map with a pattern of small interlacing 
circles, which overlap one another in the most complex way, but never 
coincide, and resemble the rings in a shirt of chain -mail. The RR species 
Fig. i. Distribution in Ceylon of the earlier VR, R, and RR species from Trimen’s Flora. 
show the same thing but with yet larger circles. The three diagrams may 
very well be looked upon as exhibiting three stages in the process of 
gradual spread of species, which commence as VR and go through the 
other stages in turn. 
The second point that shows at once in these diagrams is that the 
enormous majority of the endemic species are in the wet zone (which com- 
prises the south-western quarter of the island), as has already been pointed 
out (cf. Table XV of the Phil. Trans, paper). 
Now the important point about these distributions, from the Natural 
Selection point of view, is that nowhere do the areas occupied by the 
endemic species coincide, except in a few instances among the VR forms, 
where it often happens that two or three or more occur on the same 
mountain top. On Nillowekanda, for example, there are found, and there 
only, Acrotrenia lyratum , Stemonoporus reticulatus , and Ochna rufescens ; 
on Ritigala three species ; on Hinidunkanda three ; on Adam’s Peak ten, one 
