347 
its Applicability to the Ferns , &c. 
It appears to me very unsafe to assume that two very similar forms are 
necessarily very nearly related when they live in places as to whose former 
geological connexion we have no evidence, and between which accidental 
transport is almost impossible. Other evidence than mere structural simi- 
larity is now required to prove close relationship. We must have geo- 
graphical affinity as well. A number of groups at present accepted as 
monophyletic will probably prove to be more or less polyphyletic, as is 
proving to be the case with the old and once apparently well-defined genus 
Acrostichum (2). 
Again, it is worth while pointing out that 4 age and area ’ gives strong 
evidence in favour of mutation. If we accept infinitesimal variation literally, 
it is, as pointed out in a recent paper (13, p. 202), easy to evolve anything, 
but four provisos, and they are all very large, must be made. Unfortunately 
for Natural Selection, none of them can be shown to hold good, and the 
result has been gradually to force its supporters, as illustrated in Mr. Ridleys 
recent paper (7), to abandon infinitesimal variation, and postulate for some- 
thing larger. They have not yet fully realized that by doing this they have 
really abandoned the essential point of their position. There is no evidence 
to show that a small — not infinitesimal — variation in one direction can be 
followed by others in the same direction. To take the examples given on 
p. 203 (13), can it for a moment be supposed that the formation of bulbils in 
Asplenium and many other plants began in easy stages, not to speak of 
infinitesimal ? Can one conceive of them as beginning as rudimentary bulbils ? 
Or can one expect to see a slightly dehiscent berry, a partially reversed leaf? 
How can intermediate stages exist between cauliflory and normal flowering, 
between pollen and pollinia (especially pollinia with translators — an essential 
family character, the only one really discriminating Asclepiadaceae from 
Apocynaceae), between sympodia and monopodia? If once really infini- 
tesimal variation is abandoned, there seems nothing for it but to take the 
plunge and admit that mutations of considerable size may occur. And if so, 
Natural Selection cannot be causative. 
But to return to the bearings of age and area on the question, it seems 
clear that it shows that endemic species confined to small areas are in reality, 
in the great majority of cases, new species commencing their career. At the 
commencement of their life, and for a long period afterwards, Natural 
Selection comes in, but chiefly as an agent to determine the non-survival of 
really disadvantageous mutations. How many of these occur we have no 
means of knowing ; those which survive are but few in number — only 902 
in New Zealand since very far back in time. These one may look upon as 
having successfully passed through the Natural Selection sieve, but, as I have 
already pointed out, a very small accident may easily kill out other com- 
mencing species which, if left uninterfered with, might also have passed 
through. A fire on a mountain-top in Ceylon might easily kill a local 
B b 
