446 MR. H. B. GUPPY : PLANT-DISTRIBUTION 
inference that the primitive genera * must have commenced when there was 
still a possibility of reaching both the great land masses" (Phil. Trans. 
ibid. p. 338). This recalls Bentham’s opinion respecting the Composite, 
that the two hemispheres, the east and the west, must have possessed the 
family at its earliest recognizable stage. 
So, again, when he observes in connection with the Podostemace and 
Tristichaces that “ the only widespread genera are the non-specialised ones, 
whilst the more specialised the genus, on the whole, the less is its area of 
distribution " (Proe. Roy. Soc. 1914, p. 545), and when he remarks that 
“the larger the group and the fewer the characters on which it is based, the 
greater the likelihood of its being polyphyletic "—that is to say, of arising 
independently in localities remote from each other (Ann. Roy. Bot. Gard. 
Perad. p. 447, 1902),—he is expressing the differentiation standpoint. But 
_ the writer cannot conceive how the principle, often implied in the foregoing 
remarks, that the simpler the form the wider its range, could apply to groups 
that have been built up, as Dr. Willis infers, by the species taking the 
generic step, the genus the tribal step, and the tribe the family step (idid.). 
But, apart from this, there is a great deal that links together the two 
views of distribution; yet one may add that whatever view we take of 
distribution, whether that of the Darwinian evolutionist, or that of the 
Mutationist, or that of the advocate of pure differentiation, we all get into 
the same dilemma when we handle the larger groups. If we require for 
their development the mutability or instability of the characters on which 
the taxonomist bases his larger groups, characters that in our own age are 
relatively immutable, we cannot look to existing prevailing conditions for 
guidance in the matter. However, Dr. Willis in his account of the 
astounding modifications experienced by the Podostemacew, under conditions 
described by him as “unique” among plants, offers, as the writer has 
already explained, a way out of the difficulty. 
The statistical treatment of Distribution.—-lf the Differentiation hypothesis 
is valid, we may now ask what we should be justified in expecting from a 
statistical treatment of the main features of plant-distribution. If we listen 
to the story of the early stages in the distribution of the Composite and 
Gentianaces, as interpreted by Bentham and Huxley, we should expect that 
the larger plant-groups would to a great extent ignore the cleavage of the 
land of the globe into two large masses diverging from the north, and that 
the response made to the existing arrangement of land and sea would 
increase as we go down the differentiating scale, being least for the family 
and greatest for the species. On the other hand, we should expect a 
marked response of the larger plant-groups to the climatic differentiation of 
the latitudinal zones. 
If in our investigations with the family as our starting-point we disclose a 
method and a system that could not be brought about by a procedure so 
haphazard as that involved in commencing the genetie sequence with the 
