200 
Willis.—Age and Area. 
good ; but obviously they rarely exist in any marked degree, or the figures 
for distribution would not come out in such definite agreement with those 
that would be expected were Age and Area alone operative. Under the 
latter supposition we regard most species as not yet having necessarily 
reached their limit possible of dispersal. 
Under Age and Area species are supposed to owe their differences in 
distribution not primarily to differences between themselves , but to 
differences in the time at which they were first evolved, or first appeared 
in the country under consideration, and secondarily to differences in the 
average rate at which they are dispersed. Into this latter factor there, of 
course, enter all those differences due to dissimilarities between the species 
themselves. Trees, for example, will spread at a slower average rate than 
herbs, Cruciferae probably more slowly than Compositae, and so on. By 
taking the species always in groups of ten, and comparing these with other 
tens allied to the first, chance differences between them will have their 
effect reduced to a minimum. 
It will be seen that this method of looking at species and their dis¬ 
tribution substitutes for the essentially ‘vital’ character of distribution 
under Natural Selection a much more ‘ mechanical’ conception, that species 
owe their distribution primarily to their age, a distribution checked only by 
barriers, mechanical or ecological. On the old theory species were regarded 
in general as having reached their limit of possible distribution ; on the new 
they are regarded as usually in process of expanding their areas, and 
generally with extreme slowness. On the old theory species of very 
limited area were looked upon as owing that smallness to the competition 
of other more successful types, and therefore as the failures in the struggle 
for existence; on the new, they usually owe it to the circumstance that 
they are still comparatively young, and have not had time to occupy 
larger areas. 
Now the relative values of these two theories can be quickly tested by 
putting them to the proof, and using them to make predictions as to dis¬ 
tribution. If only one of them can be thus used, and at the same time 
proves to be successful in its predictions, then it is clear that the evidence 
in favour of that theory becomes very strong. This is exactly what 
happens in the present case. No one has ever suggested that Natural 
Selection can be used for prediction, whereas Age and Area has already 
been used over a hundred times with success. So easy is it to make pre¬ 
dictions, and so regularly are they successful, that it is almost literally true 
to say that one may at the present time ‘ shovel ’ in results in the study of 
distribution. The criticism that has been directed against Age and Area 
is of the arm-chair a priori type; but if the critics would take the trouble 
to try the methods for themselves, I feel certain that they would not long 
decry them. In the present paper I make some predictions of larger range 
