SELECTION ON THE VARIABILITY AND CORRELATION OF ORGANS. 
59 
some degenerate offshoot of a race superior to themselves than a sample of the 
primitive people from whicli the Circassian races may be supposed to have sprung. 
I'lie whole of this discussion is, of course, very hypothetical; no stress whatever is 
to he laid upon it except as an illustration of method, and a rough ajDpreciation of the 
vast amount of elimination which must he necessary to evolve one race from a second 
in tlm case of organs which we know by measurement to have continuity of variation, 
and only saltatory changes in pathological cases, which have, as far as we can judge, 
no influence on the mass-evolution which has produced the local races of mai]. 
But given fair samples of material our method will enable us to determine whether 
a race A—for of course a limited number of characters—could with less destruction be 
deduced from a race B, than the race B from A. It will not therefore follow that the 
path of least selection is that which necessarily was used by Nature. Possibly both 
A and B have been reached by far less expenditure of material from C. Still it is 
something definite in the midst of our gropings after truth in problems of descent to 
have even a rough appreciation of the amount of selective destruction wliich would 
arise from alternative suo-o-estions. That is why this special numerical illustration of 
0& 
the surface of survival has been given. 
The reader will possibly find the matter rendered somewhat clearer by the diagram. 
The femur is measured alon!>' the horizontal and the humerus alontj the vertical. 
A is the type or mean femur-humerus of the Aino population. ^yithin in the 
continuous ellipse rornid A the whole Aino popidation up to 1 in 1000 would fall. 
F is the type of the French population and the continuous ellipse round F gives the 
area within which up to 1 in 1000 of the French population fall. Since the diagram is 
drawn to centimetres of the bones, it will be seeji liow very small are the limits of 
variation within both populations. P is the ceiitre of the surface of survivals ; for the 
selection of Aino from French it makes the “ fittest to survive.” In the case of the 
selection of the French from the Aino, P is no longer the centre of fitness, but the 
“ centre of unfitness” ; the Aino are killed ofi‘ with an intensity which increases the 
closer we approach to P. Now it seems to me that these two cases, which are cjuite 
distinct in theory, ought to manifest themselves in Nature and require distinguishing 
names. A race may be modified because a complex of organs with a certain system 
of values is good for it, or because it is bad for it. The race may be modified because 
a certain element of it is fittest or because it is unfittest to survive. In the former 
case we select for survival round the centre, in the latter case we select for destruction. 
I propose to call these cases -positiv(' and negat-ive selection respectively. It may he said 
that if there be positive selection in one part of the population there will he negative 
in another. But the kernel of the matter is in either case the existence of a centre, 
a definite set of most fit or of most unfit organs, while in positive selection the less fit 
organs, and in negative selection the more fit organs are distributed over wide areas 
of the field, and do )iot reach a maximum of unfitness or a maximum of fitness 
respectively for any definite individual. 
