Gradual , Change , and Aj/ Guppy s Method of Differentiation. 615 
complete failure of the attempts to read adaptive values into the enormous 
majority of the diagnostic characters of living beings, that there is only 
rarely any handle for the operation of selection. Personally, I am inclined 
to think that change was generally effected by a single operation, but every 
compromise between the extremes is possible. But in view of the work 
done by Mr. Udny Yule and myself ( 19 ), showing that genera in their evolu¬ 
tion follow very closely the rule of compound interest, it seems enormously 
more probable that the changes were single steps. 
The current attitude of the Mendelians towards questions of evolution 
is one of an aggressive agnosticism. Since investigations upon Mendelian 
lines have not as yet been able to throw as much light upon the problem 
as had been at one time expected, they seem to think that no other line of 
attack upon the question will be any more likely to find a way that may 
possibly lead to something in the nature of a solution of the problem at some 
future date. They seem inclined to think that because they have not them¬ 
selves seen a ‘ large ’ mutation, such a thing cannot be possible. But such 
a mutation need only be an event of the most extraordinary rarity to 
provide the world with all the species that it has ever contained. As I have 
pointed out (‘ Age and Area ’, p. 212), one large and viable mutation upon any 
area of a few square yards of the surface of the earth, and once in perhaps 
fifty years, would probably suffice. 1 The chance of seeing such a mutation 
occur is practically nil , whilst if the result were subsequently found it would 
probably be called a relic. Darwin’s theory of Natural Selection has never 
had any proof except from a priori considerations, yet has been universally 
accepted, and has led to great advances in biology ; and until the Mendelians 
show us how to control mutation (a thing that will evidently be some day 
possible), the proposition now put forward will presumably go without actual 
demonstration by verified fact. What I contend is that the facts brought up 
here and elsewhere go to show that neither of the extreme suppositions— 
Special Creation and Natural Selection—contains all the truth, and that 
therefore this, or some similar, compromise between them is rendered 
necessary by the present condition of our knowledge. 
The small mutations that are all that the Mendelian school will allow 
are obviously in the highest degree unlikely to give rise to mutual inter¬ 
sterility, such as so commonly characterizes specific difference, and if they 
were to be accumulated it is difficult to see where the sterility would come 
in, for each would seem as likely to be fertile with its successor as with its 
predecessor— A with B , B with C, C with D, and so on. But let 
a big step, say from A to M, such as dropping of endosperm, be taken, 
1 Dr. Guppy has suggested that it is by no means unlikely that the many species once seen and 
never afterwards discoverable may often be such mutations. The case of Christisonia albida, 
described in Age and Area, p. 151, is almost certainly a case of a non-viable mutation, and it may 
be noted that Hooker, who was not a 1 splitter ’, accepted it as a Linnean species. 
