Gradual , Change , and by Guppy s Method of Differentiation. 609 
has probably very few adherents. Even its most strenuous supporters now 
usually begin with a mutation, though they may improve it afterwards by 
Natural Selection of fluctuating variation. 
Though by education an enthusiastic adherent of this latter theory, 
I found so many difficulties that were insoluble, and so often found simpler 
explanations, that I soon became a convert to Mutation (10, 2nd ed., p. 118 ; 
3rd ed., p. 208). It struck me at once, however, that there was a logical 
unsoundness in the new theory, for there was not the least evidence to 
suggest that a sudden change or mutation in one direction could be followed 
by another in the same direction, if the first were selected ; yet this was 
necessary if new structures of marked difference were to rise by small 
mutations, whether with or without Natural Selection. 
This supposition was largely based upon my investigation of the flora 
of Ritigala Mountain ( 12 ) in Ceylon in 1905, and in May 1907 (i. e. a year 
later than Guppy, who will be mentioned presently) I published a paper 
( 13 ) dealing with" some of the implications of the subject. This paper was 
chiefly concerned, however, with bringing up a trenchant argument against 
the natural selection of fluctuating or infinitesimal variations, an argument 
which has been used in the controversy on several occasions, and has never 
been refuted. 
It was shown that in the enormous majority of cases no use could even 
be suggested for the characters that mark specific or generic distinction, in 
spite of the great amount of work on adaptation. Also that, even if some 
use-value might conceivably be found in rare cases for the mature character, 
it could not be supposed to exist for the early stages of the same. The case 
of the weak stems that always accompany tendrils, and of the tendrils them¬ 
selves, which was mentioned above, affords an excellent illustration of the 
practical impossibility of intermediate stages. This case occurs in hundreds 
of different places in the system of the flowering plants, so that all cannot be 
supposed to have a common ancestor that acquired the climbing habit, and 
combination of tendrils with weak stem, once for all. 
The case was considered of Coleus elongatus , Trimen. This is a species 
peculiar to the summit of Ritigala, where it is accompanied by other pecu¬ 
liar species within a very limited area of not more than five acres. It was 
shown that there was not the least reason to suppose it (or them) to be of 
the nature of survivals. Consequently, even if the characters were ever use¬ 
ful or the reverse, they must have been so upon Ritigala summit, and 
nowhere else. This fact rendered unavailable the refuge to which the 
supporters of adaptation fly in such cases of difficulty—that the characters 
must have been useful somewhere else. Their other refuge—that they must 
have been useful at some time—was also rendered very precarious by the 
fact that geological and other evidence seemed to show that the conditions 
upon Ritigala had remained little altered since the Tertiary. 
S s 2 
