4 WILLIS : EVIDENCE AGAINST THE ORIGIN OF 



it to be an adaptation in this case, but a structure which has 

 been forced upon the flowers either by direct mutation or 

 by correlation. 



My own work in botany was formerly to a large extent the 

 study of adaptations, in which, as a student from 1886 to 

 1890, I was trained to believe, and, in general, the result was 

 that I found some other suitable explanation of a much less 

 fanciful kind. For instance, gynodioecism, formerly explained 

 as an adaptation to cross-fertilization, turned out to be a very 

 variable and to some extent non -hereditary character, and I 

 was entirely unable to regard it as an adaptation, its value to 

 the plant being extremely problematical. Or, again, the case 

 of the dorsiventral flowers of the wind-pollinated Podoste- 

 macese, above described, may be quoted, or the production of 

 cleistogamic flowers.* Apart from the great difficulty (in 

 many cases of adaptation or supposed adaptation) of imagining 

 of what use the first stages, which are necessary on the theory 

 of selection of infinitesimal variations, can be, a simpler and 

 more reasonable explanation can in very many cases be found. 



The general line of argument so far followed has nearly 

 always been answered by the not very good reply (were it not 

 that the theory of infinitesimal variations holds the field) that 

 we do not know where the species originated, and that the 

 characters in question may have been useful somewhere, or at 

 some time. Most British species, for example, range over a 

 vast area in Europe and Asia, and one cannot therefore feel 

 sure that this reply is not good. And it is to this reply that 

 the evidence of local endemic species seems to afford a final 

 rejoinder. 



Let us begin with the case of Ritigala. There are several 

 species on the summit, which are confined to it, i.e., to an area 

 of at most 2-5 acres. Now, as Ritigala is not peculiar among 

 the Ceylon mountains in height or other qualities, and as 

 geological evidence seems to show that its elevation has not 



* Willis, Linn. Soc. Journ. 30, 1893, p. 295. 



