404 MR. G. J. ROMANES ON PHYSIOLOGICAL SELECTION. 



sketching out the lines on whicli the work of verification may 

 proceed. 



There are two main branches of testing inquiry, the one expe- 

 rimental, and the other systematic. It is open to the systematist, 

 in any deparfchieut either of botany or zoology, to utilize his 

 knowledge as a specialist in the following way. Let him cast 

 about for closely allied species which are thoroughly well sepa- 

 rated from one another, either by geographical barriers or by 

 migration. When he has foand any two closely allied species 

 which, for either of these reasons, he feels justified in certainly 

 concluding can never have had an opportunity of intercrossing, 

 let him ascertain whether they are not fertile with one another. 

 The species ought to be as closely allied as possible, because, 

 if they differ in any considerable degree, even though the dis- 

 tinction between them is nominally specific, it really aj)proaches a 

 distinction that is generic ; and in the case of genera there is no 

 question as to sterility being due to a general difference of 

 organic type. Moreover, the specialist ought not to rest satis- 

 fied with only a few observations. His aim ought rather to be 

 to make his observations over a large number of species, tabulate 

 the results, and then see whether the average amount of sterility 

 yielded by all his selected species is not considerably lower than 

 a similar average obtained by selecting a similar number of 

 closely allied species now inhabiting the same continuous area — 

 taking care, however, to choose areas which are believed to have 

 been continuous for long periods of time. Perhaps the best rule 

 to follow (especially in the case of plants) would be to take species 

 which are peculiar to oceanic islands, and to match these with 

 allied species on mainlands, for the first set of tables ; while, for 

 the second set, allied species, both of which are peculiar to the 

 same large continental area, should be chosen. If these observa- 

 tions were made over a considerable number of cases, I should 

 expect them to show an unmistakable difference in the results of 

 the two sets of averages. But it would be necessary to make 

 them over a considerable number of cases, because by this method 

 of inquiry we could never be sure that all modifying conditions 

 had been excluded. Even if we could know the life-histories of 

 each species chosen, there would still remain the element of 

 doubt which is incidentally mentioned by Mr. Darwin in another 

 connection — namely, that " if a species was rendered sterile with 

 some one compatriot, sterility with other species would (? might) 



