rOKMATIVE AND SELECTIVE ACTION. 381 



and with the preying fish, though it is itself blind. But this 

 species is found exclusively on the outer limits of the region 

 inhabited by Periophthalmus, namely, on the south-east coast of 

 Australia and in the central region of the Pacific. Also the 

 species spokon of as living on the West coast of Africa, and 

 which is probably blind, belongs to such another frontier dis- 

 trict. Now it is, of course, not much to be wondered at, that 

 in these frontier districts a blind form should occur among 

 those that can see. But there is a much more striking argu- 

 ment for allowing this exception again to serve as confirming 

 the rule. While all the truly blind species of those regions 

 whence Periophthalmus is absent, are devoid, not merely of 

 developed eyes, but of the first elements for forming them, and of 

 the accompanying weapons — having no dorsal glands whatever, 

 since without these the eyes would be useless — the blind species 

 of the Pacific has both the glands and the cell-gi'oups inside the 

 papillcs agi'eeing precisely in structure with a middle stage in the 

 series exhibited in the formation of the veiy highly developed 

 eyes of Onchidium verruculatum. Thus here we seem to have 

 a species in which either eyes formerly existing have begim to 

 disappear from desuetude, or the construction of a true organ 

 of sight has been begun by the coincidence of the three factors 

 •which would give rise to it.'^^ 



I have treated of this example somewhat in detail, on pui'pose 

 to define as clearly as possible the limit -line where the external 

 modifying causes which we discern as giving rise to a new 

 organ — or rather as transforming one already existing into 

 another — cease to be eifective, and where those selective influ- 

 ences begin to act which determine the further perfecting of a 

 newly developed organ. 



I am, moreover, convinced that in eveiy case, by a corre- 

 sponding method of research, the same limit-line will be easily 

 detected. . That it is not yet recognised, nor even in most cases 

 supposed to exist — so that it is often difficult to avoid a con- 

 fusion between the transforming causes and the piirely selective 

 ones — is, in my opinion, entirely due to our having hitherto 

 been satisfied with mere general speculations on Darwin's theory, 

 ind having neglected to investigate the innumerable problems 



