No. 605] 



ANIMAL COLOFATION 



263 



ance, is so simple a form of protection, and yet ex liy- 

 pofhesi so absolutely complete, that it seems remarkable 

 that more species have not availed themselves of this 

 mode of defence. He argues that if once their potential 

 vertebrate enemies were driven to eat any such insects in 

 spite of their unpleasant taste, they would almost cer- 

 tainly soon acquire a relish for what was previously dis- 

 agreeable, and the insects would be in great danger of ex- 

 termination, having in the meantime become conspicuous 

 by gaining warning colors. He concludes that if this 

 reasoning is correct, it is clear that this mode of defence 

 is not necessarily perfect, and that it depends for its ap- 

 parently complete success upon the existence of relatively 

 abundant palatable forms : in other words, its employment 

 must be strictly limited. 



Dixeyi*^ encounters the same difficulty in his consid- 

 eration of Batesian mimicry. He observes that in this 

 relation the advantage is all on the side of the edible 

 mimicking species, whose existence is, indeed, a source 

 of danger to the form mimicked, inasmuch as any ex- 

 perience gained by tasting the former would be used to 

 the detriment of the latter. From these considerations 

 he believes that such an association can exist only when 

 the numbers of the one species are insignificant in com- 

 parison with those of the other. Upon this point he is in 

 essential agreement with Wallace,^^ who states that mim- 

 icry has beto shown to be useful only to those species and 

 groups that are rare and probably dying out, and would 

 cease to have any effect should the ]iroportionate abun- 

 dance of mimic and model be reversed. 



Here, then, ai;e two hypothetical t\ pes of a-s(M'intioiL 

 whose persistence admittedly depeitids upon the main- 

 tenance of definite though undetermined ratios between 

 their com]-)onents. 1 lence it devolves upon those who hold 

 that the a^suinod n^alions are real, to outline some sys- 

 tem l)y which th(^ dm^ i)roportion of protected and unpro- 



