3i6 Palmers from the Marine Biological Laboratory at Tortugas. 



of utility before selection can have operated on it. As a condition ante- 

 cedent to selection, incipient warning coloration must have reached a point 

 at which insectivorous vertebrates could distinguish warningly-colored in- 

 sects from those which did not possess that character. It is conceivable 

 that selection then operated to intensify warning coloration, but can we 

 believe that it carried it to its present stage far beyond the point at which 

 discrimination must have been easy? Should it not rather have stopped 

 so soon as warningly-colored insects were readily discriminated, long before 

 the present very conspicuous differences had appeared? Are we to believe 

 that when an insectivorous vertebrate encounters together two color varie- 

 ties of a familiar, unpalatable species which differ slightly in conspicuous- 

 ness, it will take the less conspicuous and leave the other? Does it seem 

 still further possible that this result will follow if the two varieties are en- 

 countered in succession ? Should the varieties diif er from one another 

 in conspicuousness to the extent of mutations, can one believe that a verte- 

 brate foe with experience of the equal unpalatability of both would take 

 the one and leave the other? Yet this must have been the case if warn- 

 ing colors were perfected by selection. Is it not more probable that if any 

 discrimination occurs, the vertebrate foe would attack the more conspicuous 

 insect because it is less familiar and that his attacks would thus tend to retard 

 the development of conspicuousness rather than to accelerate it? 



If warning coloration may be initiated in insects without the aid of selec- 

 tion, as indeed it must be, and if the later stages in its development may 

 not be satisfactorily accounted for by selection, then we need not invoke 

 its aid at all. The entire development of warning coloration may well 

 have been due to the action of the same forces that initiated it. That these 

 forces may be orthogenetic is obvious, but my own work does not seem to 

 afford a sufficient basis for the discussion of this subject. Mayer (1902) has 

 concluded that in butterflies of the genera Papilio and Ornithoptera and in 

 Hesperidse the color-patterns have been mainly determined by internal factors 

 (race tendency), not by external influences or natural selection. 



A SUBSTITUTE FOR THE THEORY OF WARNING COLORATION. 



If the views already expressed concerning the development of conspicu- 

 ousness in coral-reef fishes are well founded, this character has resulted 

 from the action of internal forces in the absence of counteracting selection. 

 Selection in the direction of protective resemblance has been held in abey- 

 ance by the coral-reef habitat which has effectually limited the attacks of 

 predaceous fish. Selection in the direction of aggressive resemblance has 

 not taken place owing to the nature of the food. As a result coloration 

 has developed unhampered by selection and this development has resulted 

 in definite colors and patterns, constant and characteristic of each species. 

 Among these are the colors and patterns typical of warningly-colored insects. 



