254 



THE AMERICAN NATURALIST [Vol. LI 



In some eases natural elimination is certainly in some degree selec- 

 tive, while in other cases it certainly is not, and in the most favorable 

 eases of all the selection is apparently not very rigorous. Gross terato- 

 logical abnormalities are eliminated. But the smaller deviations from 

 type, which in theory ought to furnish the basis of selection, appear 

 upon quantitative study less generally and sharply determinative of 

 survival than might have reasonably been expected theoretically. 



It is perhaps worthy of note that ever since it appeared that 

 the larger, and rarer, discontinuous variations are in no danger 

 of being lost through swamping, it has been beside the mark to 

 ascribe especial theoretical significance to smaller deviations from 

 type. Attention may also be directed to the important admis- 

 sion that in some instances elimination is known to be selective. 

 It will then be in order to examine reports of researches upon 

 the basis of which it is confidently asserted that in other cases the 

 same is not true. 



It happens that with five others one is cited which falls 

 squarely within my own field of investigation. This is Pro- 

 fessor ReighardV "Experimental Field Study of Warning 

 Coloration in Coral-reef Fishes," which Dr. Pearl seems to 

 consider of particular significance for his own argument. But 

 since it has been my good fortune to study the same material in 

 the same place, more extensively and with better facilities, I feel 

 justified in saying that Reighard's results will bear no such 

 interpretation as is here placed upon them. He proved that 

 gray snappers possess powers of discrimination and of memory 

 which would lead one to suppose that, if the bright colors of the 

 smaller reef fishes possess a warning significance, the snappers 

 should be aware of it and avoid them. He did not prove beyond 

 possibility of doubt that they do attack such fishes freely, but 

 that is of no importance in the present connection; his con- 

 clusion that tropical fishes are not warningly colored with refer- 

 ence to their commonest enemies is perfectly sound. His ideas 

 regarding immunity coloration are, however, the only ones in 

 his paper which have the remotest bearing upon the matter of 

 selective elimination, and there is no reason to suppose that these 

 are more than logical deductions from incorrect premises. 



Dr. Pearl's* own report upon the natural elimination suffered 

 3Reighard, Jacob, 1908, "An Experimental Field-study of Warning Col- 

 oration in Coral-reef Fishes," Carnegie Institution of Washington, Papers 

 from the Tortugas Laboratory, Vol. 2, pp. 257-325. 



* Pearl, Raymond, 1911, "Data on the Relative Conspicuousness of 

 Barred and Self-colored Fowls," The American Naturalist, Vol. 45, pp. 

 107-117. 



