.Iiialysis of the Theory of Warning Coloration. 313 



I. ( (!) Certain inseets are readily eaten hy iiiseeti^vrous 7'ertehrates 

 and these are, n'ith feiv exceptions, inconspicuous (protectively colored). 



Tt may be remarked apropos of this statement that it is an obvious neces- 

 sity. A palatable insect without means of defense could not be conspicu- 

 ous and exist where persecuted by vertebrate foes. The palatable forms 

 which remain and are accessible to vertebrate foes are therefore those which 

 are inconspicuous in their natural setting. 



{h) Other insects are (in one or another stage) either refused by insec- 

 tivorous vertebrates or tasted and then rejected or eaten with more or less 

 evident " signs of disgust." These are, tvifh fezu e.veeptions, zuarningly col- 

 ored. Insectivorous vertebrates learn by experience to avoid them. (See 

 especially Finn, 1897a.) 



(c) According to the recent observations of Marshall (Marshall and 

 Poulton, 1902) insectivorous invertebrates (mantids, spiders, dragon flies, 

 etc.) do not refuse conspicuous insects, but usually eat them, although 

 .\crein3e are usually refused by spiders. Insectivorous invertebrates need not, 

 therefore, be further considered in discussing the theory of warning colors. 



These statements are based chiefly on feeding experiments in which 

 captive vertebrates have been fed upon insects, but without being offered 

 a choice between conspicuous and inconspicuous forms. The results of 

 different observers are not wholly in accord. Cf. Poulton (1887) and Mar- 

 shall and Poulton (1902), with Beddard (1892), and Pritchett (1903). As 

 Finn (18970) has pointed out, experiments should be made on vertebrates 

 at liberty and they should be offered a choice. Those observations of Finn 

 that were carried out in this way are in accord with the statements made 

 above under (a) and (b), and are the most conclusive known to me. It 

 is highly desirable that there should be further observations of the same 

 sort as well as studies of the stomach contents of insectivorous vertebrates. 

 Nevertheless, the foregoing statements may be provisionally accepted and it 

 is unlikely that future work will essentially modify them. 



Upon the observations summarized above certain inferences have been 

 based and these may in turn be conveniently summarized in the following 

 statements : 



II. (a) That the conspienonsness of those insects that are provided 

 with means of defense serves not merely to zvarn insectivorous vertebrates, 

 but is advantageous or necessary for that purpose, as the insects are thereby 

 protected from attack, while their insectivorous foes are saved unpleasant 

 experiences. To conspicuousness is thus assigned an advantageous or neces- 

 sary biological function. 



(b) That warning coloration has been developed by selection, through 

 the continued destruction by insectivorous vertebrates of the least conspicu- 

 ous of the insects provided with means of defense. 



These tenets are so obviously a necessary part of the theory of warning 



