358 PROCEEDINGS OF THE ACADEMY OF [June, 



"It has already been pointed out that the acceptance of insects 

 by insectivorous animals in captivity is no proof of their normal 

 likes or dislikes in a wild state. Such acceptance only proves what 

 their action would be when they had been, from some exceptional 

 cause, kept without their normal food in its usual quantity and 

 variety. Hence the fact that Acrseas were devoured [by a ground 

 hornbill] is no evidence that they are normally eaten except in a 

 lime of unusual hunger. On the other hand, the rejection of two 

 L. chrysippus, after three Acrseas had been readily eaten, indicate 

 that the former butterfly is decidedly distasteful to this species of 



bird" (p. 348) 



" Byblia ilithyia was .... distinguished [by baboons] from an 

 Acrcea, but this by no means proves that the resemblance is not 



beneficial under natural conditions (p. 388) Considering 



what has been already argued about insect-eating animals in con- 

 finement, the acceptances (excluding the Hesperiidse) probably do 

 not justify the conclusion that the Lepidoptera were palatable, or 

 that they would be sought for in the wild state except under the 

 stress of hunger" (p. 389). 



"It has already been pointed out that the refusal or evident 

 dislike of insect food by captive animals is trustworthy evidence of 

 unpalatability, while acceptance is not proof of palatability " (p. 436). 

 It is self-evident that this oft-repeated dictum is merely special 

 pleading for the admission of as much as possible of the evidence 

 favorable to the theories, and the exclusion of as great a proportion 

 as possible of the evidence that might be unfavorable. So plain is 

 this fact that even Mr. G. A. K. Marshall, collaborator with Prof. 

 Poulton in the paper quoted from, severely criticized the Professor's 

 attitude. He says 93 in part: 



' There is too emphatic an insistence upon the possibility of error 

 where an insect is accepted; for it practically casts suspicion upon 

 every such case. On the other hand, the possibility of error in the 

 other direction is not indicated." 



The Experiments are not Trustworthy Guides to Behavior under Natural 



Conditions. — The writer is by no means the first to question the 



analogy of behavior under experimental to that under natural 



conditions. The idea is put briefly by L. W. Kline in an article on 



'Methods in Animal Psychology": 94 "Nothing so shrinks and in- 



■■ Trans. Ent. Soc. Lond., 1908, p. 140. 

 Amer. Joum. of Psychol, 10, 1898-9, p. 276. 



