to limitations of the Mi'dhrian Hypothesis of Mimicry. 103 



Alternating Besemhlance. 



Although there appear to be serious difficulties in the 

 way of accepting any idea of mutual simultaneous mimicry 

 between two unpalatable species, yet it is not impossible 

 that a certain interchange of characters might still take 

 place between them. For example, at first A might be 

 more numerous than B and a simple Miillerian approach 

 might take place from B towards A. While this was in 

 progress conditions might arise which would cause a 

 reversal in the relative numbers of the two species, so 

 that B would become the dominant partner. Then any 

 further mimetic approach that might take place would be 

 from A towards B, so that the final result might show a 

 fusion of the original colours of the two species. This kind 

 of approach is quite different from that contemplated in 

 the hypothesis of Reciprocal Resemblance, although pro- 

 ducing similar results ; it may therefore be distinguished 

 under the name of Alternating Resemblance. It is not 

 probable that this phenomenon will be of common occur- 

 rence in nature, and it is only here referred to in order to 

 show that even if the actual existence of reciprocity can 

 be demonstrated, that will not of itself afford proof of 

 Diaposematism ; for the facts can be consistently inter- 

 preted on the lines of a one-sided Miillerian approach. 



Since the foregoing remarks were written Professor 

 Poulton has kindly suggested to me a criticism which 

 appears, at first sight, to lend some support to the idea of 

 Diaposematism. As a very similar suggestion has been 

 made to me independently by another friend, it seems 

 advisable to show how it may be met. 



It has been contended that among the enemies of the 

 two distasteful butterflies, A and B, there will be some 

 (which we may call X) whose first experience will be 

 entirely of A. Having thoroughly learnt that this insect 

 is unpalatable they will then come across B. If among 

 the individuals of B there occurs a variation (B') which 

 presents a rougldy A-like appearance, then B' will have 

 an advantage over B in relation to the attacks of X. 

 Similarly other enemies (Z) will first have experience of 

 B only, so that when they afterwards attack A, the variety 

 A', which has varied in the direction of B, will be at an 

 advantage as compared with A, because it Avill recall to 



