98 Mr. G. A. K. Marshall on Diaposematism, with reference 



same area the education of young birds, lizards, etc., 

 necessitates the destruction of approximately 1000 indi- 

 viduals in each group of distinctive patterns. Again, let 

 us suppose that A is a common species and is represented 

 by 100,000 individuals in that locality, while B is more 

 scarce and has only 5000. Now, ex fa/pothesi, the 

 Mullerian factor will exact an equal toll from the two 

 species, and its selective importance must therefore 

 depend upon their relative numbers. In the present 

 instance the plentiful A will lose only 1 per cent, of its 

 individuals from this cause, while the scarcer B will lose 

 no less than 20 per cent. It is therefore reasonable to 

 suppose that if certain individuals of B chanced to 

 develop a variation in the direction of A, that variation 

 would have, in relation to the Mullerian factor, a definite 

 advantage over its own typical form, because it would 

 tend to share, at least to a small extent, in the relative 

 advantage enjoyed by A ; and the continuance of the 

 elimination would gradually tend to enhance the variety 

 and to diminish, and finally exterminate, the type. This 

 would be a case of simple Mullerian approach from B to 

 A. But if the mimicry is to be reciprocal it must be 

 shown that A is capable of approaching B by a precisely 

 similar process. In order to simplify the argument let us 

 make the supposition that 10,000 specimens of A simul- 

 taneously present a sudden marked variation in the 

 direction of B, to such an extent that young birds would 

 be liable to rank them with B rather than with A; what 

 will be the effect of the Mullerian factor on this remark- 

 able variety ? The point again to be considered is the 

 relative incidence of the destruction. There will now be 

 90,000 examples of the A pattern, and 15,000 of the 

 B pattern. If 1000 individuals of each are again destroyed 

 by experimental tasting, a simple calculation shows that 

 the percentage of loss due to this factor will be six times 

 greater in the variety of A than in the typical form of A. 

 In fact the result is exactly the opposite of what took 

 place in the case of the variety from B towards A. 

 Whereas there we found that the variety had an appreci- 

 able advantage over its own type form, here it is seen to 

 be at a distinct disadvantage. It is obviously impossible 

 for the Mullerian factor to build up a mimetic resemblance 

 on such a foundation; if it have effect at all, that effect 

 must be to gradually eliminate the variety and to establish 



