102 Mr. G. A. K, Marshall on Dia2wsematism, with reference, 



On reverting to the case where the two species have 

 equal numbers and applying the above test, we find that 

 the contention in favour of a Mtillerian approach cannot 

 reasonably be maintained. If A and B are equal, then 

 their position in regard to the Mtillerian factor is abso- 

 lutely identical, so that neither has an iota of advantage 

 over the other and the essential condition for the origin of 

 Miillerian mimicry is lacking. In these circumstances it 

 is evident that any small variation from one towards the 

 other will not practically affect the numerical relationship 

 of the two species and will therefore have no mimetic 

 value. 



The comparative nuiiibers of individuals in Miillerian 

 Associations, 



Unfortunately we have very little knowledge as to the 

 approximate individual numbers of the species forming 

 any Miillerian association; but in November 1903, Pro- 

 fessor Poulton exhibited to this Society (Proc. Ent. Soc. 

 1908, p. liv) a series of 823 butterflies captured on a single 

 day by Mr. C. B. Roberts in British Guiana, with a view 

 to giving an idea of the relative numbers of the species 

 in that spot. Out of the total no less than 295 specimens, 

 comprising 9 species, fell into the mimetic association 

 which centres round the common Melinxa mneme, all of 

 them belonging to undoubtedly distasteful genera. M. 

 mneme itself vastly outnumbered the other species, being 

 represented by 253 individuals, while the remaining 8 

 species were represented by the following numbers : 8, 9, 

 10, 9, 1, 3, 1, 1. There seem good grounds for supposing 

 that these figures do give a fair rough idea as to the 

 relative occurrence of these species in that locality, and 

 the very striking discrepancy in the numbers of the 

 rnimics as compared with their dominant model is 

 precisely what we should expect to find on the conception 

 of Miillerian mimicry here advocated. Such conditions 

 w^ould be specially favourable to the production of a large 

 mimetic association, and it seems highly probable that in 

 actual practice Miillerian mimicry will only have arisen 

 where the discrepancy in numbers has been fairly con- 

 siderable ; although the possibility of its occurrence where 

 the difference is comparatively small can, of course, be 

 theoretically upheld. 



