598 PEor. E. B. POTJLTON : natural selection 



that the results are not common, the resemblance merely 

 deceptive. And now this has even been shown for the colours 

 themselves, in some of the best known and most striking 

 examples of mimicry. 



The last example of similarity in appearance produced 

 through diversity of method is one which occurred to me a few 

 years ago when lecturing on mimicry in the Hope Department 

 at Oxford. It was worked out in detail during the summer of 

 1897, and the general results were communicated to Section D 

 of the British Association at Toronto, on Monday, August 23rd, 

 1897. A brief abstract is printed in the Eeport of the Meeting 

 (pp. 692-694). 



Although the Lepidoptera are characterized as an Order by 

 the clothing of scales upon the wings, cases are very frequent 

 in which this covering has been in part, or almost wholly, lost. 

 By comparison with kindred unmodified forms, as well as by 

 microscopic examination of the transparent areas themselves, it 

 is possible to show that this loss is recent, and to trace the 

 steps by which it has been reached from a condition in which 

 normal scales were present. 



In the very large convergent group of tropical American 

 Lepidoptera, which has sprung up round the best known species 

 of the Ithomiine genera Methona and Thyridia, transparency of 

 a large part of the wings is a characteristic feature. Through- 

 out the group the ground-colour of the wings is transparent, 

 with a black border which general ly'passes inwards as two trans- 

 verse bands in the fore wing, separating the transparent part 

 into three areas, and as one band in the hind wing, separating 

 it into two areas. The group consists of Ithomiince {Neotropince) 

 of many genera, of Danainae, of PierincE, and of moths of the 

 genera Anthoonyza and Eyelosia, belonging to the JBericopidae 

 {Rypsidce\ and the widely separated genus Gastnia {Castniidce). 

 Under the theory of external causes we should expect that the 

 transparency would be attained in a similar manner throughout, 

 by the reduction of scales to hairs, by the complete loss of scales, 

 or by some other uniform method. Under the theory of natural 

 selection we should expect that the methods would be different 

 in the different groups. There are many ways in which 

 transparency can be attained, and we should expect that one 

 group would submit one set of variations making towards the 

 resemblance, another a different set, to the operation of 



