that very difterent species have attained this end, each by the accumulatiou oi 

 those variations which were rendered possible by its pecuhar ancestral history and 

 present constitution — in other words, by the theory of natural selection. 



A more elaborate case, which I have recently investigated, is afforded by a 

 large group of tropical American Lepidoptera — moths as well as butterflies — which 

 closely resemble certain common Avide-spread species of the Ithomiine genera 

 Methona and Thyridia. The appearance thus produced consists of a transparent 

 ground with a black border to both wings, the fore wing being also divided by 

 black transverse bars into three transparent areas — the hind wing usually into 

 two. From a comparison with other species of the various families, &c., not 

 altered in this direction, we know that the transparent wings are not ancestral. 

 When we investigate the manner in which transparency has been attained, it is 

 found to be by difterent methods in the different constituents of the group. Among 

 the numerous genera of Ithomiinee {Methona, Thyridia, Dircenna, Eutresis, Ithomia, 

 &c.) the result has been attained by the reduction of the scales to a very minute 

 size, so that they hardly interfere with the passage of light. This reduction 

 aftects the two kinds of scales which alternate with each other in the rows upon 

 the wings of tbis sub-family, a common result being {e.g., in Methona and Thyridia) 

 the alteration of the more slender scales into hairs, and of the broader ones into 

 minute bifid structures, still retaining scale-like proportions in spite of their 

 extremely small size. In others, again, the two kinds of scales are reduced 

 respectively to simple and Y-shaped hairs, which regularly alternate along the 

 rows. In "the Danaincc proper, represented by the genus Ituna, the transparency 

 is chiefly due to the great diminution in the number of the scales, and those 

 which remain are neither much reduced in size nor altered in shape. In the 

 Pierince, represented in this group by only a single species, Dismorphia orise, the 

 scales are greatly reduced in size, but are neither greatly altered in shape nor 

 diminished in numbers. 



Hence in these three sub-families of butterflies transparency is attained in 

 three difl^erent ways, viz. (1) by reduction in size and simplification in shape; 

 (2) by reduction in number ; and (3) by reduction in size alone. 



When we examine the moths which fall into the group, we find a much 

 greater difference in the methods, corresponding to the wider divergence in 

 affinity. In the several species of the genus Castnia the scales lose their pigment, 

 although undiminished in size, while they are at the same time set vertically upon 

 the wing, so that light can freely pass between their rows. In the widely sepa- 

 rated genus Hyelosia the arrangement is nearly the same, except that the vertical 

 scales are much attenuated. In the genus Anthomyza, which iurnishes the group 

 with many species, the scales retain the normal size, shape, and overlap, but 

 become so completely transparent that the light freely passes through them. 



In all the numerous constituents of this large group of Lepidoptera a very close 

 resemblance has been produced by entirely different methods ; a result which, it 

 has been argued above, is only consistent with the view that natural selection 

 alone, among all the explanations which have been suggested, has been the cause 

 of the observed phenomena. 



I owe to the kindness of Mr. Godman and Mr. Salvin the opportunity of 

 studying all the butterffies of this large transparent-winged group, while Mr. 

 Herbert Druce kindly lent me those moths which are not represented in the Hope 

 Collection in the Oxfoi'd University Museum. 



