Characters as Adaptive and Specific. 211 



markable experiments of Dorfmeister and Weismann 

 in relation to this subject are well known. More 

 recently Mr. Merrifield has added to their facts, and 

 concludes that the action of cold upon the pupae — 

 and also, apparently, upon the larvae — has a tendency 

 to produce dark hues in the perfect insect ^. 



But, passing now from such facts of climatic vari- 

 ations over wide areas to similar facts within small 

 areas, in an important Memoir on the Cave Fauna 

 of North America, published a few years ago by the 

 American Academy of Sciences, it is stated : — 



"As regards change of colour, we do not recall an exception to 

 the general rule that all cave animals are either colourless or 

 nearly white, or, as in the case of Arachnida and Insects, much 

 paler than their out-of-door relatives." 



Now, when we remember that these cave faunas 

 comprise representatives of nearly all classes of the 

 animal kingdom, it becomes difficult, if not impos- 

 sible, to imagine that so universal a discharge of 

 colouring can be due to natural selection. It must 

 be admitted that the only way in which natural 

 selection could act in this case would be indirectly 

 through the principle of correlation. There being no 

 light in the caves, it can be of no advantage to the 

 animals concerned that they should lose their colour 

 for the sake of protection, or for any other reason of 

 a similarly direct kind. Therefore, if the loss of colour 

 is to be ascribed to natural selection, this can only 

 be done by supposing that natural selection has here 

 acted indirectly through the principle of correlation. 

 There is evidence to show that elsewhere modification 



* Trans. Entom. Soc. 1889, pait i. p. 79 et seq. 



P a 



