No. 605] 



ANIMAL COLORATION 



273 



Little need be said regarding the inference that the 

 constancy of the ratio in which the females of P. pohites 

 seem to have occurred for many years shows tliat natural 

 selection does not exist for this six'cics in Ceylon, or 

 else that its force is so slight that in lialt' a fciiiurx-. and 

 perhaps in a century and a half, it has produced m. effect 

 appreciable to the method of examination employed. 

 This is valid only if it is true as postulated that tlie 

 various ty^ies of P. polytes constitute *'a population liv- 

 ing and breeding together under the same conditions." 

 But it is gravely to be doubted that this indispensable 

 condition is fulfilled. We have some evidence- (that of 

 Bates and Wallace already cited) that butterflies which 

 differ in color differ in habit, and if it should appear that 

 the colors of butterflies in general are correlated with and 

 repeat those of their surroundings, Punnett's fifth point 

 is forever invalid. For it will be impossible to establish 

 by observation the universal negative that is required, 

 which is, of course, that the three types of female do not 

 differ in any constant respect in their normal behavior. 



Eegarding the sixth point, which has reference chiefly 

 to the fact of seasonal dimorphism among butterflies, it 

 must first be affirmed that although the induced changes 

 differentiating tlie broods of tlie spring and summer, or 

 wet and (lr>- <ea-(>n-. are not diiH'ctly inherited, the capa- 

 bililN' of ivspotidiim' delinitely to the physical stimulus 

 of (dianged teiiii )eratui-(^ oi- hunudity is a heritable racial 

 trait.'"^'* Tlie seasonal variations in the coloration of but- 

 terflies may be analogous upon the whole with the instan- 

 taneous color changes of tropical fishes, which also occur 

 in response to external stimuli. The latter, liowexcr, fol- 

 low more quickly than the former ujion ai»proiM !ale stim- 

 ulation; they are reversible; and are known to l>e nor- 

 mally adaptive, since they reduce the ('onspi('nonsnes> of 

 the individuals in which they appear. It may eventually 

 prove to be a fact that instantaneous adaptive color ad- 

 justments, the phenomena of seasonal and sexual dimor- 



