WARNING COLOEATIOX. 155 



ground. A magpie ate two caterpillars after it had carefullj- 

 rnbbed off the hairs. 



It is quite clear from these experiments that insects which 

 exhibit warning colours are by no means always exempt from 

 attack. The opinions of insect-eating mammals, birds and 

 reptiles, appear to vary as to the edibility of this or that insect. 



It may be said, of course, that these experiments, having 

 been conducted upon foreign animals, are of less importance 

 than those upon British animals ; this, however, cannot be 

 allowed, because the theory of warning colours implies — not 

 a special recollection of any particular type of insect — but a 

 general association of bright colour with poisonous or dangerous 

 qualities. 



But these experiments do appear to show that very generally, 

 though not always, a disagreeable taste is associated with a 

 conspicuous and \-aried coloration. 



On the other hand, precisely the same deductions can l)e 

 drawn by watching the behaviour of animals when offered in- 

 conspicuously-coloured insects. 



Mr. Poulton has directed attention to the apparent inedibility 

 of the cockroach ; this insect is, however, eaten by varioirs 

 animals, though not by all to which it has been offered. As 

 will be now mentioned, a protectively-coloured caterpillar is 

 distasteful to some insect-eating species. 



Mr. Finn found that the larva of one of our common Xoctu;e 

 — Mamestra persicarice — was not a universal favourite. This 

 caterpillar is inconspicuously coloured, and might fairly be 

 adduced as an example of protective coloration. There are two 

 varieties which occur together on the same plant ; one is brown, 

 the other green. These caterpillars were eaten, after being 

 well pinched, by a Glossy starling, and by the Greater Spotted 

 woodpecker ; if the caterpillar had been one of those which 



