1887. J VALUE OF COLOUR AND MARKINGS IN INSKCTS. 211 



of undue iufluence and thus throws out of adjustment the preexisting 

 condition of comparative equilibrium. 



Another conclusion which is demonstrated very completely hy the 

 tables is that a Caterpillar may be eaten by one insect-e.ating Vertebrate 

 although refused by another. I believe, however, that the acquisition 

 of an unpleasant taste and of conspicuous colours appealed, at any 

 rate at first, to a large number, probably all, the vertebrate foes ; 

 for if this were not so, if the species became unpalatable and 

 conspicuous to (say) half its enemies, and became conspicuous but 

 remained palatable to the remainder, it seems only reasonable to 

 conclude that immunity from the attacks of one set of foes would 

 be counterbalanced, or perhaps more than counterbalance-], by the 

 facilities afforded to the other set. On this account and for other 

 reasons which will be given helow, I think it probable that the 

 difi^erences observed between the enemies of insects in this respect 

 are of recent date as compared with the acquisition of this mode of 

 protection, and have arisen out of the great competition for food ; but 

 in most instances the change of habit has not become so far confirmed 

 that the previously distasteful food is eaten with avidity and pleasure. 

 The first table of highly eons[)ieuous larvae (including Croesus and 

 Nematus from tlie second table) can be shortly analyzed to show in 

 it the various stages of transition from the most utter disregard to 

 the opposite extreme of conduct, indistinguishable from that observed 

 when the larvae are known to be relished. The intervenins stages 

 are furnished by the details given by the different observers, and 

 are described in the headings of the vertical columns between those 

 numbered I. & V. (see page 2 i 2). 



It must be remembered tiiat these analyses represent a comparison 

 between the results of experiments carried out under different systems 

 and with the use of an incomplete number of Vertebrates in all 

 cases. Hence many of the insects would doubtless have to be 

 shifted into other columns after being offered to other Vertebrates, 

 or to those actuully employed, if it were certain that they were 

 thoroughly hungry, Allowing for this, however, the analyses provide 

 us with numerous instances of transition through all conditions of 

 failure in the protective efficacy of the method we are discussing. 

 At the same time one can see at a glance the relative behaviour of 

 different insect-eaters as far as they have been tested in the case of 

 each larva. 



Jenner Weir's suggestion that the hairs of certain larvae act as a 

 warning of other unpleasant qualities can also be tested by the 

 examination of the former tables. There are altogether fourteen 

 larvae which may be called hairy, out of a total of twenty-seven 

 (omitting the two terrifying species). Of these, two {L. rubi and P. 

 fuliginosa) were eaten, as far as any observations are recorded ; one of 

 the former and five others {L. rubi, L. quercus, L. pini, P. auriflua, 

 A. caja, and O. potaf.oria) are either known to possess irritating 

 hairs or are believed to possess them ; as many as five are gregarious 

 {V. 10, V. urtict^, P. bucephala, E. lanestris, C. neustria), and this 

 habit, together with the colour, is by far the most important factor 



