October 9, 19 19] 



NATURE 



125 



obliges them to do the ties of blood-relationship. Then, 

 again, it is obvious wh^ they occur in the same and 

 not in widely different localities; in some instances, as 

 we have seen, their bearers actually tlving in company 

 and frequenting the same flowers; for the common 

 aspect, supposing it to be in any sense protective, 

 would onl); take effect when the sharers in it were 

 exposed to the attacks of the same body of enemies ; 

 that is to say, when they inhabited the same locality. 

 .And this would be equally true, whether the warning 

 colours are shared between distasteful forms, or 

 whether they are deceptively adopted by forms unpro- 

 if'Ct<?d by inedibility; whether, in Prof. Poulton's terms, 

 ihey are synaposematic or pseuda|X)sematic. I do not 

 enlarge upon this part of the question, or upon the 

 theories which are known under the names of Bates 

 and .MiiUer respectively, because these theories have 

 been fullj dealt with elsewhere, and I think I mav 

 assume that they are familiar to the greater part of 

 my hearers. But that mistaken ideas as to what is 

 really meant by protection and mimicrv still prevail in 

 some quarters, is evident from certain remarks of 

 Fruhstorfer in dealing with the genus Prioneris which 

 we have just been discussing. "Wallace," he sa\s, 

 "regards the ' rarer' Prioneris as a mimetic form 'of 

 the ' commoner ' Delias. But I cannot accept his 

 view, since mimicry among the in all respects harmless 

 Pierids appears no sort of jirotection, and, ])roperlv 

 sjx^aking, the smooth-margined Delias should rather 

 copy the armed Prioneris if there is assumed to be 

 mimicry at all." If an\one has no better knowledge 

 than this of what is meant by the theory of mimicry, 

 it is not wonderful that he should consider the subject 

 unworthy of serious attention. 



The warning-colour theor)-, then, gives a rational 

 explanation both of the superficial character of the 

 resemblances and of the geographical factor in their 

 occurrence. But it obviouslv involves the realitv of 

 natural selection; and it is 'here that some are'dis- 

 posed to part company with the upholders of the 

 theory. I have already referred to the fact that much 

 positive evidence now <xists both that butterflies are 

 eaten and that preferences on the part of thoir enemies 

 exist between one kind and another. I will only 

 remark in passing that the objector on this score 

 sometimes adopts an attitude which is scarcelv 

 reasonable, and, perhaps, on that verv account is 

 somewhat hard to combat. The kind of" objector that 

 I mean begins by saying that the destruction of 

 butterflies by birds and other enemies is not sufficient 

 to give play for the operation of selection. You beg 

 his pardon, and produce evidence of considerable 

 butterfly destruction. To which he replies, " Oh, they 

 are eaten, are they? I thought you said thev were 

 protected." This is a good dilemrna, but the dilemma 

 is notoriously an unconvincing form of argument. If 

 a reply be called for, it may be given like this : 

 •'Butterflies are cither preyed upon or thev are not. 

 If they are, an opening is given for selection; if 

 they are not, it shows the existence of some form 

 of protection." The essence of the matter is that 

 both the likes and dislikes of insectivorous animals, 

 and the means of protection enjoved bv their prev, 

 are not absolute, but relativ-e. A bird that will reject 

 an insect in some circumstances will capture it 

 in some others; ' it will, for instance, avoid 

 insect k if it can get insect B, but will feed on .\ 

 if nothing else is to be had; and it is probablo that 

 scarcely any insect is enlirelv proof against the attack 

 of every kind of enemv. The relative nature of pro- 

 tection is readily admitted when the question is not 

 one_ of mimicry or of warning colours, but of pro- 

 tective resemblance to inanimate objects. .Ml degrees 

 NO. 2606, VOL. 104] 



of disguise, from the rudimentary to the almost per- 

 fect, are employed ; the lower degrees are allowed 

 to be of some service, and, on the other hand, a dis- 

 guise that is almost completely deceptive mav at 

 times be penetrated. This consideration applies also 

 to the objection that the first beginnings of mimetic 

 assimilation can have no selective value. If the 

 rough resemblance to an inanimate object affords 

 some amount of protection, though that amount may 

 be relativelv small, why should not the same apply 

 to the first suggestion on the part of a mimic of 

 an approach to the aposeme or warning colour of 

 its model? The position that neither kind of assimila- 

 tion is of service is intelligible, though not common; 

 but there is no reason why benefit should be afiirmed 

 in the one case and denied in the other. There are 

 further considerations which tend to deprive this 

 latter criticism of force; the fact, for instance, that 

 a resemblance to one form may serve as a stepping- 

 stone for a likeness to another; or, again, the exist- 

 ence of clusters, as thev may be called, of forms 

 varving in atfinitv, but embodying a transition by 

 easv stages from one extreme to another. In a case 

 of this sort the objection that may be felt as to two 

 terms in the series arbitrarily or accidentally picked 

 out is seen to be groundless when the whole 

 assemblage is taken together. 



Much attention has lately been given to the fact 

 that of individual variations some are transmissible 

 bv hereditv and some are not ; under the latter head- 

 ing would generallv fall somatic modifications directly 

 induced upon the individual by conditions of environ- 

 ment. Whether any other kind of variation belongs 

 to the same category need not for the present pur 

 pose come into discussion. But with regard to the 

 undoubtedh- transmissible variations, or mutations if 

 we like to call them so, there is, I think, a fairly 

 general consensus of opinion that thev need not neces- 

 .sarily be large in amount. \ complete gradation, in 

 fact, appears to exist between a departure from type- 

 .so slight as to be scarcely noticeable, and one so 

 striking as to rank as a sport or a monstrosity. .And 

 we know now that where the Mendelian relation 

 exists between two forms, no amount of inter- 

 breeding will abolish either type; intermediates, when 

 formed, are not permanent, and if one type is to- 

 prevail over the other, it must be by means of selec- 

 tion, either natural or artificial. 



In view of all these considerations, I venture to 

 think that there is no reason to dispute the influence 

 of natural selection in the production of these remark- 

 able resemblances. Other interpretations may no- 

 doubt be givcA, but they involve the ignoring of some 

 one or more of the facts. It mav fairly he claimed' 

 that the theories of Wallace, Bates, and Miiller, 

 depending as thev do on a basis of both observation 

 and experiment, corne nearer to accounting for the- 

 facts than anv other explanation yet offered. It will, 

 of course, always be possible to deny that any ex- 

 planation is attainable, or to assert that we ought to 

 be satisfied with the facts as we find them without 

 attempting to unravel their causes. But such an 

 attitude of mind is not scientific, and if carried into 

 other matters would tend to deprive the study of 

 Nature of what, to most of us, is its principal charm. 

 It is Quite true that before the validity of any 

 generalisation is accepted as finally and absolutely 

 established, every opportunity should be taken of 

 deductive verification. This has been fullv recog- 

 nised bv the supporters of the theorv of mimicry, 

 and much has been done to test in this manner the 

 various conclusions on which the theorv rests. The 

 verification is not complete, and perhaps n"ver wilT 



