November 29, 1912] 



SCIENCE 



151 



and so cause the general depression of its sur- 

 vival curve. For instance, let length of 

 feather have a differential survival value, 

 when a species of sparrov? encounters colder 

 winters. Evolution in the direction of in- 

 creased feather length would cause an in- 

 creased abundance of the species, unless there 

 was a counter-influence. Such a counter-in- 

 fluence might be a greater mortality from 

 shrikes, as more sparrows would thus come 

 under their observation and engage a larger 

 share of their attention. 



The graphic method here proposed is also 

 applicable to mutation (in the De Vriesian 

 sense) though of less value in that connection. 

 In such cases the mutation is plotted on the 

 base line at the appropriate distance from the 

 polygon of frequency and the curve of sur- 

 vival is extended past this point. The curve 

 of survival may be level in the region of the 

 polygon, but inclined outside of it, so as to 

 affect the mutation either favorably or un- 

 favorably (Fig. 15a). Or the survival curve 

 may be inclined throughout its course, in 

 which case the mutation does not have an ex- 

 clusive advantage or disadvantage. Many of 

 the complications referred to before are also 

 applicable here, but further mention will only 

 be made of the case where the survival curve 

 is strongly inclined at the magnitude of the 

 mutation and it is therefore strongly sub- 

 jected to natural selection within its own unit 

 in addition to its competition with the old 

 species. 



Coincident selection, which many zoologists 

 have apparently had difficulty in understand- 

 ing, is, I believe, made quite clear by the use 

 of this method. I follow Gulick in consider- 

 ing Lloyd Morgan's term of coincident much 

 preferable to Baldwin's meaningless phrase of 

 " organic selection." In expressing coincident 

 selection, it is only necessary to construct with 

 the survival curve two polygons of frequency, 

 one for the innate variation, and the other for 

 the resultant variation after the modification 

 by the environment. 



The action of coincident selection will 

 differ in degree according to the type of corre- 

 lation of the modification with the innate 



variations. There may be three cases. In the 

 first, the modification is always the same in 

 amount, regardless of the degree of develop- 

 ment of the innate character. This results in 

 the transfer of the original polygon of fre- 

 quency to one side or the other. Since, how- 

 ever, the modification would of course be some- 



PlG. 16. The dash line polygon represents the 

 individuals in a case of extreme coincident selec- 

 tion after modification. The solid line polygon 

 represents the individuals as they would have been 

 if not modified. Dotted lines are theoretical sur- 

 vival curves. 



what variable, there would be some readjust- 

 ment of distribution, and a corresponding in- 

 crease of the variability (Fig. 16a). This is 

 the case usually assumed in the consideration 

 of coincident selection and its consequences 

 are well knovsni. In Fig. 165, where the modi- 

 fication is opposed by natural selection, it may 

 be expected to decrease in amount. In the 

 second case, the modification is greater in 

 those individuals which have the innate char- 

 acteristic in a less degree than in those in 

 which it was larger. Thus, elephants with an 

 innately shorter trunk might have greater 

 modification than others because of greater 

 strains involved in its use. The final result 

 in this case, then, would be the shifting of the 



