756 



SCIENCE 



[N. S. Vol. XXXVI. No. 935 



toward the limit, the species must remain at 

 the limit. 



The process of natural selection must be 

 further analyzed because of the heterogeneity 

 of the species. A species is made up not only 

 of individuals of two sexes in most cases, but 

 also of individuals of various ages. This dif- 

 ference may affect the survival results in two 

 different ways. The individuals of one sex 

 may differ quantitatively so that they will be 

 affected by differing survival rates when sur- 

 vival is discriminative. Because of the sex- 

 ual difference, Weldon in his Garcinus experi- 

 ment used males only, in order that it might 

 not be necessary to increase the number to be 

 measured. 



At different ages, natural selection must 

 often operate differently, for the characteristic 

 changes quantitatively, and sometimes quali- 

 tatively, with age. Where the change is quan- 

 titative, the individual will change in its sur- 

 vival chance, where survival is discriminative. 

 Where the change is qualitative the charac- 

 teristic in question may only be in evidence 

 during part of the lifetime. In addition to 

 this, the survival chance may change with age, 

 regardless of the magnitude, because of the 

 general inefficiency of the immature. In many 

 species, the rigor of natural selection is doubt- 

 less concentrated on the younger stages, with 

 much decreased action later. Specific in- 

 stances we have in the vital statistics of man, 

 and in the large birds of prey, where they are 

 not molested by man. 



In experimental work, economy of effort de- 

 mands concentrated attention upon adults 

 only or upon the young of some restricted age 

 group. Nevertheless, such a study has a seri- 

 ous limitation, for the whole influence of nat- 

 ural selection upon a species can not be known 

 until we have an analysis of its action for 

 every age group in the species. While the 

 number of young individuals in the sparrow 

 experiment of Bumpus was too small to per- 

 mit an analysis along age lines, yet inspection 

 shows that the selection was affected by age. 

 It is quite probable that where natural selec- 

 tion is inoperative upon adults, it may be 

 very efficient in some younger stages. In Fig. 



14ffl I have illustrated the escape from the inci- 

 dence of natural selection by age. Cramp- 

 ton's results with the antenna lengths of 

 Philosamia pupw might not be found to apply 

 in imaginal life. In fact, we have the possi- 

 bility that natural selection might be nullified 

 by a reversal of action in the adult from its 

 action at some younger stage. Thus an extra 

 abdominal segment might be favored in the 

 caterpillar, but penalized in the moth. Such a 

 disparity has the effect of making the young 

 different from the adult, ending in some cases 

 in an elaborate metamorphosis. 



It would seem that a polygon of frequency 

 would move along a survival curve based upon 

 the measurement of some one character until 

 it reaches its summit, when it would remain 

 stationary. Periodic selection would then 

 ensue. This maintains the species at a fixed 

 point and decreases variation. One might at 

 first suppose that when a new structure be- 

 gins, the survival curve would start at and 

 move along with the increase of the structure 

 to a maximum. But it must be remembered 

 that this is impossible, for the height of the 

 theoretical curve must always have an aver- 

 age height of at least one, no matter what the 

 characteristic used in the construction of the 

 curve may be, unless the species is decreas- 

 ing in abundance. Furthermore, it can get 

 no higher than one unless the population is 

 increasing in numbers. What we really have, 

 therefore, when there is no change in abun- 

 dance, is a compensatory depression of the 

 whole survival curve, as the frequency polygon 

 moves up its slope. It may be likened to a man 

 climbing a step ladder in a descending ele- 

 vator. His rate of speed might be such that 

 his elevation would not change, but he would 

 have horizontal motion. In this case, there 

 would be a general depression of the survival 

 curve to compensate for the increased survival 

 from the increase of the characteristic in 

 question. Elimination would necessarily be- 

 come sharper at some other point, the weakest 

 point, or at several weak points. But the elim- 

 ination, discriminative with respect to this 

 second weak characteristic, would with the 

 first characteristic be generally distributed. 



