November 29, 1912] 



SCIENCE 



759 



60 does it become further reduced by empha- 

 sis on the importance of correlation, rather 

 than the direct usefubiess of the characteris- 

 tic in question. 



Of course, it is xmfortunate to be obliged 

 to find that so much of importance in evolu- 

 tion is not to be found in the more easily 

 understood and investigated work of direct 

 selection. It seems like a step backward, to 

 be obliged to conclude that not only do we 

 know less than we thought we did, but to also 

 have important avenues of investigation either 

 closed or made very difficult. It bewilders us 

 to realize that the most fundamental factors 

 of evolution lie in the abstruse, deep-lying 

 problems of the origia of variations and their 

 correlations. A gleam of promise lies in the 

 recent induction of variations by Tower and 

 MacDougal. Epochal as these results are, 

 they may be said to be more the evoking of 

 variations than the making of them. The 

 greatest promise for light on the origin of 

 correlation probably lies in the study of in- 

 ternal secretions and in the transplantation 

 of tissues and organs. 



This position in reference to correlation has 

 its significance for the guidance of further 

 experiments on survival values. I do not be- 

 lieve it is necessary to pick out obviously 

 adapted characters for selection experiments. 

 In proper analysis with large numbers, most 

 causes of death will probably show discrimi- 

 nation with most characteristics. We may, 

 therefore, expect to get appreciable results in 

 aquatic animals by altering salinity or car- 

 bonic dioxide content or muddiness, and in 

 terrestrial animals, by altering temperature, 

 humidity, etc. On the other hand, there 

 should be care in selecting the species with 

 reference to the following conditions : 



1. The animal should be readily obtainable 

 in great numbers. 



2. The sexes should be readily distinguish- 

 able. 



3. The adverse condition should be capable 

 of being applied evenly to any desired degree 

 with the least possible alteration of other con- 

 ditions. 



4. The species should have an allied variety 

 or species living in the new environment, 

 which differs from it in the characteristic 

 chosen. 



5. The characteristic should be one easily 

 measured or counted and show a good range 

 of variation. Measurements should be ordi- 

 narily expressed in percentage of length. 



6. The characteristic should not change with 

 age, or at least not within some adult size unit 

 to be used. 



7. A further advantage lies with a species 

 capable of being successfully bred. It may 

 become desirable to get light upon some point 

 by this method. 



Of course natural selection is determined 

 not by the death rate at any one catas- 

 trophe, but by the end result of the various 

 vicissitudes that all the individuals of a spe- 

 cies pass through from the laying of the egg 

 to the completion of reproduction. It is 

 therefore dangerous to conclude that the nat- 

 ural selection seen in any of these experi- 

 ments, which all express the result of a part 

 only of the life of the individuals, was actu- 

 ally producing evolution in these species. Of 

 course, diagrams such as these must not be 

 expected to give the entire evolutionary status 

 of a group of individuals. We may have such 

 a condition as shown in Fig. 17 and still have 

 the species stationary, for the advantage in 

 one direction may be much reduced or wholly 

 lost by selection in the opposite direction in 

 some other catastrophe or critical period of 

 life. Thus, long antennse were favored in the 

 pupse of Philosamia cynthia. It is quite pos- 

 sible that short antennse might be favored in 

 the imago, for conditions are so dissimilar. 

 Slenderness was an advantage to sparrows in 

 a blizzard, but compactness might be in dodge- 

 ing hawks. The lethal selection may also be 

 counteracted by sexual or fecundal selection, 

 modification of the germ plasm, or ortho- 

 genesis. For a time any one of these influ- 

 ences may even force the species in the oppo- 

 site direction to that in which lethal selec- 

 tion is impelling it. 



The relation to orthogenesis is especially 



