"Apeil 3, 1914] 



SCIENCE 



483 



these in turn as parental forms, from those 

 having a weight of 35-40 centigrams there 

 resulted a progeny with an average of 57.2 

 centigrams, while from those having a 

 weight of 65-70 centigrams a progeny was 

 obtained which had an average of 55.5 cen- 

 tigrams. In other words, selection had not 

 only failed to make any advancement, but 

 actually resulted in a slight retrogression. 

 Facts quite in accord with this but giving 

 much more pronounced results have been 

 obtained by Tower (1906), Jennings 

 (1908), Johannsen (1909) and others. It 

 should be noted, however, that there have 

 been several experiments, notably those of 

 De Vries with buttercups, Tower with po- 

 tato beetles, and Smith with Indian corn, 

 where a possible advance of a character 

 was recorded in a group. Heterozygotes 

 here may have been responsible for the re- 

 sult, although again the explanation may 

 consist in the elimination of the effects of a 

 determiner. 



The results in mixed races as exempli- 

 fied by corn, beans, etc., where selection has 

 gradually improved a group of organisms 

 but finally reached a limit beyond which 

 no progress appeared possible, are compar- 

 atively well understood and are due, as 

 explained by ShuU (1908), to the separa- 

 tion of the pure lines which were present 

 in the race at the beginning. This is where 

 the average agriculturist, horticulturist, 

 and animal breeder has gone far astray and, 

 having succeeded for a few generations in 

 making progress, has failed to understand 

 why he may not continue to be successful. 



Thus we find that attempts to modify a 

 character by selection within pure lines 

 within a small number of generations have 

 almost universally failed, and that the few 

 apparent results to the contrary must be 

 looked upon with the suspicion that the 

 population was a mixed race and that 

 Mendelian principles applied. 



Once again we are led to propound with 

 still greater emphasis the question, "How 

 then has evolution taken place ? " "In what 

 manner have organisms acquired their char- 

 acters ? " " Is it possible to escape the diffi- 

 culties that confront the investigator on 

 every side 1 ' ' 



III 



The application of statistical methods 

 to problems of biology has provided and 

 will continue to provide facts of decided 

 value obtainable in no other way. Never- 

 theless, the use of data "en masse" unco- 

 ordinated with experimental methods can 

 not solve the riddle of existence so easily 

 as some, at an earlier period at least, would 

 have had us believe. There are, however, 

 investigations which seem fundamental to 

 the problem under discussion and which 

 may well be approached from the statisti- 

 cal side. These relate to the influence of 

 certain factors composing the environment 

 as well as to the part played by asexual 

 and sexual reproduction, corresponding in 

 reality to close and cross breeding, upon 

 variability and size in organisms. 



Some studies undertaken in 1900 in con- 

 nection with the influence of food supply 

 on variability^ based upon the comparison 

 of groups of Chrysanthemum leucanthe- 

 mum L., the common white daisy, as well 

 as Perca flavescens Mitch., the yellow 

 perch, indicated that the difference in vari- 

 ability as evinced by the coefficient of 

 variation for a group with a maximum food 

 supply as compared with a group having 

 a minimum food supply, was extremely 

 small and well within the limits allowed 

 by the probable error. From this the in- 

 ference was that external stimuli played an 

 extremely unimportant part under normal 

 conditions as a cause producing variability 

 in general. 



Attempts were subsequently made to ob- 



3 Science, p. 728, 1907. 



