DEPARTMENT OF EXPERIMENTAL EVOLUTION. 131 



tions by which the experimenter might carry his strains definitely in 

 any desired direction. It was the opinion of the great majority of 

 experimentalists that the capacity of improvement by selection was 

 definitely limited, owing to the fact that the number of kinds of genes 

 available in the formation of a given organism is limited. On the other 

 hand, Castle, to refer only to the most striking investigator in this 

 field, accepts the view of unlimited capacity for modification through 

 individual selection. This controversy is now to a large extent settled, 

 chiefly through certain experiments of Castle himself, which have led 

 him to reject his doctrine of contamination of genes, and apparently 

 also of indefinite variability in them. 



Castle's decision is, however, a matter of the last few months only, 

 • and meanwhile Dr. MacDowell has carried through an elaborate series 

 of breedings upon Drosophila to determine whether the number of cer- 

 tain bristles on the back might be increased or diminished by selection 

 of parents with an increased or diminished number of such bristles. 

 As set forth in the Year Book for 1917, Dr. MacDowell reached the 

 conclusion by the ordinary methods of genetical analysis that an indefi- 

 nitely large change in the number of these bristles could not, as a mat- 

 ter of fact, be produced through the ordinary processes of selection. 



Dr. MacDowell has now made a further statistical analysis of the 

 data by means of the methods of correlation. Many calculations were 

 required, and revised tables and charts and the text have been com- 

 pleted under the title, ''Bristle Inheritance in Drosophila III. Corre- 

 lation." The same conclusions are reached as b}^ the ordinary method 

 of genetic analysis. In brief, the correlations indicate that there was a 

 tendency in the first five generations for the bristle grades of the off- 

 spring from high-grade parents to be higher than the bristle grades of 

 the offspring from low-grade parents ; that in no other period of the 54 

 generations did such a clear difference exist betvv^een the offspring of 

 high and low grade parents; and that, when the en\dronment was ren- 

 dered as uniform as possible, no sign of any such relationship was 

 found. In the generations where a difference was found between the 

 offspring of high and low grade parents, it is obvious that the breeding 

 of high-grade parents exclusively would raise the means of the race. 

 The means actually obtained in the different generations of selection 

 experiments show a close connection with the amount of correlation 

 found. The means rise in the generations where the closest correla- 

 tion is found. Selection has been dealing with hereditary units that 

 were present in the original flies; it has not caused the origin of new 

 units; its action has been to reduce the amount of genetic differences 

 between individuals, whereas the primary requirement for evolution is 

 that it should act so as to increase the amount of such differences. 

 Natural selection acts as a stabihzer of evolving races; it has no part in 

 the evolving itself. 



