DEPARTMENT OF EXPERIMENTAL EVOLUTION. 143 



the same inbred race constantly selected for the highest numbers of 

 extra bristles is in the forty-second generation. Since the sixth gener- 

 ation no advance has so far been determined that can be interpreted 

 as the result of the selection. 



To select in the opposite direction and so produce a race with few or no 

 extra bristles. — It has, so far, been impossible to accomplish this end. 

 Two series of return selections were made and in each of these the 

 progenies from the low-grade parents were as high as those from the 

 high-grade selections. Since the flies with few extra bristles are apt 

 to be the smaller ones, after a few generations of selecting the lines 

 become very weak, and the two series of selections were lost by the 

 failure of the selected flies to produce large enough families from which 

 to make the required selections. 



To establish a race of low-grade extra-bristled jiies by starting with 

 extras that appeared in the F2 of a cross with normals. — This has been 

 accomplished. An inbred race that has fewer extra bristles than the 

 high selected hue has been obtained by this method. This difference 

 between the two lines is clearly shown, even when both are bred under 

 conditions as nearly identical as can be obtained. The year's work 

 seems, then, strongly to support the conclusion that "selection" as a 

 method of changing the germ-plasm is futile; further support is also 

 afforded the hypothesis, outlined before, that the seeming response to 

 selection shown in the first few generations was due to a sorting of 

 factors, rather than to the modification of a single factor. If there is 

 any significance in these conclusions, the cross between the high-grade 

 and low-grade lines should provide a vital link in the evidence; but 

 pressure of other researches has not yet permitted this. 



SELECTION OF BEAN-PLANTS IN NATURE. 



Dr. Harris has continued his experiments on selective mortaUty in 

 nature of bean seedhngs, and he planted out this year about 26,000 

 seedlings selected during the germination of about half a milhon. It 

 is known that the death-rate is differential; the object of current work 

 is to determine as accurately as possible the selective value of specific 

 morphological features. 



SELECTION OF FERTILITY AND FECUNDITY IN PLANTS. 



These studies have been carried along by Dr. Harris in connection 

 with other work as usual. Further series of data confirming results 

 on fertility in Cercis stated in previous Year Books have been pub- 

 lished. An extensive investigation of the relationship between the 

 number of pods per plant and the characteristics of the pod in garden 

 beans — an analysis of the data for 127,610 pods from 19,064 plants 

 grown in 32 experimental cultures made under a wide range of environ- 

 mental conditions — has also appeared. It is shown that the correlations 

 between the number of pods per plant and the number of ovules per pod 

 in the pods produced have always been found positive in sign, but 



