116 



EDWIN CARLETON MACDOWELL 



selection; and finally that in the last period (generations 32 to 

 49) no permanent rise in the means can be found. 



Extremes of variation 



The range of variation gives further light on the nature of 

 the changes during selection. If the race as a whole is being 

 changed, upper and lower extremes should change as well as 

 the means. In this race the lower limits show very sight varia- 

 tion. In all the generations (excepting the 24th and 25th) the 

 low limit for the male is either grade or 1 ; in 33 generations 

 it is 0; in 12 generations it is 1. The females are a little higher; 



EXTRA BRISTLES 



^IGH RACE 



EANS OF MALES 



GENERATIONS 



I I I LI. 



J \ \ \ \ \ \ \ L 



I I I 



2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 



Fig. 4 Highest grade son and daughter in each of the generations of the high 

 selected race, compared with the means of all the sons and daughters. Selection 

 has not resulted in the raising of the high limit of variation. 



in 20 generations the low limit is 0; in 18 generations it is 1; in 

 7 generations it is 2. From this one may say that the low ex- 

 treme is a relatively fixed point. The high extremes are more 

 variable. These are shown in figure 4. The highest grade 

 recorded (16) was found in generation 8; the highest grade re- 

 corded (13) in the 41 generations of selecting that followed this, 

 appeared in the first generation raised in the constant tempera- 

 ture room. Throughout the whole series there appears a tend- 

 ency for the highest female to be grade 9 and the highest male, 

 grade 7. Selection then has not called into being any new 

 grades that were not obtainable very near the beginning of the 

 experiment. 



