Effect of Selection Obliterated by Nutrition. 559 



them I chose, after harvest was over, as many of the 

 best as were sufficient to supply the seed for the next 

 sowing. 



From this summary of the general arrangement of the 

 experiment, let us proceed to the description of our se- 

 lection culture with the dill {Anethiuii graveolcns, V\g. 

 117). 



For this experiment seeds were bought from the 

 trade and sown in 1892 over 1 square meter, llie cnjp 

 consisted at harvest time of 56 plants. The number of 

 rays in the umbels varied between 12 and 38 and, as a 

 rule, in direct proportion with the vigor of the ])lant. 

 Six plants with 12-16 rays in the terminal umbel were 

 chosen as seed-parents and their seeds sown in 1893 over 

 an area of 8 square meters. The number of plants when I 

 came to select them was 541, and as was to be expected 

 in so much larger a number the minimum and maximum 

 were further apart; 9 and 43. Five plants with 10-13 

 rays in the terminal umbel, were chosen as seed parents 

 which constituted a notable advance in the minus direc- 

 tion, as compared with 1892. 



In 1894 the culture extended over 6 square meters 

 and the number of plants when I came to select them was 

 162. 



The table on page 560 gives, for each of the three 

 crops, the number of individuals whose terminal umbel 

 had the number of rays given in the first column. 



The figures (p. 560) show that, in spite of the fact 

 that each year plants were chosen with a markedly smaller 

 number of rays than the mean of the group from which 

 they were chosen, the mean number of rays clearly in- 

 creased during the experiment. The better manuring had 

 therefore more effect than the selection of weak plants. 



