The Operation of Nutrition and Selection. 540 



tude of variation Iq = ^^^\ increases in this case but 



not in the former. 



In connection with this result, there are two further 

 points relating to the selection of continuous varia- 

 tions. 



In the first place both agricultural and horticultural 

 selection is usually accompanied with moderate manur- 

 ing, and the most desirable individuals are usually found 

 amongst the strongest ones. Inasmuch as we have here 

 a contrast similar, though working in the reverse wav, 

 to that which obtains in our experiments, we may expect 

 an increase in the amplitude of variation as a result of 

 this contrast. 



In the second place if I had mixed the two races in 

 1894, or if I count the figures in the tables on pp. 545 

 and 547 together and calculate Qiy M and Qo, from the 

 whole lot I get (for 118+101=219 plants): 



Qi ^1 Qz Q= 



Qr+Q^ 



2 

 1894 5-0 28.6 3.8 4.4 



whereas Q for the short-fruited race alone was 



2-7 + 5.2 

 2 ~ 



That is to say, the amplitude of variation is so much in- 

 creased by the opposite action of nutrition and selection, 

 that it can only very slightly increase further by lump- 

 ing the extreme variants in both directions together. In 

 other words Q (in the case before us) is increased far 

 more by the changing conditions of nutrition in the in- 

 dividual plants on the same bed than it can be by selection 

 in two opposite directions. 



Summary. I append the values for 0\, M and 0- 



