The Operation of Nutrition and Selection. 539 



at random two samples of seed, each of wliich liad been 

 well mixed, from those harvested in 1890. One was 

 sown on two beds of about 2 square meters each ; which 

 had received a dressing composed of 5 kilos of horn- 

 flour, and 10 kilos of this manure per square meter 

 respectively. This culture gave 123 healthy plants with 

 ripe fruits. The greater part of the seed was sown on 

 five beds, of 2 square meters each, some of which re- 

 ceived no manure at all, whilst others were given % to 

 2% kilos of ordinary guano per square meter. Of this 

 crop 14-15 plants were chosen at random from each bed 

 and employed for determining the second curve. 



The total length of the five lower fruits was meas- 

 ured in millimeters (see p. 532) and the mean length of 

 fruit calculated from this. 



In this way I obtained the data in the table on p. 538. 



The following figures can be calculated from the 

 first generation of this experiment, raised from the same 

 lot of seeds : 



Qt M Q, 



With horn meal 1.3 25.2 1.5 



Without horn meal 1.9 27.2 2.4 



Hilversum 1.9 24.1 2.0 



In the third row I have written the corresponding 

 values for O. Lamarckiana (1893), for the sake of com- 

 parison (see pp. 530, 531 and Fig. 115). 



The cultivated plants have, it will be seen, a slight 

 advantage over the wild ones, which is smaller in the 

 case of the plants which had horn-flour than in those in 

 the control experiment. The horn-flour culture shows a 

 slight decrease in the amplitude of variation; the control 

 experiment a trifling increase. 



The horn-flour culture was the only one which was 



