558 Curves of Compositae and Umbclli ferae. 



given to the Oenotheras in the same year, that is, % Kilo 

 of dried cow manure and % Kilo of steamed horn flour, 

 per square meter. This mixture was spread over the 

 beds as uniformly as possible a few days before sowing 

 and well dug in. I considered the much richer horn flour 

 manure which I had given to the Oenotheras in 1891 as 

 superfluous for this experiment. 



This constant high nutrition supplied during the three 

 years of the experiment would lead us to expect a pro- 

 gression of M ; and the selection a retrogression of M. 



I chose as seed-parents the plants with the smallest 

 number of ,rays in the primary umbel or head. In the 

 Umbelliferae this could be counted before the plants 

 flowered and, inasmuch as the remaining plants were 

 pulled up before the selected seed-parents began to 

 flower, crossing could be prevented. The composites 

 on the same bed did not flower all at once; each plant 

 was recorded as soon as its rays could be counted and 

 pulled up if it was not wanted as a seed-parent. The 

 possibility of crossing was thus diminished as much as 

 possible ; in addition to this, the seed-parents were, when- 

 ever possible, deprived of all the heads that were over 

 or still flowering, as soon as the selection was finished, 

 in order to get only purely fertilized seeds for sowing. 

 Such cultures are often threatened by numerous pesti- 

 lences and misfortunes which only too often sweep away 

 every single seed-parent, after the others have been 

 pulled up. For example in 1894 the experiment with 

 Coriandrum sativum was thus brought to an untimely 

 end. Occurrences of this kind led me to spare a larger 

 series of seed-parents than would otherwise have been 

 necessary. Occasional cross-fertilization could, of course, 

 take place among them as a result of this. From among 



