OENOTHERA GRANDIFLORA AIT. 223 



second generation along with the trials of the first. This, of course, 

 is the best means of thoroughly comparing the types during the tests. 



The size of the cultures is too small to give reliable proportions 

 for each of them. Their aim is to show that the three types arise 

 from every combination without exception, and that the mutants 

 arise only occasionally. The size of the whole group, however, is 

 large enough to warrant the reliability of the average proportions. 

 In calculating these I have reckoned the gigas plants with the ovata 

 and the lutea respectively, on the ground of their general appearance. 

 The contraria were calculated separately. No other types occurred, 

 especially no lorea, no ochracea, no dwarfs, and none of the special 

 types afforded by later generations. 



Some details may be given now concerning these experiments. 



0. grandi flora x 0. Lamar ckiana. — Cross of 1914 between two 

 specimens of my races. I cultivated 60 offspring until July, but 

 retained only one-half of them during August and September. This 

 half had been planted on a bed in April, before the distinguishing 

 marks were clear. In July I counted 30 ovata, 11 lutea, and 18 brun- 

 nea, but could not distinguish the contraria. For this reason only 

 the results of the counting on the bed in August have been given 

 in table V. The three types were exactly the same as in the other 

 cultures. Of each of them one plant has been fertilized in order 

 to study their second generations in 1916. 



0. grandiflora lorea x Lamarckiana (cross of 1914). — In July I 

 had on the bed 10 ovata, 12 lutea, 8 brunnea, and 3 doubtful ones, 

 and in the box 25 ovata, 9 lutea, and 6 brunnea. All of the first and 

 the larger part of the second culture were annual. I retained only 

 those on the bed and controlled their numbers during August, when 

 they flowered, and then found that the doubtful specimens belonged 

 to the type of contraria. One specimen of each type has been fer- 

 tilized and has yielded a second generation in 1916. 



A second cross was made in 1915, with the second generation of 

 lorea. The culture embraced 60 plants, almost all of which flowered, 

 and which could easily be counted in August. No contraria was 

 observed during the period of flowering, but a mutant gigas appeared, 

 as has been described. I repeated the counting of this group at 

 different periods, in order to be sure that the same figures were 

 obtained. The plants reached a height of 1.5 m. about the middle 

 of August. 



0. Lamarckiana x grandiflora (cross of 1913). — Seeds sown in 

 1915, after a first trial in 1914. Besides the plants mentioned in the 



