LAMARCKIANA SEMIGIGAS. 685 



in the first, and 5 in the fourth, experiment, or 14 and 15 percent. 

 This fact is easily explained by assuming that their parent plants 

 in 1923 were half-mutants for this character and split according to 

 the ordinary rules. The remaining mutants were pallescens, spathulata, 

 liamata in No. 1, oblonga in No. 2 and No. 3, and four auricula in 

 No. 3. Besides those given in our table, there was a cana with 15 

 chromosomes, which produced 154 seedlings, almost all of which 

 showed the type of the parent plants, only two having the marks 

 of Oe. Lamarckiana. Moreover, there was a large amount of barren 

 grains (76 percent). I also sowed the seeds of two specimens of cana 

 linearis derived from the cross of Oe. semigigas x blandina. As was 

 to be anticipated, they produced the type blandina laeta instead 

 of Lamarckiana, and of this 25 and 17 specimens out of a total of 

 66 and 60, and besides these 30 and 39 plants of the type of blandina. 

 The remainder were 8 and 4 cana linearis, 2 pallescens and 1 spa- 

 thulata. These figures are therefore to be considered as a corroboration 

 of the main result. 



Oenothera semigigas Mut. liquida. 



From the cross with velutina, I selfed five plants out of the seven 

 of liquida of a pure type and with 15 chromosomes (de Vries 1924 b, 

 p. 261). There was some difference in their spikes; that of one plant 

 (No. 1) was as compact as in my old race, while the four remaining 

 specimens (Nos. 2 tot 5 of the table) had looser racemes. The same 

 difference was seen among four plants from the cross with tardescens 

 (de Vries 1924 b, p. 224), of which one (No. 6) had a dense spike and 

 three (Nos. 7 to 9) a loose spike. This variation, however, did not 

 seem to have any influence on the genetic constitution, as will be 

 seen from table 3. 



For Nos. 2 to 5 the number of barren seeds has been determined 

 and found to be 61, 75, 86 and 81, or on the average 76 percent. 

 This is rather a high figure, bur for my race of liquida it was almost 

 exactly the same, namely, 78 percent (de Vries 1916 b, p. 261). 

 Besides these, the constitution of my race was, on the average, 67 

 percent Lamarckiana, 32 percent liquida and 1 percent mutants 

 (de Vries 1916 a, pp. 268-269), agreeing sufficiently with the figures 

 of table 3. In the cultures of 1924 the mutants were for a large part 

 oblonga, and besides these, one pallescens, one hamata, one nanella, 

 six pulla, one semigigas and some belonging to types not yet des- 

 cribed. 



