22 



BOTANICAL GAZETTE 



[JANUARY 



them, but the percentage figures are very variable. I found them 

 as follows: 



The reciprocal crosses cannot produce any oblonga, since this is 

 assumed to be suppressed in the pollen. The only exception is 

 O. scintillans, which gives rise to a high amount of oblonga after 

 self-fertilization, and therefore may produce the same mutant after 

 a cross. I found as follows: 



The pollen of O. oblonga must produce, after crosses with differ- 

 ent species, only velutina, as is easily seen from our formula. No 

 oblonga and no laeta are to be expected. O. biennis, O. syrticola, 

 O. Cockerelli, and O. Hookeri fecundated with O. oblonga uniformly 

 gave this result. The reciprocal crosses, however, must give a 

 splitting, the laeta hybrids assuming the characters of O. oblonga. 

 The percentages should be about 50, but in my experiments there 

 was much fluctuation in this respect. I found as follows: 



