388 Tricotyloiis Races. 



LxcJuiis fill gens. The tricotylous seedlings of this 

 species are as a rule weakly ; their culture, therefore, is 

 difficult and their harvest poor. In 1892 I had a tri- 

 cotylous plant whose seeds gave a proportion of 5%. 

 From these I reached, in the spring of 1894, a ratio of 

 13% containing one tetracotylous plant; most of the 

 tricotyls afterwards remained ternary. In 1895 they 

 produced tricotyls in proportions varying from 3 to 11%, 

 with a mean of 6%. In the next, i. e., the fifth genera- 

 tion (spring of 1896), I counted from 2 to 8% tricotyls 

 per seed-parent, and from a particular individual 21 tri- 

 cotyls amongst 110 seedlings, i. e., about 19%. But the 

 numl)er of seedlings in this case was too small to signify 

 a real advance. 



Penstemon gentianoides. In 1892 I had raised four 

 tricotylous plants from bought seed. They produced 

 respectively 0.3, 1.0, 2.6 and 3%^ tricotyls in 1893. I 

 planted out the tricotylous seedlings of the best seed- 

 parent, but only six managed to flower. Their seeds gave 

 ratios varying from 4% to 12% (March 1892), with a 

 mean of 7%. The tricotylous seedlings of those seed- 

 parents only which had ratios above 10% were planted 

 out. Of these 8 tricotyls, 6 hemi-tricotyls and 2 tetra- 

 cotyls have flowered. The seeds of the former gave 

 ratios of tricotyls, ranging from to 2>.Z% with a mean 

 of 2.8%; the hemi-tricotyls from 1.27^) to 2.4% with a 

 mean of 4.8% ; and the four tetracotyls 10% and 11%-, 

 amongst which, however, only a single seedling had four 

 cotyledons. The offspring of both of these tetracotyls 

 and of the best of the remaining seed-parents were 

 planted out in 1895. Only in the case of eight plants, 

 however, w^as the harvest a sufficient one and gave a 

 ratio which as a rule was between and 12% and which 



