TricotyluHs Half Races. 391 



through the course of four generations (1891-1894), 

 and found it heritable to the extent of from 1 to 2%, 

 but I have not made any detailed records. 



Spinacia olcracca. The spinach also contains tri- 

 cotylous seedlings. I employed the Dutch spinach with 

 smooth round seeds without thorns, a fine and perfectly 

 constant type. In 1892 I found a tricotylous specimen, 

 and grew this amongst some normal i)lants, because the 

 species is dioecious. When the harvest was first exam- 

 ined there proved to be five tricotyls and one hemi-tricotyl 

 amongst 1000 seedlings, i. e., 0.6%. Then the best tri- 

 cotyls v^^ere selected from several thousand seedlings ; 

 thirteen being met with. During the flowering period, 

 several of them proved to be monoecious, a phenomenon 

 which sometimes occurs in this species and has been ob- 

 served also in Cannabis sativa, Mercttrialis annua, and 

 others. I harvested the seeds separately from 5 female 

 or monoecious plants, and obtained ratios of 0, 0, 0, 1 and 

 2% tricotyls amongst the seedlings, of which only 4 

 managed to flower. They were one male and 3 female 

 plants which latter set an abundance of seed. They gave 

 ratios of 0, and 2%, the latter occurring amongst 430 

 seedlings. In the summer of 1895 the tricotyls flowered, 

 and I collected the harvests of the various female plants 

 separately, and thiis was able to determine the proportion 

 of tricotyls for each parent in the spring of 1896. This 

 fluctuated between and 3% and in one case reached 

 4% (mean 1.5%). 



Summary: The results of the experiments described 

 show that, in the cases dealt wn'th, a stringent selection, 

 extending over three years, failed to effect any definite 

 and certainly any considerable advance. The individual 

 instances fall into two categories ; as a rule the propor- 



