The Isolation of Tricotylous Interuiediatc Races. 433 



spring" (1893) the proportions 2 — 4 — 7 and 7*/o of tri- 

 cotylous seedlings, counted in lots of from 300 to 500. 

 I only planted out the seedlings of one of the plants with 

 7 Jo, but the experiment was not successful, and only 

 three tricotylous plants gave a sufficient quantity of seed. 

 Their hereditary values were 2 — 8 and S/c. The seed- 

 lings of one of the best plants were planted out in 1894. 

 In this year I had twelve tricotylous plants for which I 

 could calculate the values. With two exceptions, they 

 were distributed between 7 and 19%, but the exceptions 

 attained 2?) and 25%, the mean of the whole group being 

 13%. In 1895 the tricotylous seedlings of the two best 

 parents were planted out in separate groups. The mean 

 numbers of these differed considerably ; but both attained 

 a proportion of 41% as maximum. The parent with 

 23% had produced offspring the values of which were 

 distributed between 7 and 31% (18 plants with a mean 

 of 17% and one plant with 41% tricotylous grand- 

 children; whilst the parent with 25% gave values from 

 15 to 31% amongst 12 offspring, with a mean of 26%, 

 and two specimens with 41%. In conformity with the 

 principles of selection of the grandparents, the three 

 plants with 41% were not regarded as analogous, and 

 only the offspring of the parent with 25% and with a 

 mean, calculated from its grandchildren, of 26% were 

 planted out. The progeny of the third plant with 41 % 

 were considered as of inferior value and thrown away. 

 In the following spring, when the values were cal- 

 culated, no essential difference could be detected between 

 the two grou]:)s of my culture of 1806. In percentage 

 calculated from lots of 300 seedlings these figures were 

 as follows : 



