436 Tricotylous Races. 
plants of these groups and the result in the following 
spring, 1898, was 23 to 65%, with a mean of 40% among 
17 seed-parents, and 26 to 75%, with a mean of 47 /r , 
amongst 13. 
Phacelia tanacetifolia, Fig. 87. My race arose from 
the same set of bought seed which included Clarkia, Hcli- 
chrysiun and Papaver. In the summer of 1895 I had 
20 tricotylous plants, raised from bought seed, in flower. 
Their values constituted a two-fold group like that which 
occurred in Papaver. Three seed-parents had 12- -12 
and 14%, but the rest had values between 1 and 10%; 
the three former were alone used as the basis of my 
race. The three cultures derived from them gave 30 6 
and 9 separate harvests, which were evaluated in the 
spring of 1897 in the usual way. The lowest, middle 
and highest values for the three groups were 5 26 58, 
212842, and 61-4 16%. Obviously the two first 
grand-parents had given better offspring than the last 
one. I selected the former group for the continuation 
of the race, employing the two best plants with 54 and 
58%. They gave two groups of tricotylous plants, the 
harvests from which exhibited a great advance on the 
average, but which did not differ essentially from one 
another. With the exception of the extremes these fig- 
ures constituted a closely circumscribed group of 35 val- 
ues distributed between 35 and 72%, with a mean of 
57%. 
The extremes were 20 and 22%, and on the other 
side 80 85 and 90%. The two former figures, which 
were probably the result of incomplete isolation, occurred 
in the same group. The higher figures, however, were 
distributed over the offspring of both grandparents. 
Obviously the mean value of 57% of the intermediate 
