432 BOTANICAL GAZETTE [DECEMBER 
frequently beet roots are separated into selects and discards on 
small differences. A satisfactory method of predicting seed yields 
would not only alter these results but would furnish a means of 
obtaining sufficient seed to make several plantings of each family. 
foot 
WH 1905 
Weight of seed per beet iin ounces ins906 
LS HS 5S T4595 1 
Rercenta 
sugar in 
$135 15,5 17,5 12.5 24,5 
ef 
19.8 
Fic. 2—To accompany table IT 
Tables II~XI (summarized on p. 463) and figs. 2-11 show the 
relationships which obtain between these characters. Some of the 
tables show slight irregularities, but they are not sufficient to invali- 
date the results. The effects of the irregularities may be seen in 
Bree. 
te 
i=] 
ma 4 
i) » 2 
be P4 Yield of seed in 1909 
% s< Ounces 
che ’ 
Sw 
<3 Yield of seed in i907 so ft Oo .2 
eR aces 2 
Be 223 
Hoa 2c 2 
pep? 22.8 
"4 
20.55) 23.3 
se al . Be 
2155 23. 
2 24.3 
Fic. 3.—To accompany table III Fic. 4.—To accompany table IV 
the deviation of the dots representing the empirical means of the 
separate classes from the straight line regressions. The divergence 
is not marked except where the numbers are small.. Moreover, 
the general trends of the broken lines connecting the empirical 
means agree fairly well in direction with the straight line regres- 
