GENETICS: GOODSPEED AND CLAUSEN 
335 
of 75 plants of N. Tabacum var. macrophylla at the beginning of the 
blooming period was 28.49 =±= 0.09 mm. and the length 45.60 ± 0.07 
mm. Measurements of flowers taken on an average one month later 
gave a mean corolla spread for the same plants of 24.19 =±= 0.22 mm. 
and for length 42.25 =*= 0.15 mm. For the same plants means derived 
by grouping together the measurements of 25 of these plants early in 
the season, 25 in the middle of the period, and the remaining 25 toward 
the end of the period of measurement gave for spread 26.69 0.20 mm. 
and for length 43.96 ±0.18 mm., which figures lie very close to the 
means for all the plants for the entire season which are 26.83 =±= 0.09 
mm. and 44.27 =»= 0.07 mm. respectively. 
The latter arrangement brings up a practical matter of considerable 
importance in connection with genetic investigations of flower size in 
Nicotiana. In this case although the means were practically identical 
for the tv/o arrangements, the coefficient of variability where each third 
of the plants was measured at a different time was 9.77 =*= 0.54% 
whereas that based on the distribution of means for all the plants for 
the entire season was less than half as great, 4.45 =t= 0.24%. For 
length the corresponding coefficients of variability, 5.18 =±= 0.29% and 
2.13 =t 0.12% respectively, show a closely proportionate correspondence 
in this respect. This higher variabihty is exactly what would be ob- 
tained if a large F2 population were measured progressively, that is 
beginning at one end of the field and working through until the last 
plant had been measured. Such a method of measurement used in 
comparing large F2 populations with relatively small parent and Fi 
populations might very easily lead to erroneous conclusions on the 
critical point of the experiment, namely the extent of increase of varia- 
bility in F2. 
In the case of the semi-sterile hybrids of N . Tabacum varieties with 
N. syhestris another factor enters in, namely the lack of seed production 
and the consequent prolonged blooming period. There is, therefore, 
not so sudden a decrease in corolla size as in the case of N. Tabacum var. 
macrophylla; but nevertheless a marked decrease occurs rather later in 
the blooming season. In the case of spread of flowers of F1H137 there 
was a decrease during the period from August 9 to September 19 from 
38.33 ±0.11 mm. to 31.67 =t= 0.26 mm. For length the decrease was 
less striking, from 56.22 =t 0.09 mm. to 53.44 ± 0.28 mm. Similarly 
in 13 (11) F1H38 there was a decrease in corolla spread from August 4 
to November 7 from 39.10 ± 0.31 mm. to 33.10 =t 0.22 mm., and in 
length from 57.10 =«= 0.22 mm. to 52.60 ^ 0.29 mm. Flowers of UFiH^s, 
genetically the same but a different population blooming the second 
