FACTORS INFLUENCING FLOWER SIZE IN NICOTIANA 337 
worked through the lot during a period corresponding to the one 
considered here. This distribution was obtained in this particular 
instance by grouping the means for the first 25 plants on the first dates, 
the second 25 on a mid-date, and the third 25 on the last date together 
in one distribution. Table lb has to do with corolla tube length of the 
same plants, and the same explanations of course apply to it. 
Table lb 
Frequency Distributions for Length of Corollas in U. C. B. G. 13^^/07, 
Class Centers 
in M 
illimeters 
Designation 
No. 
Mean 
Coef. Var. 
38 
39 
40 
41 
42 
43 
44 
45 
46 
47 
48 
Means for totals 
I 
3 
6 
34 
28 
2 
I 
75 
44.27i.O7 
2.I3zt.I2 
Means for first dates . 
I 
5 
29 
29 
10 
I 
75 
45.6Ozt.O7 
i.96=t.ii 
Means for last dates . 
2 
2 
10 
17 
9 
15 
8 
10 
I 
75 
42.25zt.i5 
4-58±.25 
Means for different 
dates 
2 
5 
6 
4 
7 
8 
24 
13 
3 
2 
75 
43.96 =t. 18 
5.i8=t.29 
The period during which measurements were taken was from July 
28 to August 30, although a few of the plants were measured again on 
September 22. Measurements of flowers borne among the developing 
seed capsules on the terminal inflorescence and also of open flowers 
which had not yet begun to shed pollen, when such were taken, were 
separately designated in the records. These classes of flowers show 
significant differences in flower size, as will be shown later on in this 
paper, and a haphazard selection of the flowers to be measured might 
easily give undue prominence to some one of these classes. They were 
not however excluded in the calculation of the total mean for each 
plant, but the error introduced in this manner is very slight on account 
of the small number measured in proportion to those fully open and 
borne on the later flowering laterals. 
Certain difficulties and in some cases the arbitrary nature of the 
groupings are quite evident. As has been noted the measurements 
were taken in connection with a number of separate experiments and 
obviously the dates chosen for making the measurements were dictated 
by convenience rather than with the present use of the data in mind. 
The reason for the groupings will be explained in the pages that follow 
and the fact that, notwithstanding the above situation, the data 
bring out the facts to which we desire to call attention is sufficient 
justification for any slight discrepancies which may be apparent. 
Table la brings out the fact that the mean spread of corolla limb 
