336 T. H. GOODSPEED AND R. E. CLAUSEN 
in every case for each date was calculated and also a mean for the 
period during which the measurements were taken. This last is the 
mean designated under "means for totals" in the tables. It was 
calculated simply by taking the average flower size of all the flowers 
measured during the entire period. On account of the different num- 
bers of flowers measured on the different dates, this calculation would 
differ somewhat from a mean based on equal weighting of each date, 
but such a procedure does not appear to have any particular advantage 
for the purposes of this discussion. 
Table la 
Frequency Distributions for Spread of Corolla in U. C. B. G. ij^^/oi, 
Class 
Centers 
in Millimeter 
Designation 
No. 
Mean 
Coef. Var. 
19 
20 
21 
22 
23 
24 
25 
26 
27 
28 
29 
30 
31 
32 
Means for totals 
7 
25 
28 
7 
6 
I 
I 
75 
26.83dz.O9 
445±-24 
Means for first dates 
I 
13 
25 
25 
8 
I 
2 
75 
28.48rh.O9 
4.iizb.23 
Means for last dates 
I 
3 
8 
II 
15 
5 
10 
10 
I 
3 
3 
3 
2 
75 
24.i9zb.22 
ii.59±-64 
Means for different dates . . 
I 
2 
4 
6 
2 
6 
6 
12 
18 
12 
4 
2 
75 
26.69 dz. 20 
9.77=t.54 
Tables la and lb give the frequency distributions for spread and 
length of corolla, respectively, of 13 22/07 under designations which 
are in part at least self-explanatory. For example the line in Table la 
designated "means for totals" gives the distribution of the means 
of the 75 plants under consideration for the entire number of measure- 
ments taken. Thus for Plant No. i of 13 22/07, 28.73 mm. represents 
the mean size of corolla spread as determined from the measurements 
of flowers on seven different dates from July 28 to September 22, 
and including a few flowers late in the season from the terminal 
inflorescence. In a few cases some open flowers not yet shedding 
pollen have been included in these calculations. An average of about 
40 flowers per plant was measured during this period, not more than 
80 nor less than 25 on any one plant. The "means for first dates" 
simply gives the distribution of means calculated from the flowers 
measured on the first date and likewise the "means for last dates" 
gives the distribution of means calculated from the flower measure- 
ments on the last date. The distribution designated "means for 
different dates" is one such as might be obtained if a large number 
of plants were measured, each on a single date; i. e., if a worker began 
measuring the flowers on his plants at a certain date and gradually 
