Stout & Boas: STATISTICAL STUDIES IN CICHORIUM 391 
year-old F;'s and after the first sixty days for the one-year-old 
Fəs. In other words, most of the decrease has taken place during 
eighty days in the first case and during sixty in the second. It is 
quite obvious then, when purely mathematically considered, that 
there should be a large negative correlation between rate of de- 
crease and length of blooming period. 
2. SIGNIFICANCE OF THE RANGE OF VARIABILITY IN FLOWER 
NUMBER PER HEAD 
A question which arises in this connection, and which appears 
to be of considerable biological significance, is whether the total 
amount of decrease or the actual range in variation in flower 
number throughout the season bears any relation to the highest* 
flower number in the plant. In other words, do plants with high 
flower number show a larger total decrease than those with low, 
or is the amount of decrease about the same in each? The ques- 
tion was suggested by the comparison given above of the data 
for different periods of bloom, when it became evident that the 
variability during the later periods was lower than that of the 
earlier. 
The following tables, one for three-year-old F, plants and one 
for one-year-old F; plants (the same used above) will perhaps best 
TABLE. 2I 
RANGE OF VARIABILITY OF FLOWER NUMBER PER HEAD OF 106 THREE-YEAR-OLD 
PLANTS OF THE Fi GENERATION 
| Minimum flower number | Difference between 
Number of Maximum flower | maximum and 
plants number Range | Average average minimum 
| 
1 30 15.0 15.0 
2 29 I3 I2 12.5 16.5 
3 28 17-14 15.3 22-4 
5 27 15-12 14.2 12.8 
14 26 16-12 14.0 12-0 
8 25 17-13 14.2 зн 
17 24 I5-II 13.8 10.2 
19 23 16—10 12.6 10.4 
21 22 17-11 TR г. 
13 21 16-11 da їй 
3 30 13-11 12.0 Bc 
answer the question. Here the total range from highest to lowest 
flower number has been recorded, the plants with the same highest 
