356. BOTANICAL GAZETTE [way 
facts we know that all the classes occurring in any individual, fall 
most probably on one part (perhaps on one side) of the curve in the 
racial variation, and that the classes in the individual variation are 
almost continuous. 
rt. NINETY PLANTS COLLECTED AUGUST 7 
Total 
go 
II 
I 
SMP: 
18 16 4| 2 
Total | D 
/ 
2 Io 
rie] sa] s|e]7|e]o|% 
/ 27 
ee 
) 
22 | 15 
a a ie 9 
I] 4 |13| 24 14 LL D-lL.00 
From this point of view, we should take into account the flowers 
in an equal number of individuals, if we wish to compare two col- 
lections in two different conditions, and not an equal number of 
flowers, because in the latter case we would probably obtain a curve 
that declines to one side or the other. For example, when in the 
second collection of rays 1431 heads from 70 plants were taken into 
account, the mean was 17.777; and when in the third collection 
1366 heads from 49 plants were taken, the mean was 17.694; these 
are due undoubtedly to the smaller number of individuals that were 
in the first collection. 
I have examined also whether the shifting of R and D occurs 
with the change of the flowering season. 
2. NINETY PLANTS COLLECTED AUGUST. 11 
lsl4[s [16 |7|8|o|Tom|| p alsl[als[o|7|8| 9 [tol 
Elaty 25 | 28 19 | 5| 3 90 || f | 3] 4 o}a4}18) 11) 7 | 3} 9? 
3. NINETY PLANTS COLLECTED AUGUST 19 
R| 3 s|s|ol7] slo 10 Total || D| 2 s}als|o|7| 8 [o| | | 33] tome 
j|x|2]9|23 3° | 14 7| 4| 9° lf} x] 2|5| 22| 20 II 12 5 ri Phi es 
In the second collection only a slight change occurred, while in 
the third collection the number of different classes in each individual 
increased to 7 (the most probable one), and accordingly the number 
of D shifted from 5 to 6. These increases result from the addition of 
classes of lower value, to which the seasonal change of rays is due. 
Correlation between the number of rays and of disk florets 
The material of the third collection was used in this calculation. 
A glance at the correlation surface in the accompanying table shows 
