.346 
T. H. GOODSPEED AND R. E. CLAUSEN 
Turning now to a consideration of these data, we find the situation 
somewhat different from that in the preceding tables. For the spread 
of corolla, the size is greater at the end of the period than at the 
beginning. As shown in Table IVa flowers on the "mid dates" of 
measurements are larger than on the "first dates," but smaller than 
on the "last dates" — there is a progressive increase in corolla spread 
during the period of measurement. The differences in size in both 
cases for the three measurements are within 2 mm. and differ from 
the mean of the "means for totals" by a still lesser amount. On the 
other hand, with reference to length of corolla in both Tables IVb and 
Yb, the means for the first, mid, and last dates show a slight increase 
followed by a more marked decrease. This point will be discussed in 
a different connection below. 
Table VI 
Influence of the Age of the Plant on Corolla Measurements 
Difference between 
nDesignation 
spread . . . 
length. . . 
13 Fi H137: 
spread . . . 
length. . . 
13(11) FiH 
spread . . , 
length . . . 
asFiHss: 
spread . . . 
^length. . . 
13 Fi H44: 
spread 
, _ 53±-38 
length 54.05dz.26 
55 
38.; 
48 ±.09 
6o±.07 
33±-ii 
,22d=.09 
IOzir.31 
.IOzh.22 
.8i±.3r 
.i9±-i5 
36.67 ±.44 
55.oo±.i8 
36.10rb.43 
54.10zb.29 
39.05 =b.40 
55.43i.15 
40.16zb.44 
54.53±-28 
24.19zb.22j 
42.25=b.i5| 
3i.67zb.26^ — i.66zb.45 
53.44zb.28 -I.22zb.2I 
33.10zb.22l -3.oo±.53 
52.60zb.29 -3.00zb.37 
40.05 zb.35 
53.81 ±.24 
39-32 zb.64 
52.84zb.49 
+o.24d=.5i 
+ 0.24zb.2I 
+ i.63±.58 
+o.48zb.39 
-5.00 zb. 51 
-1.561b.34 
-3.00zh.49 
-1.50zb.41 
+ 1.00 =t. 53 
-I.62zb ' 
- .84zb.77 
-1.69zb.57 
— 4.29db.24 
-3.35i.17 
-6.66dz.28 
-2.78dz.30 
— 6.oozb.38 
-4.50i1.37 
+ i.24zb.47 
-1.38dz.28 
— .79±.74 
— 1.21^.56 
In Table VI the situation with reference to the differences in size 
of flowers borne on a plant at the beginning, toward the middle, and 
near the end of the flowering season or the period during which flowers 
were measured is summed up. This table is self-explanatory. The 
probable errors of all constants have been included. These differences 
should be significant if they are at least three times as great as their 
probable errors. This requirement is fulfilled in all cases except for 
13 Fi H38 and 13 Fi H44, an explanation for which we have suggested 
above. 
