J. A. Harris 
453 
The probable errors were calculated on the basis of the actual number of 
locules, n = 5000. 
Apparently differences (a) in the vigour of the ovules, or (b) in the chances 
of fertilization, or (c) in the supply of seed forming substances, from fruit to fruit 
tend to render the locules more alike in the number of seeds they produce than 
they are initially in number of ovules*. The discussion of these points must, of 
course, be reserved until all the data can be set forth. 
■ Illustration {III — A a) 3. Intra-ovarial Correlation for Seed Weight. 
Case (Aa) may be further illustrated by a series of seed weights to be 
discussed in a forthcoming paper on Crinum. Ten seeds from each of 200 fruits 
taken from a fine row of plants in the Missouri Botanical Garden in the spring 
of 1906 were weighed. From Illustration 2 it appeared that in Hibiscus the 
locules of the same fruit have a tendency to produce similar numbers of seeds. 
Here the problem is to determine to what extent the seeds from the same 
fruit are alike in weight. The seeds range from 1 — 18 half-gram unitsf. This 
would give asymmetrical intra-fruit correlation table of 18 2 = 3 24 compartments 
and 200 x 10 x 9 = 18,000 entries. The work is very rapidly carried out by my 
formulae. The total weight in units for each of the 200 fruits is obtained by 
simple addition. These, shown in Table III, are of course the class 2 (V)' s around 
TABLE III. 
Total weight of seeds (in half-gram units) in Individual Fruits of Crinum. 
10 
14 
19 
21 
24 
27 
29 
31 
33 
36 
39 
41 
43 
45 
49 
54 
61 
70 
85 
96 
10 
14 
20 
21 
24 
27 
29 
31 
34 
36 
39 
41 
43 
46 
49 
55 
62 
71 
86 
97 
10 
15 
20 
21 
24 
27 
29 
.32 
34 
36 
39 
42 
43 
46 
50 
55 
62 
72 
87 
100 
11 
15 
20 
21 
24 
28 
29 
32 
34 
36 
40 
42 
44 
46 
50 
57 
62 
73 
89 
103 
12 
16 
20 
22 
24 
28 
30 
32 
34 
37 
40 
42 
44 
47 
50 
58 
63 
75 
89 
107 
12 
17 
20 
22 
25 
28 
30 
32 
34 
37 
40 
42 
44 
47 
51 
58 
64 
76 
90 
115 
12 
17 
20 
22 
25 
28 
30 
32 
35 
38 
41 
42 
44 
47 
52 
59 
64 
78 
91 
122 
13 
17 
21 
23 
25 
28 
30 
32 
35 
38 
41 
42 
45 
47 
52 
59 
64 
80 
93 
124 
13 
18 
21 
23 
25 
28 
30 
32 
35 
39 
41 
42 
45 
48 
53 
60 
66 
81 
94 
125 
14 
19 
21 
23 
26 
29 
31 
33 
35 
39 
41 
43 
45 
48 
53 
61 
69 
84 
95 
126 
0 as origin. The 2000 seeds weighed are distributed as follows : 
1 
2 
o 
4 
5 
6 
8 
9 
10 
11 
12 
IS 
u 
15 
16 
17 
18 
219 
396 
352 
290 
212 
147 
106 
76 
55 
39 
27 
26 
29 
13 
5 
6 
1 
1 
* This is also the conclusion reached for Sanguinaria. See Harris, Biometrika, Vol. vn. 
pp. 327—328, 1910. 
t Class 1 = 0-0 — 0-5 gr., class 2 = 0-6 — 1-0 gr., etc. For details see final paper, Annual Report 
Missouri Botanical Garden, Vol, xxiii, 1912, 
Biometrika ix 58 
