KiRSTiNB Smith 
275 
represent o-^^^, the squared standard deviation of the arrays of same height, to 
obtain a reasonable description ; this is shown on the diagram. The marginal 
frequencies of the height variate could be expressed fairly well by a Gaussian curve. 
These theoretical values of o\^j and are given in Table XI together with the 
weights 
calculated from them. 
TABLE XI. 
MiUims 
theoretical 
theoretical 
v' 
lip 
observed 
nip 
observed 
»;,/ 
frora 
minimum 
nip from 
least 
squares 
102-25— 104-25 
8-456 
4-73 
4-7123 
2 
5-00 
9-92 
9-99 
104-25—106-25 
8-748 
6-62 
6-3809 
10 
10-40 
10-18 
10-25 
106-25- 108-25 
8-987 
13-89 
130282 
10 
11-10 
10-45 
10-51 
108-25— 110-25 
9-172 
26-80 
24-6339 
27 
11-54 
10-72 
10-77 
110-25—112-25 
9-304 
47-59 
43-1217 
56 
11-71 
10-99 
11 03 
112-25—114-25 
9-382 
77-76 
69-8648 
59 
11-81 
11-26 
11-29 
114-25— 116-25 
9-408 
116-90 
104-750 
115 
11-62 
11-53 
11-55 
116-25-118-25 
9-380 
161-71 
145-332 
142 
11-70 
11-80 
11-81 
118-25-120-25 
9-298 
205-83 
186-597 
244 
11-80 
12-06 
12-08 
120-25-122-25 
9-164 
241-06 
221-744 
265 
12-15 
12-33 
12-34 
122-25-124-25 
8-976 
259-78 
243-960 
261 
12-52 
12-60 
12-60 
124-25—126-25 
8-735 
257-59 
248-580 
265 
12-83 
12-87 
12-86 
126-25—128-25 
8-441 
235-02 
234-710 
219 
12-98 
13-14 
1312 
128-25—130-25 
8-093 
197-30 
205-508 
197 
13-78 
13-41 
13-38 
130-25—132-25 
7-692 
152-41 
167-023 
131 
13-85 
13-67 
13-64 
132-25-134-25 
7-238 
108-33 
126-167 
88 
13-78 
13-94 
13-90 
134-25—136-25 
6-730 
70-85 
88-7361 
77 
14-28 
14-21 
14-16 
136-25—138-25 
6-170 
42-64 
58-2529 
52 
14-40 
14-48 
14-42 
138-25—140-25 
5-556 
23-61 
35-8204 
20 
14-05 
14-75 
14-69 
140-25—142-25 
4-888 
12-03 
20-7416 
16 
14-56 
15-02 
14-95 
142-25—144-25 
4-168 
5-64 
11-4040 
11 
14-95 
15-29 
15-21 
144-25-146-25 
3-394 
2-43 
6-0407 
4 
18-00 
15-55 
15-47 
146-25-148-25 
2-567 
]-49 
4-8835 
1 
19-50 
15-82 
15-73 
The usual regression line is 
= 12-7007 + -130489 {y^, - 124-0467), 
and the line for which is a minimum is 
mj = 12-7071 + -1342345 {y.^ - 124-0467). 
For were found in the two cases the values 44-411 and 44-109 and for the 
'goodness of fit' P the values -0047 and -0051*. 
* A case was purposely chosen in which the regression was known to be far from linear, in order to 
ascertain whether this fact itself would separate at all widely the least square and regression lines. 
