Raymond Pearl and Wilbur N. Fuller 
225 
possible to deal with a variety of problems which could not before be attacked by 
biometric methods on account of inherent peculiarities of the data. 
At the outstart the following contingency table (Table VIII.) between length 
of worm and extent of clitellum was formed : 
TABLE VIII. 
Length. Class unit 2'5 cm. 
o 
C- 
c+ 
D- 
D + 
E- 
F- 
F+ 
Totals 
|6| 
5 
35 
43 
1 
129 
120 
2 
1 
8 
1 
38 
8 
1 
379 
3 
1 
72 
8 
2 
|6: 
|7| 
2 
14 
23 
19 
3 
6 
3 
1 
|7; 
: 7 • 
|8| 
|8; 
: 8 : 
6 
4 
2 
1 
2 
1 
2 
1 
1 
14 
1 
1 
Totals 
7 
49 
73 
155 
135 
53 j 
9 
1 
482 
This gives a table of 80 sub-contingencies, a comparatively large number for 
this method. Following the notation of Pearson in the memoir referred to, we 
find for the square contingency from this table 
101-5385, 
whence the mean square contingency 
(^==•2107. 
The mean square contingency coefficient 
1 + <ji' 
4172. 
hence equals 
From the same grouping 
whence, interpolating by the diagram in Pearson's Table I., the mean contingency 
coefficient 
C, = -34. 
It is at once evident that these values do not approach very closely the 
relation C'i = C*„, which should be the case if the proper grouping is used in the 
Biometrika iv 29 
