Karl Pearson and J. F. Tocher 
in 
or the odds against Liverpool and Birmingham general mortality experiences being 
samples of the same population are gigantic. This test which takes the distributions 
as wholes seems to us absolutely conclusive. The differences in deathrates in 
Liverpool and Birmingham are not due to difference of age constitutions, but are 
fundamental*. 
(ii) Cancer Deathrates in Edinburgh and Dundee, 1891-1900 {Males)'f. 
Edinburgh 
Dundee 
Age Group 
Population 
Deaths, Cancer 
Population 
Deaths, Cancer 
OS 
149,763 
5 
89,775 
6 
5—15 
280,655 
8 
168,510 
2 
15—25 
274,343 
18 
142,917 
7 
25—35 
214,063 
47 
98,953 
10 
35—45 
158,133 
117 
76,132 
47 
45—55 
115,206 
295 
59,066 
114 
55—65 
69,954 
356 
.37,337 
151 
65—75 
32.966 
266 
16,958 
109 
75—85 
10,311 
93 
4,625 
30 
85 and over 
1,045 
8 
535 
3 
Totals 
1,306,439 
1213 
694,808 
479 
Proceeding as before we find = 28-4:837. The factors for the age classes 
to give xo^ from the terms for the separate age groups are given below. They 
differ more from unity, but some being in excess and some in defect, there is no 
substantial difference between Q^- 
We have Xo"" = 28-0393, 
or, xo^ is 1-6 % less than Q^. 
Age Group 
„ , a, a', / 
b actor : — J / 
VP 1 
2 
Age Group 
Factor: 1 
VP / 
{p+p' J 
Kp+p' ) 
0—5 
5—15 
15—25 
25—35 
35—45 
1-03386 
1-03427 
•99360 
•95387 
•96788 
45—55 
55—65 
65—75 
75—85 
85 and over 
•98855 
1-00711 
-98966 
-94318 
•98812 
We now apply the same tests as before: we have 
{W - M)/aM'~M =Q = 5-3370. 
Hence i (1 + a) = -99999,99527, 
* This'only means that the differentiation cannot be accounted for by age differences, it might well 
be accounted for by class or occupation differences. 
•j- The results are sums of ten years' population found on assumption of arithmetical progression. 
