Ethel M. Eldbrton and Karl Pearson 
561 
Both of these vahies may be considered significant and negative, and hence 
when the attack-rate is constant there is a sensible, if not very close relationship 
between increased isolation and reduced general mortality from diphtheria. 
This confirms the view already reached that while isolation is associated with 
higher attack-rate its effect is to lessen the number of deaths whether they be 
reckoned as case-mortality or general population death-rate. 
(6) What is the meaning of the Association between Isolation and increasing 
prevalency of Diphtheria'? The analysis of this problem is more complicated. 
The obvious answer of those who advocate increasing isolation would be that 
it has been adopted in those districts where the disease is most prevalent, and 
this of course may turn out to be correct. But we may ask in turn upon what 
statistics they depend to demonstrate their view that isolation lessens the preval- 
ence of the disease and is therefore advantageous, if our data demonstrate that 
where there is more isolation, there there is more diphtheria ? It can only be by 
an analysis of no simple character that it is possible to deduce from such data 
that the practice of isolation has lessened the amount of the disease. 
There is, however, a preliminary problem to be dealt with. The isolation -rate 
has been increasing very sensibly from 1904 to 1912, the attack-rate has lessened 
although very slightly, the case-mortality has lessened and the mortality on the 
population is considerably less. These facts are exhibited in the following table : 
Variate 
Symbol 
Means 
Standard Deviations 
1904-1908 
1909-1912 
1904-1908 
1909-1912 
Attack-rate per 1000 population 
A 
1-33 
r28 
■657 
•639 
Isolation-rate per 100 attacked 
I 
42-4 
55 ^7 
25^52 
25^18 
Mortality per 1000 population 
M 
•174 
•138 
•080 
•061 
Mortality per 100 attacked 
m 
14^6 
12^1 
5-72 
5^01 
Now it may well be, since the attack-rate has changed so little, that in the 
towns with increasing attack-rate there has been increasing isolation, both 
quantities changing with the time, but having no causal relation the one to the 
other. It is of some interest therefore to consider the type of districts in which 
isolation is most practised. In the first place we ask if any known bad social 
conditions are associated with prevalence of diphtheria. We took as our measure 
of sanitary conditions (i) the infant death-rate, or the deaths of children under one 
year per 1000 births, (ii) overcrowding, or the percentage of the population in 
private families with more than two in a room. We found the following results: 
First Period Second Period 
Variates Correlated 1904-1908 1909-1912 
Attack-rate and Infant Death-rate -•206±^074 -•206 ±^072 
Attack-rate and Overcrowding ... -•153±^075 — •136±'074 
