564 
A Study of the Ejfects of DixMheria Isolation 
It will be seen at once that this conclusion opens up new problems : (i) Is the 
greater isolation the outcome of greater incidence, the only remedy suggestable 
for greater incidence being a more complete isolation ? (ii) Is the greater 
incidence in some manner a result of the greater isolation, and does it really tell 
against isolation as a remedy against the spread of diphtheria ? The association 
of greater isolation with local prosperity would then be merely a measure of the 
economic capacity of the district for carrying out the accepted sanitary code, 
(iii) If (ii) is to be answered in the negative, then is there any factor in 
prosperit}^ which makes for a greater diphtheritic incidence ? The final answers 
to these problems can probably not be given on the basis of the present data. 
The correlations under discussion although significant are not of such a marked 
character as to provide more than provisional statements, or indeed more than 
suggestions for further inquiry and tabulation. 
(7) Does greater Isolation follotu increasing Incidence, or greater Incidence 
follow increasing Isolation? The problem is a much more subtle one than appears 
at first sight. What we have established is that those towns with the higher 
isolation-rate have the higher attack-rate. It does not follow from this that the 
individual town which increases its isolation-rate will increase its attack-rate. 
To determine whether this is so we took as our variates : increase in isolation- 
rate (/) between the periods 1904-8 and 1909-12*, and the similar increase A of 
the attack-rate. We found : 
rAi= + -256 ± -072, 
a value probably significant, although not quite so large as that found for the 
inter-urban relation : 
r,ij= + -427 ± -063 (for 1904-8), 
= + -290 ± -069 (for 1909-12). 
We can, we think, therefore conclude that the towns which increase their isolation- 
rate are those with increasing attack-rate, just as the towns with higher isolation- 
rate are those with higher attack-rate. 
But this does not answer the question as to which is "the cart" and which 
"the horse"! Does the increased attack-rate precede or follow the increased 
isolation-rate ? To answer this question we divided our material into three 
periods each of three years, let us say T^, and T^. Then the attack-rate 
increase between T, and was correlated with the isolation-rate in T^, and the 
isolation-rate increase between and with the attack rate in T^. In other 
words we asked whether towns with most rapid increase of attack-rate in the 
* That is the total number treated in hospital x 100 and divided by the total number of attacks was 
taken for the first period and for the second period, and their difference (second period — first period) 
was treated as increase in isolation-rate. In the same way the sum of the totals attacked for the years 
of the first period x 1000 and divided by the sum of the calculated intercensal populations for the same 
years was treated as the attack-rate, and the difference of second and first period values taken as the 
increase in the attack-rate. 
