Ethel M. Eldbrton and Karl Pearson 
559 
less than the mean attack-rate (I'SO), but they have a case-mortality of 19-04 as 
against the average case-mortality of 13-26; the 17 towns* with no isolation at 
all give a case-mortality of 19-4. It would thus appear that the towns with little 
or no isolation are those with a lower average attack-rate, but with rare exceptions 
their case-mortality is high. 
Isolation-rate. 
0 10 20 30 40 50 60 70 80 90 100 
Diagram II. 
To test the influence of these towns with little or no isolation, we have 
removed the column 0 — 10 isolation-rate group and recalculated and r},i,j ; we 
find 
= - -4120 ± -0484, r},,,^ = -4810. 
Thus while the correlations are somewhat reduced by excluding the towns with 
little or no isolation there is still in the towns which do isolate a very sensible 
relation between the degree of isolation and the case-mortality, and this relation 
exhibits rather more skewness. 
We may sum up as follows : The relation between greater isolation and a 
lessened case-mortality appears to be a real one. We have shown that it is hardly 
due to spurious correlation, as this would have produced a positive correlation and 
further no great changes are made when we correct for inequality in the numbers 
* South Shields (1st and 2nd Periods), Sunderland (1st and 2nd Periods), Barrow (1st Period), 
Preston (1st Period), Wigan (1st Period), Smethwick (1st and 2nd Periods), Walsall (1st and 2nd 
Periods), West Bromwich (1st and 2nd Periods), Coventry (1st and 2nd Periods), Barnsley (1st and 2nd 
Periods). Of these towns West Bromwicli in the 1st period had the highest ease-mortality recordeu of 
any of our 80 towns, while Smethwick in both periods, and Coventry and Barnsley in the 2nd period 
with no isolation had case-mortalities below the general average. 
