Alice Lee 
533 
There is, however, a much graver criticism to be made of Dr Newsholme's 
- method in this measure of segregation. He proposes to correlate 
lOO</)£/0 and 1000</>/P, 
and interprets the high correlations as a sign of the value of segregation in 
reducing the phthisis deathrate. We have not his data to test his conclusions by, 
but we can compare them against certain results for 38 years in (i) England and 
Wales, (ii) Scotland, and (iii) Ireland. Here they are : 
Correlation of Phthisis Deathrate and Ratio of Institutional 
Phthisical to all Phthisical Deaths. 
Years District Correlation 
1866—1903 Scotland - -9815 ± -0040 
1866—1903 England and Wales --9750 ±'0054 
1866—1903 Ireland --8720 ±-0262 
1876—1905 Sheffield - -80 ±'0443 
1884—1904 Salford - -67 ± -0811 
The reader may imagine in this table a confirmation of Dr Newsholme's 
results, for the larger material gives higher values of the correlations. On the 
contrary, these correlations have been obtained by taking as the measure of 
segregation the ratio 
Mean Institutional deaths per annum from phthisis 1866-1903 
Annual Total deaths from phthisis 
Now it is clear that this index never varies with the increasing percentage 
of institutional deaths from phthisis. Yet all the correlations are greater than 
Dr Newsholme's ! We have little doubt that he would get higher values than he 
has done, if he replaced the actual institutional deaths per annum by the constant 
mean value. In other words the results reached by him are of no significance, 
for we get higher correlations by putting a single fictitious value for the annual 
institutional deathrate. 
The real source of his result is not the strong influence of segregation on 
phthisis, but the spurious correlation introduced by using the phthisis deaths, ^, 
in the numerator of one variate, 1000<f)/P, and in the denominator of the other, 
lOO0i/<^. Thus no scientific results of value can be found from Dr Newsholme's 
second measure of segregation. 
In discussing this second measure of segregation, Dr Newsholme lays great stress 
on the part played by asylums for the insane in segregating the tuberculous. He 
notes that the percentage of lunatics treated privately with relatives and others was 
18'4 in 1859 and fell to 5"5 in 1902, thus marking increasing segregation during the 
period of fall in the phthisis deathrate. He states (p. 274) that : " the deathrate 
from tuberculosis in borough and county asylums in 1901 was 15-8 per cent, of 
the inmates, and over ten times as great as in the general population." Now 
68—2 
