536 
Tuberculosis and Segregation 
problem must be dealt with afresh, and the modern methods of partial correlation 
and variate difference correlation applied to its various aspects. We have taken 
the period used by Dr Nevvsholme, 1866-1903 inclusive, and have used the figures 
for each individual year thus obtaining 38 entries, which are few indeed, but the 
best vve can probably do with data of this kind, and therefore directly comparable 
with Dr Newsholme's results, for he seems to have used individual years for his 
correlations although he does not always say so (cf. pp. 271 and 280), and notwith- 
standing that his tables are all given for five-year periods. 
The population numbers for England and Wales (Table A) were taken from 
the Registrar-General's Annual Report for 1909, and the phthisis deaths from the 
Reports for 1866-1903 ; the average of each five years' period agrees with Dr 
Newsholme's values for phthisis, but the values for indoor and for total paupers 
do not quite agree with his. Dr Newsholme was therefore written to and asked 
whence he obtained his numbers. He was kind enough to reply, but said that 
he was unable to refer at the moment to the original tables, but that undoubtedly 
the data were the statistics given in the Annual Reports of the Registrars-General 
for England, Scotland and Ireland. We then examined the Local Government 
Board returns and found that Dr Newsholme apparently had used the pauper 
returns for the January quarter of each year. We kept therefore to the Registrar- 
General's Report, as the numbers there given are based on the Local Government 
Board's returns for the whole year, which are a fairer measure of pauperism than 
those for the January quarter alone. 
For Scotland, our numbers (Table A) agree with Dr Newsholme's for both 
phthisis and indoor paupers, except when we take the first five-year period 
(1866-70), where they differ slightly. In the case of total paupers for the periods 
1866-70, 1881-85, and 1896-1900 our figures do not agree*. We cannot find 
any reason for these divergences except a slip in his or our arithmetic, or the 
possibility that a wrong number of outside paupers has been taken by one or other 
of us. We do not think the differences in the values are such as to invalidate 
a comparison of results. 
In Ireland the only serious discrepancy in our values is in the total number 
of paupers for the period 1876-80. 
These discrepancies, however, emphasise the very necessary rules for statistical 
treatment : (i) that the ultimate raw data should be pubhshed with every inquiry, 
and (ii) it should be stated exactly where they are taken from, and how they have 
been treated. 
Table A gives our raw data, Table B our deathrates and indices based thereon. 
We have correlated the phthisis deathrate taken as 1O°0/P with lOOpi/p^ and 
* We are unable to compare bis and our data for individual years, because Dr Newsholme has only 
published his data for five-year periods. 
