538 
Tuberculosis and Segregation 
TABLE I. 
Correlation of 
with 
Crude Indices 
Ai 
A., 
A3 
A4 
A5 
Afi 
A7 
As 
England with Wales 
lOV/P 
h 
+ 
•946 ± 
■090 ± 
•201 ± 
•335 ± 
■407 ± 
•475 + 
•538 ± 
•584 ± 
•614 + 
•012 
•134 
•149 
•153 
•155 
•153 
•149 
•145 
•143 
10V/-P 
•966 ±^007 
■258 ± •126 
•461 ±-123 
■508fl27 
■518fl36 
■528fl43 
•543 ± •U? 
■ •562±^150 
•587+ •ISl 
It 
+ •930 
+ ^340 
+ ^542 
+ ■567 
+ •547 
+ •529 
+ ^583 
+ ^539 
+ ^557 
+ 015 
•120 
•110 
•116 
•130 
•142 
•151 
•156 
•159 
Scotland 
•952f010 
•265 ± -126 
•240fl47 
••205± -164 
•186±-179 
•182+^191 
Ir 
+ •920 + 
+ ^250± 
+ •182 + 
+ •086 + 
+ •024 + 
-•003 + 
Ireland 
•017 
- ^881 + • 
•127 
-•280+^ 
•151 
- ^264 + • 
•170 
- -226+ • 
•185 
- ^182+^ 
•198 
- -145+ • 
- •112+^ 
•024 
•125 
•145 
•163 
•179 
•194 
•206 
It 
+ •893 + 
+ •235 + 
+ •180 + 
+ •162 + 
+ •133 + 
+ •108 + 
+ ^081 ± 
+ -044 ± 
- ^004 + 
•022 
•128 
•151 
■167 
■182 
•195 
•208 
•219 
•230 
It will be seen from this table that whether we use the index Ii or its 
inverse 1^, we get practically the same results — naturally with changed sign. 
But the results themselves are of extraordinary interest. For both Scotland 
and Ireland, when we proceed to annul the time-factor by correlating successive 
differences, we find that the high correlations interpreted by Dr Newsholme as 
marking a relation between pauper segregation and phthisis deathrate entirely 
disappear or become less than their probable errors. There is thus no organic 
relation between these variates as measured by the above indices. In the case 
of England and Wales, however, while there is a reduction on annulment of the 
time-factor to roughly two-thirds of the high value noted by Dr Newsholme, 
this value does not tend to disappear with increasing differences. Thus in 
England with Wales, as apart from the remainder of Great Britain, there would 
at first sight appear to be an organic relation between segregation of paupers 
and the phthisis deathrate. But our first column under the England with Wales 
section shows that if we fix the percentage in the general population of these 
indoor paupers and then annul the time-factor, we reach a slightly higher value 
of this apparent organic relation. It has therefore nothing to do with segregation. 
Thus Dr Newsholme's interpretation of his original high correlations appears in 
every case fallacious. 
There are two methods of testing this result, i.e. the absence of organic 
relationship between indoor pauperism and phthisis. Suppose we correlate the 
crude numbers of phthisis deaths per annum and of indoor paupers per annum, 
the resulting coefficient will have very small logical value because both these 
variates are continuously changing with the time*. But now suppose we annul 
* It is noteworthy that the England with Wales and the Scotland correlation coefficients for these 
crude variates are high and negative, but for Ireland the coefficient is moderate and positive. Thus 
the factors at work must be totally different in the two Islands. Since indoor paupers relative to 
the population have remained singularly constant the increase of phthisis deaths must have been much 
slower than the population increase in Great Britain, but somewhat faster in Ireland. 
