372 StatisticaJ Studrj of Anti-Typ1ioi<l Tiwcnlation 
TABLE VII. 
India. 1/3/06 to ^SjljOI. 
Iiiooiiliiti(in 
+ 
Recoveries 
7 
669 
676 
Deaths 
1 
101 
102 
Totals 
8 
770 
778 
/i = - 2-315823, ^•=1-121288, 
•01027/-' + •44556/-' + -18709;" - 1 •29835/-2 + /• - -010804 = 0, 
/= -01 10 ±-0725. 
TABLE VIIL 
Seven Large Indian Stations. Jann.arij to June, 1907. 
luociilation 
+ 
Recoveries 
Deaths 
12 
3 
131 
42 
143 
45 
Totals 
15 
173 
188 
h=-\ -406510, k = -708378, 
- •01003/-5--10598/'^--08122r'i- -4981 7/'H»-- -071686=0, 
r= -0745 ±-1469. 
In both cases the coefficient of correlation is positive but negligible in comparison 
with the probable error. It will be observed that both these tables are from 
Indian figures, and although the Stations Abroad show a much lower correlation 
between inoculation and freedom from attack yet the death-rate in the inoculated 
and attacked is nil. I have therefore omitted this table from my series for the same 
reason as is given in the case of the English figures mentioned above. This result 
would rather tend to strengthen the supposition that some of these latter cases 
are para-typhoids, for the case-mortality is much lower in this disease than in true 
typhoid ; on the other hand it must not be overlooked that the total number of 
cases is so small that a death is hardly to be expected. 
It is of interest to note that although these two tables give a zero correlation, 
yet both the groups from which these figures are obtained give a sensible 
positive correlation in respect to inoculation and freedom from attack. This opens 
up an important question in the study of immunity. May a process which gives 
no protection after attack reduce the chance of attack ? These tables are too 
small to draw any definite conclusion from, and although there is unassailable 
