J. McD. Troup and G. D. Maynard 
403 
1 
Balti- 
a 
more 
m 
am 
n 
P2 
(i) 
(2) 
(3) 
(4) 
(5) 
(6) 
i 
162 
1730 
1730 
10 
0-029 
2 
875 
9340 
18680 
109 
0-629 
3 
1313 
14010 
42030 
245 
2-105 
4 
1570 
16760 
67040 
391 
4-455 
5 
1564 
16690 
83450 
487 
6-901 
6 
1380 
14730 
88380 
516 
8-727 
7 
1056 
11270 
78890 
460 
8-807 
g 
757 
8080 
64640 
377 
8 -394 
9 
510 
5440 
48960 
286 
7-133 
10 
314 
3350 
33500 
196 
5-409 
13 
416 
4440 
57720 
337 
11-855 
17 
52 
550 
9350 
55 
2-461 
22 
31 
330 
7260 
42 
2-300 
Totals 
10000 
106720 
601630 =N 
3511 
69=^ 2 
The probable error of q. 2 may be assumed to be at least as great as that of p 2 , 
which we have found to be 5. Hence it is an even chance that the value of q 2 will 
lie between 74 and 64. The observed value 78 would occur on the average at 
least once in four observations on a similar scale. There is, therefore, on the above- 
mentioned assumption, no clear evidence of house infection in this example. 
Neither the conclusions drawn from the results, nor the results themselves, as 
found by the method adopted by Dr Niven, seem to us to be justified. Dr News- 
holme while approving of the method does not give any mathematical proof of its 
validity, and the results differ considerably from the results found by the method 
described in this paper. 
Although we have been able to eliminate the error arising from variations in 
number of persons to a house, there still remains the disturbing factor introduced 
by variations in age and sex. It is difficult to see how this can be dealt with 
statistically, as it is almost impossible to obtain the necessary information. In some 
diseases it will not be of so much importance as in others, viz. those which are 
most prone to attack the very young, e.g., scarlet fever, or the aged, as, for instance, 
cancer. Statistics of " cancer-houses," if suitably collected and obtainable in 
sufficient quantity, would form a useful and interesting subject for mathematical 
investigation. 
In conclusion, we would observe that while the formulae given in this paper 
have, we consider, a field of practical utility in connection with some public health 
problems, yet the results obtained must be interpreted with caution. The 
unknown factors will tend to increase the value of the probable error, that is to 
say, observed variations from the calculated values may be greater than would be 
expected, without justifying the conclusion that house infection, for instance, is 
a factor in their causation. 
Pretoria, March 26, 1911. 
51—2 
