F. M. Turner 
489 
I have paid much attention to the subject of age in these statistics, as the 
subject seems to me important from two points of view. It is known that the 
protection afforded by vaccination against attack is not permanent, bvit diminishes 
with lapse of time. A quantitative estimate of the amount of protection soon 
after vaccination and after the lapse of different intervals of time would be a useful 
guide as to the proper period at which to perform revaccination. Since no 
statistics are available as to the protection against attack, it seems reasonable to 
enquire into the time-decrement of the protection against death and take this as a 
guide. The second reason for enquiry as to age of the patients is the hetero- 
geneity of the figures given above as regards age. If we take the correlation 
coeflficient between vaccination and severity to be a measure of the protection 
afforded by the operation, it seems reasonable to demand that the two groups, 
vaccinated and unvaccinated, should be similar in all respects, so far as that is 
possible of attainment. But this is far from the case in respect to age in the 
above tables. Table IV. shews that the ages of the vaccinated and unvaccinated 
differed greatly, and this would tend to alter the value of the correlation. Using 
the language of multiple correlation there is a very high correlation, among the 
patients treated in hospital, between age and vaccination. That being so, we 
cannot affirm a priori a close correspondence in magnitude, nor even in sign, 
between the total correlation Vyg between vaccination and severity, and the partial 
correlation pyg between the same factors. 
To ascertain the time-decrement of protection I have divided up the cases 
into five age groups and worked out the value of r in each. Nearly all the cases 
who were vaccinated at all were vaccinated in infancy, and the number of these 
who had been revaccinated was small. The assumption that the age of the 
patient measures the number of years that had elapsed between the last operation 
of vaccination and the attack of smallpox is in most cases correct. 
The following shews the number of revaccinated cases admitted : 
TABLE XII. 
Number of Revaccinated Persons Admitted in 1901 and 1902. 
Interval in Years since Revaccination. 
0 — 111 
111—20 
20 and f)ver 
Uncertain 
Totals 
1901 
12 
24 
45 
1 
82 
1903 
34 
79 
161 
2 
276 
Totals 
46 
103 
206 
3 
358 
The figures for 1902 are not complete in that no record was kept of revaccina- 
tion at Gore Farm Hospital. As the cases admitted to that hospital numbered 
565 only, as against 7316 to the two hospitals where the record was kept, the 
omitted cases cannot have been numerous. 
Biometrika iv 62 ■ 
