W. F. Harvey and A. M'^Kendrick 
65 
sion but rather some one quantitative relation of these facts. The improvement 
in the condition of the patient expressed in terms of feeling-of-well-being, the 
check in the spread of disease processes, the extension in the duration of life 
produced by particular methods of treatment have scarcely at all been taken into 
account because they are not available in a form suitable for statistical treatment. 
These data may be grouped under the commonly used term, experience. But 
experience is largely a personal matter and can only carry limited weight in argu- 
ment or at most appeal to the comparatively few. Professor K. Pearson has 
shown how the degree of association of non-quantitative characters can be worked 
out. It is incumbent upon us therefoi-e as medical men to provide material for 
the calculation of coefficients of contingency, as they are called, as well as for 
coefficients of correlation in future considerations of efficacy of treatment and the 
indications afforded by special tests. 
Another argument which enters implicitly or explicitly into medical disserta- 
tion is the argument from analogy. Logicians warn us against the pitfalls of such 
a method, and yet it is, when discreetly used, a most valuable adjunct to evidence. 
Thus the prophylactic value of vaccination against smallpox having been estab- 
lished it seems reasonable to suppose that a similar method would be capable of 
giving some result in other diseases such as rabies, anthrax, cholera, and typhoid 
fever. If the evidence for the therapeutic value of inoculation in cases of furun- 
culosis, erysipelas and acne is accepted we may reason further to the tiial of 
specific vaccines in the treatment of other affections. If improvement in the 
condition of a patient and the development of a feeling of well-being usually 
synchronise with a rise of the opsonic power of his serum we may — until it is 
disproved — infer that there is a causal connection between them. Final! y if over 
a wide range of diseases, tuberculosis, staphylococcal, streptococcal, pneumococcal, 
and gonococcal diseases, glanders, dysentery, typhoid and Malta fevers and others 
there would seem to be accumulating evidence in favour of the arrival of a new 
era of treatment we may reasonably anticipate that these hopes will ultimately 
prove, if properly recorded, to be based on statistically significant foundations. 
But we must discard the fetish of experience, and set to work at the expense of 
some labour to record the essence of that experience so that all the world may 
judge of its real value. The work requires co-operation, and the realisation of this 
fact may speed the advance of exact measurement in the domain of medicine. 
In the following treatment of the subject-matter of this paper I have en- 
deavoured to indicate some of the aspects from which this recently much criticised 
quantitative measurement may be viewed rather than to deal with it exhaustively. 
The phagocytic count affords a peculiarly suitable case for figure analysis. We 
have here a typical frequency distribution, that is to say, the numerical degree to 
which a given character is possessed. This is most conveniently shown by stating 
the actual number of individuals, in a sufficiently large and unselected sample, 
that possess the different gradations of the cliaracter in question. The individuals 
Biometrika vii 9 
