The Rules of Sociological

particularly if the hypothesis that such a- link exists contradicts laws 
already proved, we must set about finding a third phenomenon on 
which the two others equally depend or which may have served as 
an intermediary between the two. For example, it can be estab­
lished absolutely certainly that the tendency to suiCide varies 
according to education. But it is impossible to understand how 
education can lead to suicide; such an explanation contradicts the 
laws o_f psychology. Education, particularly if confined to ­
elementary knowledge, reaches only the most superficial areas of 
our consciousness, whereas, on the contrary, the instinct of 
self-preservation is one of our basiC tendencies. It could not 
therefore be appreciably affected by a phenomenon so remote and 
with such a feeble influence. Thus we are moved to ask whether 
both facts might not be the consequence of one single state. This 
common cause is the weakening of religious tra�itionalism, which ' 
reinforces at the same time the desire for knowledge and the 

---

Rules for the Demonstration of Sociological Proof 153 
tendency to suicide. 
But another reason exists which makes the method of concom­
itant variations the supreme instrument for sociological research. 
Even when the circumstances are most favourable for them, th� 
other methods cannot 
be 
employed usefully save when the number 
of facts to be compared is very large. If it is not possible to find two 
societies which resemble or differ from each other only in one 
single respect, at least it can be established that two facts very 
frequently go together or mutually exclude each other. But for this 
statement to have scientific value it must be validated a very great 
number of times; we would almost have to be assured that all the 
facts had been reviewed. But not only is such an exhaustive 
inventory impossible, but also the facts accumulated in this way 
can never be established with sufficient exactness, precisely be­
cause they are too numerous. Not only do we run the risk of 
omitting· some which are essential and which contradict those 
already known, but we are also not sure that we know these latter, 
which are known, sufficiently well. Indeed, what has often discre­
dited the reasoning of sociologists is that, because they have 
preferred to use the methods of agreement or difference -