But a single case like this does not bring one very far on the way to deciding the question as to whether there are persons who are genuinely and successfully guided to the discovery of subterranean water by strange sensations and by spasmodic movements of their limbs or of hazel-twigs held in the hands, due (as they declare) to an obscure influence which emanates from subterranean water and from buried metal. The fact is that we have in the belief in the guidance of the douser by occult influences a troublesome case of the fallacy in reasoning expressed by the words, "post hoc ergo propter hoc," or, to put it in English, "after this, therefore caused by this." Primitive man found that this mode of forming a conclusion very often led to a correct discovery of the connexion between two events, and he adopted it as a ready method of guidance, although it was frequently fallacious. It has taken ages, literally ages, to make people discard this mode of arriving at a conclusion in serious matters, and it is still usual in less vital affairs. To show that B followed upon the occurrence of A, even once, is, of course, a proper and useful way of forming a guess or a suggestion as to the cause of B, but still more is your guess legitimate if the sequence has occurred several times in your experience. But it is only a guess: a conclusion must not be accepted on that basis, although lazy and hasty people do adopt such conclusions. You must find out the details of the nature of A and also of B, and if possible how the one is connected with the other. And if you cannot do that you can still establish your conclusion and confirm your guess by showing that B _invariably_ follows upon A, or that (in a long experience) only when A has been present, and never when A has not been present, has B occurred. If you cannot prove the truth of your guess by this experimental demonstration of the exclusion of other causes than A or by the experimental demonstration of the invariable occurrence of B after A has occurred, then you have to seek for evidence of a real connexion between A and B, though not an invariable one, by collecting a vast number of instances of the occurrence of B and finding out whether A has preceded it in such a large proportion of cases (as compared with those in which B has occurred without the previous occurrence of A) that the cases in which B follows A cannot be considered as accidental, but indicate a real causal relation of A to B.