When we speak about “the” cause of a “disease”, in a generalised or conceptual sense, as when we say that Koch’s bacillus is “the cause of tuberculosis”, we are really defining our concept of the disease in terms of _one_ correlative. We are saying that tuberculosis is a disease in which Koch’s bacillus is invariably present. A _circulus in definiendo_ is only just escaped because we happen to know that, if Koch’s bacillus is injected into certain animals, the “disease” as we say, develops. Koch’s bacillus is the one constant correlative found in all cases of the kind that we agree to call tuberculous, by reason of certain clinical and pathological signs that we find. Possibly even this statement is not to be taken as absolutely true; though it represents what we find it convenient to say. But, when we thus declare Koch’s bacillus to be “the” cause of tuberculosis, we have by no means exhausted the study of all the correlations that may be called causal in respect of _particular_ cases. Of ten cases of tuberculosis, each one exhibiting Koch’s bacillus, we may say that _for each particular case_ “the” cause of the illness is different. Thus: A. is tuberculous because he was gassed in France; B. is tuberculous because he was infected by his sick wife; C. is tuberculous because he drank tuberculous milk; D. is tuberculous because he worked in an ill-ventilated factory; E. because he was exposed to wet and cold; and F. because he drank and was dirty.