287 



upon this analogy we found an inference of causation ; the point 

 of analogy is between the frequent and the invariable, consequently 

 the analogy must be established or presumed upon, in proportion 

 to our experience of the frequency of the succession of like conse- 

 quences to like antecedents. These grounds of the inference of a 

 cause as is just stated are imperfect, and must admit frequent error, 

 for we cannot define what number of successions of like conse- 

 quences to like antecedents, are an adequate number to prove 

 causation. 



11. Hence then, although we infer causation from succession, 

 we are obliged to confess that we can do this only in certain cases; 

 before we can admit the truth of an inference of causation, we must 

 have had an experience of a sufficient frequency of a like succes- 

 sion. Different men will hold different opinions with regard to 

 what constitutes a sufficient frequency, and the want of a possible 

 definition in this matter admits a great diversity of opinion upon 

 important points, and gives room for the distinction of close, and 

 loose reasoners. 



12. But when once we have had experience of what is con- 

 sidered a sufficient frequency of like succession, we then infer 

 some difference (where it is not perceptible) in cases in which the 

 same consequences do not succeed the same antecedents. In such 

 instances, we balance an account between like, and dissimilar sue- 

 cession; and we assign a cause only, where the frequency of the 

 same succession (approaching to the invariable) exceeds that of 

 the exceptions. Thus (not to quit our subject), if the exhibition 

 of a particular medicine should be followed by recovery from 

 phthisis pulmoualis in one instance, this succession would, where 

 men are disposed to catch at straws, indicate a possible causation; 

 if the same event succeeded to its exhibition in ten instances, its 

 credit would be better supported; if in a hundred, better still. If 

 it should succeed in five and fail in five, we should hesitate perhaps 

 to assign it as the cause of recovery in the first five; if afterwards 

 it should fail in fifty cases, we should say that in the five in which 

 it was followed by recovery the cure was owing to other causes. 

 If it should succeed in a hundred and fail in fifty, we should then 

 perhaps judge the hundred to amount to an adequate number to 

 establish the relation of the medicine, as a cause of recovery ; 

 while we should explain its failure in the other fifty, by supposing 

 some diversity of circumstances, by which its relation as a cause 

 was modified, to have prevailed. The conclusion amounts to this : 

 we infer that a secondary is produced by a primary disease, upon 

 an experience of a frequent succession of the one to the other^ 

 provided at the same time that our experience furnishes us with 

 no stronger analogies to sensible causation, by which we are rather 

 justified in considering them distinct. 





