11 



Inference always proceeds upon analogy, it assumes things which 

 the senses do not witness in the present instance, from the 

 similarity of those which they do witness with some which have 

 been perceived before with which the adjunct, the matter of the 

 inference, was found to be connected. In fewer words, we 

 believe things not seen upon an experience of their connection 

 with things seen; or, from analogy which is perceptible in some 

 respects, we infer analogy in others. To multiply our examples 



25. Fire is luminous and hot: whenever we see the former 

 property we infer the latter. Now on what does the truth (by 

 which I would be understood, when I employ this term, that belief 

 which is not likely to be superseded by another), on what does 

 the truth of this inference depend ? If heat is necessary to a 

 luminous body, that is, if a luminous body cannot exist without 

 heat, then we should be right in concluding always their con- 

 nection; but if a luminous body may exist without heat, then we 

 are not certain of the presence of heat in such luminous body 

 until we have examined it by another sense. As a body may be 

 luminous without being hot, we may be deceived in this inference. 



26. When we see water congealed into the form of ice, we 

 infer that it is cold. In this inference we are not likely to be 

 deceived, because we are aware beforehand that ice cannot exist 

 without the coW which we expect to find. We have ascertained 

 that there is a relation of cause and effect between cold and the 

 congelation of the water. Where this relation is known we can- 

 not be deceived, if we infer the cause from seeing the effect. This 

 then is a necessary connection: but there are connections which 

 are not necessary, and it is in our affairs with these that we are 

 so liable to error. 



27. What proof, it will be asked, have we of the necessity of 

 a connection? Briefly, universal experience. But if it should be 

 discovered that there is a property in nature, hitherto unobserved, 

 capable of producing such a congelation of water, in a summer's 

 atmosphere, without diminishing its temperature, we should then 

 conclude that the property producing the congelation of the water 

 was commonly associated with cold; that cold was not essential to 

 this property. Such a property being discovered, our inference of 

 the coldness of the substance, from seeing a piece of congealed 

 water, could not be free from the possibility of error. 



28. But notwithstanding the discovery of a property capable 

 of congealing water, independent of cold, we should be very 

 much disposed to believe a piece of ice to be cold; and as the 

 independent property had not been noticed in the lapse of 

 previous ages, the sense of feeling would confirm our inference, 

 perhaps, every time it was made in the course of most men's lives. 

 And why should we be right so often? Because the association 

 of cold with the property of congelation would be proportionally 

 frequent. 



