THE FOUR EXPERIMENTAL METHODS. 439 



going three methods require ; we cannot ascertain, by those 

 methods, what portion of the phenomena exhibited by any 

 body is due to the heat contained in it. If we could observe 

 a body with its heat, and the same body entirely divested of 

 heat, the Method of Difference would show the effect due to 

 the heat, apart from that due to the body. If we could observe 

 heat under circumstances agreeing in nothing but heat, and 

 therefore not characterized also by the presence of a body, we 

 could ascertain the effects of heat, from an instance of heat 

 with a body and an instance of heat without a body, by the 

 Method of Agreement ; or we could determine by the Method 

 of Difference what effect was due to the body, when the 

 remainder which was due to the heat would be given by the 

 Method of Residues. But we can do none of these things ; 

 and without them the application of any of the three methods 

 to the solution of this problem would be illusory. It would 

 be idle, for instance, to attempt to ascertain the effect of heat 

 by subtracting from the phenomena exhibited by a body, all 

 that is due to its other properties ; for as we have never been 

 able to observe any bodies without a portion of heat in them, 

 effects due to that heat might form a part of the very results, 

 which we were affecting to subtract in order that the effect of 

 heat might be shown by the residue. 



If, therefore, there were no other methods of experimental 

 investigation than these three, we should be unable to deter 

 mine the effects due to heat as a cause. But we have still a 

 resource. Though we cannot exclude an antecedent altogether, 

 we may be able to produce, or nature may produce for us, 

 some modification in it. By a modification is here meant, a 

 change in it, not amounting to its total removal. If some 

 modification in the antecedent A is always followed by a 

 change in the consequent a, the other consequents b and c 

 remaining the same ; or vice versa, if every change in a is 

 found to have been preceded by some modification in A, none 

 being observable in any of the other antecedents ; we may 

 safely conclude that a is, wholly or in part, an effect traceable 

 to A, or at least in some way connected with it through 

 causation. For example, in the case of heat, though we can- 



