NATURE AND CHARACTERISTICS OF INFERENCE 397 



ficance of this question will be better grasped when it is pointed 

 out that all inference involves this paradox : On the one hand, in 

 all inference we advance to something &quot; new,&quot; something hitherto 

 &quot;unknown,&quot; some &quot;new truth,&quot; some item of knowledge &quot;addi 

 tional &quot; to what is given us in the premisses : otherwise there 

 would be no progress, no advance of thought at all, no inference. 

 But on the other hand, since in all logical inference the conclusion 

 follows necessarily from the premisses, it must be somehow or 

 other already contained in those premisses : otherwise we could 

 not get it out of them. Now, these two characteristics of infer 

 ence the &quot; novelty &quot; of the conclusion as compared with the pre 

 misses, and the &quot; necessity &quot; with which it follows from them seem 

 to be mutually incompatible. 



If we interpret the former characteristic to mean that the 

 conclusion must be something entirely &quot; new &quot; and &quot; additional 

 to &quot; the premisses, do we not deny that all those inferences which 

 we describe as &quot; formally or logically necessary &quot; are inferences at 

 all : seeing that in these the conclusion is recognized to be some 

 how contained or involved in the premisses, so as to follow neces 

 sarily from the latter ? And if we interpret the second characteristic 

 to mean that the conclusion must so necessarily follow from the 

 premisses that it must be seen to be actually and explicitly con 

 tained in them, do we not render all such &quot;necessary&quot; inferences 

 absolutely worthless as means of making progress in knowledge : 

 seeing that in all such inferences the conclusion cannot be any 

 thing &quot; new,&quot; but must be merely a part of what was already con 

 tained in our knowledge of the premisses ? 



J. S. Mill, who is an exponent of the Empirical school of 

 philosophy, seems to have adopted the extreme interpretations 

 just suggested. For he teaches, as we have seen, that the formal 

 process of thought, by which we pass necessarily from the pre 

 misses to the conclusion of the syllogism, is not a process of in 

 ference at all, but a mere record of an inference already made : 

 that the real inference consisted in &quot; collecting&quot; the universal pre 

 miss of the syllogism &quot;by induction&quot; from previous particular 

 facts : that the syllogism itself, if regarded as a process of infer 

 ence, is always a petitio principii^ since the conclusion which it 

 purports to prove is already assumed, being contained in the uni 

 versal premiss. 



While we, on the other hand, admit, with Mill, that the first 

 Source and starting point of all our knowledge is the particular 



