emisses of a syllogism, because in
the major premiss there is already asserted what is afterwards asserted
in the conclusion. Mr. Balfour's reply is, that 'So long as in fact we
do assert the major premiss without first believing the conclusion, so
long will the latter be an inference from the former.' Now, Mill's
express contention is that we never do assert the major premiss without
first believing the conclusion; and the dispute resolves itself into one
as to the proper meaning of 'inference.' Mill is at this point guarding
against erroneous conceptions of proof; his thesis being that the
'proof' of the conclusion is not given in the major, but in the body of
evidence on which that is founded, and which carries the conclusion at
the same time. As the kind of syllogism in question is the old one
about the mortality of Sokrates, Mill here takes as 'proof' the evidence
which all men now reckon sufficient to establish the fact of universal
human mortality, though, as aforesaid, it is not literally a complete
'proof' at all. Mr. Balfour is arguing, if anything relevant to his main
thesis, that a so-called 'inference' which is merely a statement in one
particular of what is believed of all such particulars, is a 'real'
inference, and therefore somehow more valid than inferences not so
drawn. Perhaps he does not mean this: if so, the argument has no bearing
on his main case.

Concerning 'inference,' the proper development of Mill's position would
be that the processes of reasoning properly to be so called are either
hypotheses still to be tested or beliefs held by the tenure of
uncontradicted experience. And inferences of the latter kind are in fact
of the most various degrees of certainty. We 'infer' that we shall all
die, not from the generalisation that all men are mortal, but from the
accepted fact that all men hitherto have been. The major premiss in the
typical syllogism is itself the inference. But we also infer, from a
much narrower experience, that inasmuch as pitchblende, say, has been
found to yield radium in certain very small quantities, other
pitchblende will do so in future. Here the certainty is distinctly less:
few men would wager heavily on it. And we may at once grant to Mr.
Balfour that in this and many other cases 'scientific beliefs' fall far
short of 'certainty,' as that term is established for us by other
beliefs. As Mill put it, inference from particulars never can be
formally cogent. He might have