Full Idea
Propositions about 'some' may be proved empirically, but propositions about 'all' are difficult to know, and can't be proved unless such propositions are in the premisses. These aren't in perception, so forgo general propositions, or abandon empiricism?
Gist of Idea
Perception can't prove universal generalisations, so abandon them, or abandon empiricism?
Source
Bertrand Russell (An Inquiry into Meaning and Truth [1940], 5)
Book Reference
Russell,Bertrand: 'An Inquiry into Meaning and Truth' [Penguin 1967], p.86
A Reaction
This is obviously related to the difficulty empiricists have with induction. You could hardly persuade logicians to give up the universal quantifier, because it is needed in mathematics. Do we actually know any universal empirical truths?