Full Idea
We would do better to go straight from the evidence that some men have died to the mortality of Socrates, than to go via 'all men are mortal', for the probability that Socrates is mortal is greater than the probability that all men are mortal.
Gist of Idea
The mortality of Socrates is more certain from induction than it is from deduction
Source
Bertrand Russell (Problems of Philosophy [1912], Ch. 7)
Book Reference
Russell,Bertrand: 'The Problems of Philosophy' [OUP 1995], p.45
A Reaction
Russell claims that deduction should stick to a priori truth, and induction is best for the real world. Interesting. To show that something is a member of a set (e.g. planets) you need an awful lot of knowledge of the set.