Full Idea
Shall we say that natural properties are the ones whose instances are united by resemblance? - Not if we are going to say that resemblance is the sharing of natural properties.
Gist of Idea
We can't define natural properties by resemblance, if they are used to explain resemblance
Source
David Lewis (On the Plurality of Worlds [1986], 1.5)
Book Reference
Lewis,David: 'On the Plurality of Worlds' [Blackwell 2001], p.63
A Reaction
The target of this appears to be the proposal of Quinton. By now I have totally given up on so-called 'natural' properties. Lewis says the circularity (also in Idea 15743) is a reason to treat 'natural' here as primitive (though he rejects that).
Related Idea
Idea 9407 Properties imply natural classes which can be picked out by everybody [Quinton]