Full Idea
The friend of universals has an account of similarity relations as relations of identity and partial identity; the friend of modes must regard similarity relations as primitive and irreducible.
Clarification
'Modes' is Heil's term for the more usual 'tropes'
Gist of Idea
A theory of universals says similarity is identity of parts; for modes, similarity is primitive
Source
John Heil (From an Ontological Point of View [2003], 14.5)
Book Reference
Heil,John: 'From an Ontological Point of View' [OUP 2005], p.157
A Reaction
We always seem to be able to ask 'in what respect' a similarity occurs. If similarity is 'primitive and irreducible', we should not be able to analyse and explain a similarity, yet we seem able to. I conclude that Heil is wrong.