Full Idea
If a resembles b, in general, they resemble in certain respects, and fail to resemble in other respects. But respects are uncomfortably close to properties, which the Resemblance theory proposes to do without.
Gist of Idea
Resemblances must be in certain 'respects', and they seem awfully like properties
Source
David M. Armstrong (Properties [1992], §1)
Book Reference
'Properties', ed/tr. Mellor,D.H. /Oliver,A [OUP 1997], p.163
A Reaction
This is a good objection. I think it is plausible to build a metaphysics around the idea of respects, and drop properties. Shall we just talk of 'respects' for categorising, and 'powers' for causation and explanation? Respects only exist in comparisons.