Since my talk here at ANU last Thursday I have been thinking a lot about how to define 'relevant consequence'. There are various reasons why one might want a notion of relevant consequence (in addition to the standand notion of necessary consequence). For example, various attitudes are closed under some consequence relation. But they are not closed under necessary consequence. For example, 'I believe that snow is white' does not entail 'I believe that Goldbach's conjecture is true'. There is a lot of literature on how to define 'relevance'. One notion which I believe is due to (or at least inspired by) Graham Priest goes like this. q is a relevant consequence p iff q is a consequence of p, and q does not introduce any new non-logical constants (of course, it may be that new non-logical constants are introduced in the derivation of q from p, e.g. if existential elimination is applied). I like this way of defining 'relevant consequence'. But it is too strict for my purposes. I want 'I own a car' to be a relevant consequence of 'I own a Porsche', but 'car' is a new non-logical constant. So, I was thinking the following might do as a definition of 'relevant consequence'. q is a relevant consequence of p iff q is a consequence of p, and if q introduces a new non-logical constant P2, then P2 is a minimal predicate, and for some minimal predicate P1 in P, necessarily, all P1 are P2. Given this notion, 'I own a car' is a relevant consequence of 'I own a Porsche', because it is true that necessarily, all Porsches are cars. I can't think of any obvious counterexamples to this way of defining 'relevant consequence', though I am sure there are some.