Here’s an odd asymmetry. We can have past tense epistemic modals, as in (1), that are true even though we now know the embedded sentence is false.

(1) It was possible that the invasion would come from the north.

Imagine that the invasion came from the south (as had been planned for several months) and we are trying to justify why we diverted resources to the northern front. I think in such a case we might use (1), and use it truly.

Now imagine that we have discovered that in two days time there will be a giant tidal wave that will kill everyone. We also know that we are the only ones to know this. Further (and this is where the story gets odd) we know that all our geological knowledge will be magically wiped at midnight, so that very soon no one will know that humanity is doomed. I don’t think in that case (a very intuitive case!) we can say (2).

(2) It will be possible that humanity will survive until next week.

Final case before I quit for the night. (Why haven’t you quit already? I was watching the late show of A Clockwork Orange – very suitable midnight movie I think.) Imagine we have a community of infinitely many insectologists, each with a serial number. Each of them knows two things about ants. First, they each know that there are some ants. Second, insectologist number n knows that it is not the case that there are exactly n ants. I think that if they really all are part of the speech community, (3) is false.

(3) It is possible that there are finitely many ants.

I fear that if I start talking about infinite speech communities, my linguist friends will laugh at me, or worse still no longer be my friends. It’s hard being a philosopher sometimes!