Thony Gillies has “a new paper”:http://www-personal.umich.edu/~thony/counterfactual_scorekeeping.pdf up defending a strict implication account of subjunctive conditionals. That is, he says that (1) can be analysed as (2), with the quantifier in (2) being restricted by context, as many quantifiers are restricted by context.

(1) If it were the case that _p_, it would be the case that _q_.
(2) In all worlds where _p_ is true, _q_ is true.

The usual argument against this, tracing back to Lewis, involves what Thony calls Sobel sequences, such as the following examples from Lewis.

(3) If the USA were to throw its nukes into the sea tomorrow, there would be war; but of course, if the USA and all the other superpowers were to throw their nukes into the sea tomorrow there would be peace.

It is possible that both conjuncts of (3) are non-trivially true, but this is not possible on the strict conditional analysis. There’s a few responses one can make to this argument of course. As Jason Stanley has pointed out, quantifier domains can move around fairly quickly; certainly they can be different either side of a semi-colon. So it isn’t obvious the strict implication theory has this consequence. Another response notes that on Lewis’s own account, (4) should be just as good as (3), but it isn’t.

(4) If the USA and all the other superpowers were to throw their nukes into the sea tomorrow there would be peace; but of course, if the USA were to throw its nukes into the sea tomorrow, there would be war.

Now here’s the question I’m getting to. Thony credits this observation to Irene Heim, as reported in a 1999 paper of Kai von Fintel’s. Is that really the earliest source? I thought it was made in the 1996 McCawley paper that Thony cites, if not before – but I don’t have that paper on me so I can’t tell. Anyone out there with an instant encyclopedic knowledge of the history of conditionals who can help?

I should say that I very strongly suspect this point is one that many many people independently discovered, so given the speed (or lack thereof) with which some things get into print, I expect that there will be many good candidates for the honour of having discovered the distinction.

*UPDATE* – It seems I was totally wrong about my historical recollections. See the comments. And of course see Thony’s paper for a nice way to handle the data within a strict implication theory.