Many writers agree that ( 1 ) is importantly different from (3). It is not just that we deny ( 1 ) and affirm (3) - a matter everyone agrees about - rather the explanation of why we deny ( 1 ) and affirm (3) is to be found in an important semantic difference between ( 1 ) and (3). 1 The details vary (needless to say), but there is considerable agreement that the key to the meaning of ( 1 ) lies in the fact that ( l ) ' s truth conditions are to be given in terms of possible worlds in the style of Robert Stalnaker or of David Lewis, or of some reasonable variant thereon.2 By contrast, the key to the meaning of (3) lies in the fact that the justified assertability of (3) is given by the conditional probability of (3)' s consequent given its antecedent, or in terms of some reason­ able variant thereon.3 This approach to (3)' s meaning is some­ times associated with the doctrine that (3) does not have truth conditions, and sometimes with the doctrine that (3) has the truth conditions of the material conditional.4