intervals. For the computer now on the desk, at the end of this process, is made up of modules yi... yio, while the computer which sat on the desk ten months ago was made up of the entirely distinct modules xi... xio. If these are the same computer, then by OC there are worlds u and v where, respectively, the given computer is made up ofxi... xio as long as it exists and yi... yio as long as it exists. But there is also the world w where, amongst other things, there are two computers, one made of yi... yio as long as it exists and the other of xi... xio as long as it exists, both machines existing simultaneously in different places. We may therefore pose the question: (3) Which, if either, of w’s computers is the one on the desk in the actual world? No answer to (3) is consistent with reasonable views about identity, for whatever answer we the following will be true:
2013
But it was the supposition that the computer now on my desk is the same com puter as the one which sat on my desk ten months ago that led to this quandary; hence it is that supposition which should be rejected.11 For reasons that I have urged in detail elsewhere,12 we are from this point inevitably led to the perdurance construal of persistence for computers and for anything else for which the Sorites set-up makes sense. For it is unsatisfactory arbitrarily to partition the class of possible com puter kits (10-membered collections of possible modules each of which, at some world, a hobbyist can assemble into a computer) into equivalence classes under the relation ‘assembles into the same com puter as’. And presumably, as a m atter of logic, there is no finite lower bound on the interval of time required to effect a change in a computer of a sort iteration of which eventually yields a computer with no modules in common with the one on which the first change of the series was effected. So as a m atter of logic, the R-relation that is going to settle the facts of persistence must be taken to have only instantaneous thing-stages in its field.13 (I note that the strategy of this argum ent avoids the objections of E. J . Lowe to attem pts to construct tem poral Sorites analogous to modal ones [Lowe 1986]. For only O C is required to turn the tem poral p a ra dox into a m odal one, so if we agree, as Lowe does, that there is a m odal paradox, we can hardly follow him in denying that tem poral cases like that of the com puter are just as puz zling.) I claim that all this amounts to a a good argum ent for the perdurance construal of persistence. And we have yet to see how the endurance theorist can do as well.