The wp_posts table seems to retain all revisions of the same, ok very similar but presumably different, posts/pages/whatever.
I'm somewhat conversant with SQL but not WordPress. I need to extract just those records which would appear on the public facing site; so just the most recent revision, and not all the superceded rows. Not sure how to filter the fields. Obviously something more complicated than:
select * from wp_posts where post_status in ('publish','revision') order by post_modified desc
which has 'duplicates' and seems to miss some stuff.