The peopled record

The second volume replaced a static picture of representation with a record of what authors do: every act written down in the order it was performed. Then it proved a defect in itself -- one record, shared by all, credits every author with everything anybody has entered on it. This volume repairs that, and adds almost nothing to do it. Each act now records which earlier acts it was performed on, and from that one note comes an author's view: the part of the shared record he is entitled to -- what he entered himself, and whatever that rests on, back as far as it goes. Not a second record, but a restricted way of reading the one there is. The rest follows. What an author may be credited with and what he may not. How a retraction can leave a man resting on ground he was never told the name of, and the single act that puts it right. Take the addition away and every result of the first two volumes stands as it stood. Every claim here was put to a machine, and searched for the counterexample that would sink it, before it was proved.