The Second Signature

The rule that two signatures are required on a release, rather than one, is older than anybody now working in the department, and because it is older than they are it has come to be treated as a formality, which is precisely what it was designed not to be. The second signature was never intended as a second confirmation that the first person was honest; it was intended as a second reading by somebody who had not spent the previous three weeks becoming convinced that the thing was right. That is a different question from the one it is now taken to ask.

Absalon, who signed second on perhaps forty releases a year and who had never once refused, would have said, if anybody had asked him before this spring, that his signature meant that the paperwork was complete and that the person who had signed first was somebody he trusted. Both of those things were true. Neither of them was the question the rule had been written to ask.

What made the difference in March was not a failure but an absence: the person who usually signed first was away, the release was urgent in the ordinary sense in which everything is urgent, and Absalon read the file for the first time in the position of somebody who had not already decided, which is the position the second signature exists to create.

He found, in a table that had been copied forward through four versions of the same document, a figure that was internally inconsistent with the paragraph beneath it, and although the inconsistency turned out to be harmless, the plain fact that four people had signed the four versions without noticing it was not harmless at all. It was the reason Renate rewrote the guidance.

The rewritten guidance says almost nothing that the old one did not say, and it does not add a third signature, which had been proposed and which would have made the problem worse by dividing the responsibility further. What it does is state, in one sentence at the top rather than in a paragraph at the bottom, what the second reader is being asked to do — and the department discovered, in the nine months that followed, that a sentence at the top is read and a paragraph at the bottom is not.