He followed an unchanging rule: there had to be at least three or four confirmations from sources he could trust, both about the event itself and about the people involved—including those connected to this or that episode. Human memory, he believed, was a very unreliable instrument. Over time, deformation is inevitable; few people avoid it. Perhaps fearing that, he treated everything written with such harshness, constantly strengthening it. “The written must lie until it ripens,” he used to say. He didn’t consider the novel ready for publication, but two large chapters—“In K.riger” and “An Evening in Levendorf”—were published in magazines as separate novellas, with every right to exist on their own.