by Eric Rauchway

all narrative forms are 'fictions' and so are liesThe New Yorker
When you undertake historical research, two truths that sounded banal come to seem profound. The first is that your knowledge of the past--apart from, occasionally, a limited visual record and the odd unreliable survivor--comes entirely from written documents. You are almost completely cut off, by a wall of print, from the life you have set out to represent. You can't observe historical events; you can't question historical actors; you can't even know most of what has not been written about. What has been written about therefore takes on an importance that may be spurious. A few lines in a memoir, a snatch of recorded conversation, a letter fortuitously preserved, an event noted in a diary: all become luminous with significance-even though they are merely the bits that have floated to the surface. The historian clings to them, while, somewhere below, the huge submerged wreck of the past sinks silently out of sight.
The second realization that strikes you is, in a way, the opposite of the first: the more material you dredge up, the more elusive the subject becomes.... One instinct you need in doing historical research is knowing when to keep dredging stuff up; another is knowing when to stop.
[W]e can't wait to run out of sources before ending a line of inquiry. We can't even pretend to do so. Instead, we have to focus on how much new information we are getting, on average, from what we're learning. Following the work of CS Peirce and William James, Gregory Bateson famously defined information as "the difference that makes a difference." When it stops making a difference, it's no longer information.

But there remains a slippery point, here--how do you know when there's no more, how do you know the next folder won't contain the document that changes the story? I am myself cautious to the point of neurosis about calling time on my research, not least because of my own experience. Trips to the archives are finite--you have only so much money, and only so much time, and only so much patience. You pay for as much airfare and as much hotel as you can. Then you go, and you spend as much time in the archive as they will let you. (I generally skip meals during the hours when the archives are open, so I can stay glued to my seat. Neurosis, I told you.) On one trip, some years ago, I thought I had reached the point of severely diminished returns, and was flipping more or less idly through the remaining material during the last hour of my time at the archive. For the most part the letters and notes had been organized chronologically, and I had gone well past the period that interested me. Then I found a folder outside the chronological experience, which the letter-writer had apparently himself set apart, labeled "Correspondence relating to the n matter," or words to that effect. With, now, 50 minutes left I had to race through this highly pertinent material. This was my encounter with the ever tempting prospect of The Smoking Gun, the one folder or one document that will Change Everything. As I've found, sometimes such folders or documents actually exist. But I've come also to conclude, you can't let them drive your research beyond the finite limits that let you get on with life. Otherwise you make yourself crazy. So how do you know when to stop? Here's Menand, again:

You stop when you feel that you've got it. The test for a successful history is the same as the test for any successful narrative: integrity in motion. It's not the facts, snapshots of the past, that make a history; it's the story, the facts run by the eye at the correct speed.

This is similar to, if maybe a bit more satisfying than McNeill's explanation, as quoted by John Gaddis as quoted by Turkel:

I get curious about a problem and start reading up on it. What I read causes me to redefine the problem. Redefining the problem causes me to shift the direction of what I'm reading. That in turn further reshapes the problem, which further redirects the reading. I go back and forth like this until it feels right, then I write it up and ship it off to the publisher.

Both Menand and McNeill are relying heavily on that non-intellectual feeling. But Menand, I think correctly, ascribes it to your sense of narrative: Did you get the story? It's not just your gut directing you; it's your whole experience of stories from your earliest bedtimes down to the present. You can stop collecting data when you can begin confidently to talk about your subject by starting "Once upon a time," and proceed at a plausible pace through a satisfying middle to "The End."