My mother crossed in the winter of 1959, aged eleven, in a group that left from a village she has named the same way every time. Everything else about her account has moved.
In the version I was told as a child there were four days of walking and a river crossed at night. In the version she gave to a documentary crew in 1997 there were eleven days and no river. In the version she tells now, at seventy-eight, there is a woman who carried her for part of a day, who did not appear in either earlier telling and about whom she is more specific than she has ever been about anything — a green coat, a particular way of speaking, a name she is not sure of.
I have turned this over for more years than it probably deserved, and I would like to work out what I actually think, because the two available positions both seem to me obviously wrong.
The first position is that the account is unreliable and therefore not evidence. This is the position of anyone who wants the crossing not to have happened as described, and it is also, more uncomfortably, the position of a great deal of ordinary scholarly caution. If the details move, the report is defective. That is what a report is: a claim about what occurred, assessable by its consistency and its corroboration, degraded by both.
The second position is that the account is beyond assessment — that it is her truth, that memory is not a filing cabinet, and that asking whether the river was there is a category error and probably a hostile one. I have heard this said, warmly, at gatherings, and I understand why it is said, and I think it concedes everything. If nothing about the account can be wrong then nothing about it can be right either, and a thing that cannot be wrong is not the kind of thing anyone was ever willing to die for.
So neither. What I want to argue, without much confidence, is that a memory maintained by a community for sixty years is a different sort of object from a report, and that the standard epistemology of testimony — reliability, corroboration, the assessment of a witness — is built for reports and does not know what to do with the other thing.
Consider what has actually been happening to my mother's account over six decades. It has been told, perhaps a thousand times, to different audiences with different needs. To children, it was told to be survivable, which is why in my version there were four days: eleven days of walking is not a thing you hand a six-year-old at bedtime. To a documentary crew, it was told to be usable, in a form that would fit a sequence, with the emphases the interviewer's questions produced. To me now, it is being told by a woman who is running out of time and has stopped managing the effect on the listener.
None of those tellings is the raw record retrieved with more or less fidelity. Every one is a performance shaped by what it was for, and the account we have is the sediment of all of them.
That does not make it false. It makes it a different kind of true, and the useful analogy — I offer it carefully, because analogies are how these arguments go wrong — is not the witness statement but the path. A path across a plateau is real. It is also made by walking, adjusted by weather, moved when a slope goes, and no individual crossing of it is authoritative. You would not say a path is unreliable because it is not where it was in 1970. You would say that the thing it is tracking is the fact that people keep needing to get from here to there.
The woman in the green coat appeared, I think, because my mother is now closer to what she was frightened of at eleven than she has been at any point since, and the memory that has become useful to her is the one about being carried.
Now the objection, and I am going to state it at its strongest, because the soft version is what people in my position usually offer and I have come to think the softening is a form of cowardice.
The objection is that this is precisely the reasoning that protects fabrication. Every fabricated testimony in the world has shifting details and every fabricator, when confronted, says that memory is not a filing cabinet. There are people with an active interest in the claim that exile accounts are unreliable, and I have just handed them a paragraph that says the shifting is not a defect but a feature. Worse: my argument has no failure condition. If consistency supports the account and inconsistency also supports the account, then I have described something unfalsifiable and dressed it in philosophy.
I cannot defeat that. What I can do is narrow it, and the narrowing is the only original thing I have.
There is a distinction between the details that move and the details that do not, and it is not random which is which. Across sixty years and a thousand tellings, my mother has never once varied on the name of the village, the season, her age, the fact of leaving, or the two people who did not arrive. Those are load-bearing. Everything that has moved is scenery: duration, terrain, weather, the river. And that pattern — a fixed core with a moving surround — is what you would expect from a memory maintained by use, and is not what you get from a fabrication, which tends to be either rigid, because it is rehearsed, or to drift at the core, because there is nothing there to hold it.
That is a testable claim about the shape of the variation, and it is the one thing here I would defend against somebody hostile. It is much less than I would like to be able to say.
I asked her about the river last year, directly, which I had avoided doing for twenty years because I was afraid of what asking would do. She thought about it for a while and said she did not know whether there had been a river, and that she had been carried, and that both of those had been true for a long time.
She is not troubled by the question. I am the one who is troubled by it, and I have started to think that is a fact about people who inherit an account rather than about people who have one — that the second generation needs the crossing to be a report, because a report can be defended, and we are the ones who will have to defend it after she is not here to be asked.
