The old village is still there, roofless, about two kilometres below the new one. Everybody agrees that the move happened in the 1890s. Almost nothing else about it is agreed. The official account, recorded in a provincial report of 1903, attributes the relocation to flooding. This is plausible; the lower site is on a floodplain and the river has a documented history of overtopping its banks. It is also the explanation an administrator would reach for, because flooding is legible to the state in a way that most reasons are not, and because a relocation attributed to a natural cause requires no further inquiry. Local accounts, collected by a folklorist in 1961 from people whose grandparents had made the move, give three other reasons. One is a dispute with a neighbouring landowner over grazing. One is an outbreak of disease that the villagers associated with the low ground. One is a quarrel within the village itself, following which a majority left and the minority eventually followed rather than remain alone. These are not necessarily rival explanations. A village of two hundred people does not decide anything for one reason. The most likely reconstruction is that the flooding was real, chronic, and tolerable; that it became intolerable when combined with a grazing dispute that made the lower fields less valuable; and that the timing was determined by a quarrel that had nothing to do with either. This is how most collective decisions are actually made, and it is precisely the sort of account that no archive preserves, because none of the three causes was ever written down as a cause. What makes the case unusually well documented is the folklorist's method. She recorded not only what people said but who had told them, and her notes make it possible to trace each version to a family. The flooding account is strongest among descendants of households that moved first. The quarrel account survives almost exclusively in families that moved last. Neither group was lying. Each remembered the reason that had applied to them. This is a general finding disguised as a local one. Collective memory is not a single degraded copy of what happened; it is a set of individually accurate memories of different parts of it, which conflict because the event itself was not uniform. Treating the disagreement as noise to be averaged away discards the most informative thing in the record. The old village was surveyed in 2011. The survey found the church, the mill, and the foundations of thirty-one houses, which is roughly the number the folklorist's informants had described. It also found that the mill had been repaired, extensively, at some point after the village was abandoned. Nobody has an explanation for this. The folklorist died in 1989. Her notes contain no reference to anyone returning, and the families she interviewed are, by now, unavailable to ask.