Finding

Margaret belongs on the mother axis as a substantial soft hit, because the corruption runs both ways

Forename variants are a weighted scale in exactly the way surname variants are, and Margaret belongs on the mother axis as a substantial soft hit rather than as a different name. In seventeenth-century hands the letters g and y carry near-identical looping descenders, and Margaret was routinely contracted by clerks to Marg. or Mgt. to save paper and ink - so the single letter distinguishing the written token from Mary is precisely the letter most likely to be ambiguous. A Marg. written with a looping g reads as Mary to a later clerk, a nineteenth-century transcriber or a modern indexer, and a Mary is equally extendable to Margaret: the corruption is introduced by the transcription chain and runs in both directions. The corpus supplies a worked example - Tittleshall with Godwick, Norfolk, 1620 returns the same child twice, father Henry in both rows, with the mother indexed Anne in one transcript and Mary in the other. The adopted weighting gives full weight to Mary, Marie, Marye, Maria, Marya and Mria; substantial soft-hit weight to Marg, Marg., Mgt, Margaret, Margarett and Margret; partial weight to Margery and Margerie; and excludes Molly and Polly as eighteenth-century hypocorisms that cannot appear in a 1620s register. The same logic governs the father axis, where Jone and Joan in a father field are treated as a contracted or Latin Johannes rather than as a woman - the county sets demonstrably do both, indexing the father of Joseph Garnam at Norwich St Giles in 1618 as Jone and the father of Gregorye Gurnall at King's Lynn in 1615 as Katheryne.

Confidence: High

Topic file location: Identity — the shape, the matrix, and the two families