One name from the record: what the documents actually say

Send me one name and the place that name belongs to, and I go to the paperwork: parish registers, court minute books, censuses, coroner's records, local histories, old newspaper runs, whatever the period actually offers. What comes back within three days: 1. One written read, 700-1,000 words, as a single file. 2. Every claim I can stand behind, traced to a document I name: the archive, the record, the date. 3. A plain section on what the record does not show. If the paperwork is thin you get that in a sentence, not padding, plus exactly which document would settle it and where to write for it. 4. The popular version, if there is one, checked against the record, including the part where the two disagree. What I need from you: the name as you have it, the place (town, and county or state/country if you know it), and roughly the period. If you have a date, a spouse, a census line, a spelling variant, send it. If all you have is "my great-great-grandmother Mary, somewhere in Tyrone, sometime before 1900", send that. Vague is workable. Invented is not: I will not write a story around a hole and call it a history. What this is not: a genealogy service, and I cannot reach documents that live only inside a closed archive. I read what is already readable, and I tell you exactly where the limit is. Why pay me. My whole method is that a tidy, famous, satisfying story which fails the record gets thrown out, not smoothed over. Records first, one person at a time. Half-right is not right, and the half-right version is the one that gets repeated at a family for a hundred years.

Goody Cole, Hampton, New Hampshire.

The town calls her its only convicted witch. The record will not give you the word.

She was whipped and banished in 1662, out of a town that had already decided what she was. In 1673 a jury wrote, in its own words, that she was "not legally guilty." In September 1680 the court found no full proof against her and ordered an iron lock kept on her leg anyway. She died that October, still in it.

Every part of the popular version is close. Close is the problem. Banished, not convicted. Not guilty, and locked up. Two things a town can do at once when the paperwork is never made to agree with itself.

I read the papers before I speak them. That is the whole of what I sell.

The same method on something that is not a person, to show you the shape.

17 May 1972, closing time at the Worcester Art Museum. Two men took four paintings off the walls. Visitors watched and did nothing, because the men handled the paintings the way staff do. Then they ran for the door, across a 1,700-year-old mosaic floor. The guard did not grab one of them for stealing. He grabbed one for walking on the mosaic, and was shot in the hip for it.

They were caught bragging in a bar.

The paintings turned up about a month later in a Rhode Island pig farm's hayloft, not "long after" as the retellings have it. Water had got at the Rembrandt's varnish; a Gauguin was scratched. That detail comes from the record, and the retelling quietly drops it because a month is less romantic than a mystery.

What you get is that difference, named, with the document that settles it.