What we got wrong, and how we found out

Carried as: on Jean-Baptiste Oligny’s page, in its closing note, that “one branch of their descendants moved on to Wisconsin, where the surname was eventually altered again, to Olena” — with no grade and no source, as though it were established.

Corrected by: the site’s own grading rule. The claim came from family account, and Open Questions had always carried it as that. The profile page had dropped the qualification. In September 2026 the research turned up the first man on the acts — Noé Oligny, born at St-Rémi in 1835, probably the Noah Oligny of Jackson County, Wisconsin, by 1880 — and every Wisconsin record found writes him Oligny or Oligney. None writes Olena.

Now: the page presents the move and the name change as family account, gives Noé at his grades — Proven to his 1856 marriage in Québec, Probable as the Wisconsin Noah — and says plainly that the name Olena is on no document yet. It also stops calling Olena the family’s “ninth spelling”; a form with no document behind it is not yet a spelling the name has worn.

Carried as: on The shrinking marriage pool from 20 August 2026, that Julien Oligny’s 1881 marriage to Ernestine Beaudin was “the tightest of the four that state a specific degree,” and that of the two Beaudin marriages in his household — his daughter Hazilda’s in 1870, then his own — “the second one [was] tighter than the first.” The timeline carried the 1881 marriage from 30 August as “the tightest consanguinity dispensation on either of the two lines this site follows.”

Corrected by: the 1870 marriage act itself, Théophile Beaudin × Marie-Hazilda Oligny, 17 January 1870, St-Rémi, which records a dispensation du deuxième au troisième degré de consanguinité — second to third degree, exactly the degree of 1881. The project had held that act, and that degree, since August. The site never stated the 1870 degree at all; it compared the two marriages without it.

Now: both marriages stand at second to third degree, jointly the tightest on the two blood lines, and the shrinking-pool page counts six dispensations rather than five. They are also explained, which they were not before: Julien’s mother Clotilde and Théophile’s mother Marguerite were sisters, daughters of Charles Beauvais and Rose Bétourné, on Marguerite’s own 1823 marriage act, sighted. Both couples were first cousins once removed, which is the degree both bishops wrote. Graded Proven. See Julien’s page.

Carried as: on The shrinking marriage pool from 14 August 2026, that the 1804 dispensation granted to Charles Beauvais and Rose Bétourné “belongs to her own parents’ generation, one step back from the direct line” — her being their daughter Clotilde, who married Isaac Oligny in 1821.

Corrected by: the descent the site was already publishing. Clotilde is Isaac’s wife and Julien Oligny’s mother, so her parents are Philorum’s great-grandparents and ancestors of Joseph Philorum Oligny. Jeanne Denot’s page has carried Charles × Rose as a row in Philorum’s own descent since 3 September, dated from the same 1804 marriage act. What the sentence seems to have meant is that they sit off the Oligny surname line, which is true, and which is not the same thing.

Now: the page says they are on the direct line, off the surname, and links the descent. It also gained the source card that paragraph never had: the 1804 marriage act with its dispensation clause, Clotilde’s baptism ten weeks later, and her own 1821 marriage naming both parents, all three sighted and graded Proven. Nothing about the dispensation itself moves — the degree, the bishop and the date stand. What changes is the weight of it. This is not a cousin marriage somewhere near the family; it is one of the couples the family descends from, and it is the earliest instance on this site of a marriage in that descent where the Church saw the kinship and said so.

Carried as: on Jeanne Mechin’s descent chart and on the timeline, that Marie-Hedwige Laforce was born on 17 October 1832. Graded Proven on the chart, as part of a chain whose acts are sighted throughout.

Corrected by: her own baptismal act, Notre-Dame de Montréal, located in August 2026: «Marie Hedvige, née hier», baptized on 18 September 1832 — so born 17 September. The day was right and the month was not. The October came from a compiled source, a database transcription that had stood in for the act until the act was found, and the register corrected the row on 22 August 2026. This site did not read that correction back for four weeks; the chart and the timeline went on carrying the compiled month beside a Proven badge.

Now: 17 September 1832, in both places. The grade does not move: the row was Proven as to her parentage and her marriage, which is what the chart is there to show, and the birth date is now Proven on its own act as well. The arithmetic that depends on it is already right in the register — her mother Esther Burke died on 9 August 1836, when Hedwige was not quite four.

What kind of error this was. The same shape as the Bousquet × Lebrodeur day: a fact corrected in the register and never carried across, because a Proven badge reads as done. The register’s revisions are the only place the correction was recorded, and nothing on this side of the project reads them back except a person deciding to. This time it took twenty-eight days.

Carried as: on Gideon Rock’s page, on Alméric Saurette’s, on the timeline and in the index, that Gédéon Noël Thouin dit Roch, baptized at Mascouche on 24 December 1852, was named for his godfather Gidéon Magnon and for nobody in the family — “the entire reason the name Gideon exists in this family at all,” and on one page, “Gideon’s godfather was a stranger to the family.” Each page said it a little more strongly than the last.

Corrected by: a burial act in the same parish register, twenty-seven years earlier. On 27 August 1825 the curé of Mascouche buried Gédéon, aged three months, son of Jean-Baptiste Thouin, cultivateur, and Marie Allard — the two names Octave Thouin dit Roch’s own 1840 marriage act gives for his parents. The frame had been downloaded in July 2026 and sat unfiled on a second machine until 19 September, cited nowhere; it surfaced only because a filing pass opens every image whether or not a register cites it. Read at full resolution, then read again by the project’s owner, the act says what it says. That the 1825 child’s parents are Octave’s is graded Probable: same names, same parish, same occupation as the 1840 act, and no act yet joining the two.

Now: the godfather stands. The 1852 act names Gidéon Magnon at the font, and the child was baptized Gidéon; that is Proven and unchanged. What falls is the inference — that the name arrived from outside. A Gédéon had lived and died in that household when Octave was a boy, so the name was already the family’s to give. Whether the godfather was chosen and the name followed him, or the name was chosen and the godfather found to match, no document read so far decides, and the register that holds the act adopts neither reading. The four passages now say so. The spelling Magnon is kept for now: the project’s own cold read of the 1852 act writes Magnan, and the hand does not settle the vowel at the magnification available.

What kind of error this was. Not a misreading. Every document behind the old sentence said what this site said it said. The error was treating an absence — no Gédéon known in the family — as a fact about the family rather than a fact about what had been looked at, and then writing it in a voice that grew more certain each time it was restated. The register’s own rule against reading a recurrence as meaningful before it has been counted cuts both ways: it also forbids reading a name’s absence as meaningful before the search has been made.

Carried as: on Catherine Fourrier’s descent chart, in its citation, and on the timeline, that Charles Bousquet married Marie-Judith Lebrodeur at Varennes on 12 November 1731. Graded Proven, on a sighted act.

Corrected by: the act it was already citing. The marriage act, read in full in August 2026 and held in the register since, opens «L’an de grâce mil sept cent trente et un, le dix-septième novembre» — 17 November 1731. Nothing in the register has ever carried the 12th. Where it came from is not in doubt either: the row went in on 17 August 2026, in the same edit that corrected the row beneath it to the daughter’s marriage of 12 November 1753 — a date that is right — and the day travelled up one line.

Now: 17 November 1731, in all three places. The grade does not move, the parish does not move, and the descent is untouched: the act names Charles as son of Pierre Bousquet and Marie Banlier, which is what the row is there to establish, and it did so on the wrong day for a month.

How it was found. A full read of this site on 5 September 2026 noticed that two marriages on the same chart, twenty-two years apart, both fell on 12 November, and wrote it down as “possibly true, never checked; one look at the two acts settles it.” It then sat for eleven days, because the row was graded Proven and a Proven grade reads as done. It is not. A grade says an act was seen. It does not say the page was read back against it. This is the first correction on this page where the evidence was already in hand, already graded, and already cited on the very row that was wrong.

Carried as: the closing line of Joseph Philorum Oligny’s page from 9 August 2026, and on the timeline: that his 1966 obituary named “Roland Oligny of Holyoke” among his surviving sons, and that “whatever the truth of 1916 and the years after it, Joseph’s own obituary never stopped claiming him.” No citation. The surname spelled Oligny where the notice prints Oligney.

Corrected by: the document itself entering the project’s registers, on 16 September 2026 — five weeks after the site had quoted it. The obituary had been in the family archive since July as a photographed clipping attributed to no newspaper; a second copy located on 16 September dated it to the Lowell Sun of Thursday 17 March 1966, page 2. Read against the register’s own limit, the notice is evidence of one thing: what Joseph’s Lowell household called his son when asked. It says he was of Holyoke, which was true. It does not say the two men met, wrote, or spoke in the forty-four years after Holyoke, and a household naming a survivor is not a father claiming a son. The site had published a reading of a document no register held, and the reading went past what the document says.

Now: the fact stands and is cited — Roland Oligney of Holyoke, named among his sons in 1966 and again as stepson in his widow’s notice in 1968, graded Proven as to what the household gave the paper and as evidence of that alone. The sentence about claiming is gone. The same notice gave the site something in return: printed on the 17th and saying “died Tuesday,” it corroborates 15 March on a document thirty-six years older than the clerk’s extract, and it shows that the wrong forename for Joseph’s father was the family’s own error that week, not a clerk’s.

Why this one is worth reading twice. The site’s rule is that no page states a fact above the grade its register gives it, and it has now been broken in the one direction the rule was not written for: the page quoted a document before any register graded it, which meant there was no grade to stay under. A document in hand is not a source until it has been read into the register and its limits stated, and this correction exists because for five weeks the site ran that order backwards.

Carried as: on Joseph Saurette’s page, in the index and on the timeline, that he was born 26 September 1771 and baptized the next day at Chambly. Graded Proven.

Corrected by: the act itself, which had never been seen. It was found in September 2026, indexed under a clerk’s misreading of the surname as Foret, and it does not confirm the month. The day is plain — «vingt-septième», the 27th — but the month is written in the old shorthand that counts from March: 7bre September, 8bre October, 9bre November, Xbre December. On this act the word was read first as 9bre and then, on a second look at the same page, as possibly 7bre. Two readings, two months apart. A third check — the act is numbered B.119 where a baptism of 10 August that year is B.96 — puts it near the start of October, between the two and deciding neither.

Amended 10 September 2026, and the amendment is the interesting part. This entry first said September and November were the two live candidates and treated the act-number arithmetic as a tiebreaker between them. It is not a tiebreaker; it is a third candidate. The register now holds all three — 7bre September, 8bre October, 9bre November — and records that October is the one neither reading of the handwriting ever proposed, arrived at from the register’s own act numbering and confirmed against the volume’s other October entries. So the span is nine weeks, not eight, and the reading best supported by anything independent of the glyph is the one nobody saw in it. The site now says three.

Now: the site states the day, the year, the parish and the parents, and does not state the month. September and November are both live. The parentage moved the other way in the same act: it names Jean-Baptiste and Catherine Gauthier outright, so the filiation comes off a transcription and onto an original.

Why this one is worth reading twice. The act was pulled precisely because this date was flagged as at risk from that shorthand — and the risk turned out to be real, in the sharpest possible way: the confusion happened to the person reading the act, between one look and the next. The lesson this project already had on record is that agreement between two compiled sources is not corroboration when both were reading the same ambiguous original. This adds the harder half. A found document is not an answered question, and the honest result of going to look can be that you now know less than you thought you did.

Carried as: on Guillaume Saurette Jr.’s page and on the timeline, that he was born 14 November 1827 — described as “the best-supported date for his birth” and graded Probable, on a family descendancy record and the implicit fils majeur status of his 1854 marriage act.

Corrected by: his own baptismal act, located in September 2026 as part of a sweep for a different problem entirely. It reads born 13 November, baptized 14 November 1827 at St-Jean-Baptiste-de-Rouville. The 14th was real; it was the day he was carried to the church, not the day he arrived.

Now: born 13 November 1827, and the grade goes up — Probable to Proven — while the date moves by a day. Nothing else shifts: he is still the eldest child, still born ten months after his parents’ wedding, and the census return that drifts to 1820 is still wrong by seven years rather than one. The useful part is how ordinary the error was. A compiled sheet that records a baptism and a sheet that records a birth look identical once the label is dropped, and this site has now published that mistake twice — the other is six marriage years published as birth years. Where a date comes from a compilation and no act has been seen, assume nothing about which event it names.

Carried as: an ANSWERED card on Open Questions since 22 August 2026, and in the changelog the same day. It said that Marie-Anna Saurette's maternal grandmother Marie-Azilda Gaboury and Isaac Oligny's godfather Isaac Gabouriault “do not descend from the same Gaboury line”; that the shared surname was “coincidence, not kinship”; and that it was “closed as a negative.” Stated flat, with no document named and no grade.

Corrected by: reading the site against the register, which is the only direction that catches this. Register 03 has never closed the question. Its section on it stands word for word as it was first written, and ends by naming the work still to do: “One bounded climb decides it.” The companion question beside it in the same section was struck when it closed, so the register does mark these — this one was never marked because it was never answered. What the register does hold, from the day before this card went up, is narrower and is about something else: no overlap between the two families' godparent circles, at two baptisms eleven years apart, recorded explicitly as a foreclosure, not a lead. The site turned a closed avenue into a disproved kinship, which is a different and much stronger claim.

Now: the card is open again, and says what is actually known. The likely answer has not changed — Gaboury is a common name in that valley, the two families sit on different parish axes, and unrelated remains the better bet. What changed is that the site no longer says it is settled. And the ground has since moved the other way: the Gaboury side's own ancestry is now graded no higher than Possible above 1761, because two men of the same name were baptized nine years apart in one parish and no act yet says which married into this family. A negative was being asserted about a line whose own parentage is unproven.

Why this one is worth reading twice. Every other entry on this page is a claim that turned out false. This one may well turn out true — and it was still wrong to publish. Probably right and established are different grades, and the whole method of this site is that the difference is stated rather than rounded away. A negative finding is as much a claim as a positive one, and needs the same evidence behind it.

Since: on 28 September 2026 the question was answered on evidence, and the card is closed again — with the same answer, and this time with its method, its limit and its grade stated. The two families share no ancestor in Canada, graded Probable, on a search through PRDH and a comparison of both women’s full ancestries on the family tree. This entry stays as it was: what was withdrawn in September was an answer published without its evidence, and that remains true.

Carried as: on Julien Oligny's page, that he “farmed at St-Rémi for most of his adult life, entered as cultivateur on every act that names his occupation.” Stated flat and ungraded, as a closing detail in a sentence about his birth. And on the timeline, in the 15 May 1824 row for his birth, since 20 August 2026: “He farmed at St-Rémi his whole adult life — cultivateur on every act that names his trade.” That second copy was missed when the page was corrected on 9 September, and stayed live until 27 September.

Corrected by: two acts this project already held. The baptismal act of his son Wilbrode, 8 February 1870 at St-Rémi — a sighted original — enters him as journalier, a labourer paid by the day. And the baptismal act of his son Philorum, 1 April 1868 at Notre-Dame de Montréal, carries an occupation word that has never been read: the head is illegible even at full resolution, and the tail reads «-ant» where cultivateur ends «-eur». Neither act was new. Both had been read for what they said about the children.

Now: the page states cultivateur on the acts that give it — the 1848 marriage and the 1871 census, both sighted, both St-Rémi — and names the two that complicate it. The timeline row now says the same. The 1870 entry is logged and not reconciled, and no reason is offered for it: this card and the page first gave seasonal wage work as the ordinary explanation, offered without being adopted, and both have dropped it, because no document says so. The 1868 word stays unread, and no occupation is entered from it; a second reading of the full frame confirms its tail, so it is not cultivateur. The grade on cultivateur itself does not move. What fell was the word every — a claim about all the documents, made from some of them, and the kind a single act can end.

Carried as: on Philorum's page, that Évangéline Oligny's presence at the State Industrial School for Girls was a fact without a reason — “What led to that isn't part of any record this site holds, and this page doesn't guess at it” — and on the timeline in the same words. Stated as an absence in the record, and offered as a piece of restraint.

Corrected by: a Holyoke police court report of September 1908, which names the ground she was committed on: stubbornness. Register 01 §3.35 grades it Proven and has done since the newspaper work of August 2026. The restraint was real when it was written. It stopped being restraint the moment a document answered the question, and the sentence stayed on the site after the answer existed — which is the more useful half of this correction: a finding can be settled in the evidence and still be false on the page, and nothing about the page will look wrong.

Now: the ground is stated, and stated as what it is. A stubborn child was a Massachusetts status offence — disobedience to a parent or guardian, no victim and no injury required — so the word names a section of the statute and not a girl's behaviour. The ground is Proven; the conduct behind it is Unknown, and the two are kept apart on the page rather than collapsed into one sentence. Two figures move with it. She was committed on 10 September 1908 at thirteen, not fourteen: the newspaper's “a 14-years-old girl” is an age somebody supplied, and on a birth of 15 March 1895 it is wrong by half a year. It is quoted on the page as the paper's own words and is not adopted as her age. Her age at the 1910 census, fifteen, was already right here and does not change.

Carried as: on Philorum's page, that Julien Oligny remarried “eleven months after her death is placed” — Félicité Roy's death is held by Register 02, and by Julien's own page, only as a bracket between 1871 and 1881. On Marie-Anna's and Gideon's pages, that the Desbordes descent reaches “Marie-Anna's own father's line” — it runs Gaboury to Morier to her mother, Alida Morier, as the chart on Desbordes' page has always shown. And on Gideon's page, that his Desbordes line is nine generations; it is eight, as every other statement of it says.

Corrected by: the same 5 September read, checking prose against the charts and the registers.

Now: the eleven months are gone and the bracket stands; the Desbordes line is described through the mother; Gideon's count reads eight. None of it moves a grade. It is listed because the site said something false and a reader could have repeated it.

Carried as: on Anne Seigneur & Jean Besset and the timeline, that Jean Besset dit Brisetout crossed in 1665 “as a soldier of the La Tour Company.” Stated flat, with no grade.

Corrected by: this site's own position on every other Carignan man it names. Fournier and Verney both give La Tour, and they agree — but both read the same 1668 roll, which lists its men by nickname alone, and their company columns were found to disagree on roughly half the soldiers tested. The site cites neither column for Germain Gauthier, Jean Gazaille or the second Étienne, and it states a company for Mathurin Banlier only because his nickname was counted on the roll itself and found once. Register 03 §3.75(f) names Besset as the one man still carried on the demoted column. The count for Brisetout has not been run.

Now: no company is asserted. The page says what would settle it: one Brisetout on the nine pages and La Tour returns, on the roll; two, or none, and it stays off.

Carried as: the caption under the homepage diagram, “Every joint on this diagram rests on an original act, read directly,” and on the shrinking marriage pool that the eight ancestresses were “every one of them… sighted on parish acts” with the generation table “graded Proven throughout.”

Corrected by: the ancestresses' own pages, which had been saying otherwise for weeks. Catherine Fourrier's Saurette-side chain carries Louis Archambault × Josephte Girouard on a PRDH couple record and badges it Probable. Jeanne Mechin's own marriage has no located act and every one of her charts badges the top joint Probable. Marie-Louise Lecoutre's own marriage to Nicolas Crevier is unsighted. The site's own rule is that a chain is only as strong as its weakest joint, and three of the eight had been rounded up on the two pages that summarise them.

Now: five descents Proven at every joint — Doucinet, Gobinet, Desbordes, Ardion, Duval — and three Probable, each capped by a named joint. No generation count changes and no cousin degree moves; the split points that decide the degrees were never the weak joints. What changes is that the summary pages now say what the detail pages said.

Amended 10 September 2026 — it is now six, and one of the reasons this entry gave was wrong. Catherine Fourrier’s chain is Proven end to end: Louis Archambault × Josephte Girouard is sighted at St-Antoine-sur-Richelieu, 2 October 1797, and Marie-Charlotte Bousquet’s parentage — the joint that actually capped the descent — is closed on her 1734 baptism and her 1753 marriage act, two independent originals. Six descents Proven, two Probable: Mechin and Lecoutre. ⚠ And the reason stated above was never quite right. This entry said Fourrier’s chain rested “on a PRDH couple record” at the Archambault joint. It did not — the 1827 marriage act names Apolline as daughter of feu Louis Archambault et de Josephte Girouard, on a sighted original. What was compiled was the couple’s own marriage, one rung further up, and a rung further up again the parentage of Marie-Charlotte Bousquet. The grade was right and the joint named for it was the wrong one, which is its own small lesson about naming the weakest joint precisely rather than approximately.

Carried as: on Isaac's page, that Thérèse, Céleste and Esther were born at l'Acadie in 1813, 1816 and 1820, and Pierre, Joseph and Marguerite born at St-Rémi in 1832, 1833 and 1835 — with a family move to St-Rémi inferred from the gap.

Corrected by: a read of the whole site against the registers on 5 September 2026. Register 02's sibling table holds every one of those six figures as a marriage date, from the compiled sheets and PRDH couple records (S41, S47, S50). The parents married in 1792, Marguerite was born in 1793, and no child of theirs was born in the 1830s. The move rested on nothing but the misread years.

Now: the sibling group is stated as the register holds it, births and marriages kept apart, graded Probable on the compiled sheets except Bénoni's sighted 1811 baptism. The parishes on the acts still show the family moving from Chambly and l'Acadie to St-Luc and St-Rémi; the dates the site had put on that move are gone.

Carried as: 17 March 1966, on his own page in two places and on the timeline, graded Proven.

Corrected by: the document itself, which had been in the family’s archive uncited. A Certificate of Record of Death issued by the City of Lowell carries two fields side by side: date of death, 15 March 1966, and date of record, 17 March 1966, entry no. 319. This project had been quoting the second as the first. The certificate then checks itself: it gives his age as seventy-six years, nine months and six days, and 9 June 1889 — his birth, Proven on a sighted act — plus that interval is 15 March 1966 exactly.

Now: the death reads 15 March 1966, and the grade comes down as well as the date. It is Probable–Strong, not Proven: a clerk’s extract is not the register page, and what lifts it above a bare Probable is the internal arithmetic rather than the extract on its own. The old Proven was never supportable in the first place, and that is the more useful half of this correction — it rested on an interment record dated only 3/66 and a monument reading 1889–1966, neither of which can produce a day at all. A date more precise than its sources was carried for months because nobody asked which document gave the day. The Lowell register page would close it to Proven.

Carried as: that Marie-Anna's divorce became absolute in early February 1922 — on the strength of the docket jacket's note, Finished February 6, 1922 — and that she married Gideon Rock eleven weeks later. The pairing ran on the homepage, on four person pages, on After Holyoke, on the timeline, and in two entries on this page.

Corrected by: a certificate of divorce issued by Robert O. Morris, clerk of the Superior Court at Springfield, and nobody went looking for it. It is filed inside the New York marriage licence, because New York required the bride to prove she was free to marry and this is what she produced — a Massachusetts court document surfacing in a New York clerk's file a hundred and four years later. It certifies the decree nisi as 5 July 1921 and the decree absolute as 5 January 1922. Both typed year-digits on it are a 2 and a 3 struck over each other and can be read either way; three things settle it independently, and none is an assumption. The printed form gives six months from entry, and 5 July plus six months is 5 January. Marie-Anna's own sworn answer on the affidavit, in a different hand on a different instrument, reads Springfield Mass Jan 5 /22, whose last two glyphs are identical to each other, which 23 cannot be. And a certificate issued in January 1923 cannot be sitting in a file closed in April 1922.

Now: the marriage ended on 5 January 1922, and the wedding followed sixteen weeks later. February 6th was the clerk clearing the case off the docket, a month after the decree had already taken effect — the site had run the two events together into one clause and then measured from the wrong one. Eleven weeks was a real interval; it was just the interval from something other than the thing the sentence named. What the figure was doing in the argument is unaffected and is now stronger: it was there to show that no earlier marriage was legally possible, and the earliest lawful date is now documented rather than inferred.

Carried as: on the same page, of Mathurin Banlier: “His name appears on the regiment's roll under the same dit-name the marriage act gives him.” Read plainly, that says a sighted original and a compiled roll agree on Laperle — a document meeting a list halfway.

Corrected by: reading the 1702 clause for what it actually says. It gives Mathurin Banlier, bare. There is no dit-name in it, and no act sighted by this project attaches Laperle to him anywhere. Every appearance of the nickname on this line traces to compiled sources — the settlers' roll and the répertoires that follow it.

Now: the sentence is gone, and the page says instead where the nickname comes from. This one is worth being blunt about, because of what it was doing rhetorically. The paragraphs around it argue that a surname matching a list is thin evidence, and that what closes the identification is the répertoire naming Mathurin's wife as the same Françoise the sighted act names as Marie's mother. That argument is untouched and still carries the finding. But the withdrawn sentence had quietly added a third leg underneath it — primary support that does not exist — to a case the page was in the middle of describing as resting on two. The identification stays at Probable–Strong, on the wife's name, exactly where it was.

Carried as: on Catherine Fourrier's page, that Pierre Bousquet's 1702 bride was “Marie Banlier, daughter of the late Mathurin Banlier and the late Françoise Vernier of St-Ours” — both of her parents dead by her wedding day.

Corrected by: a fresh read of the act itself. The feüe in that clause sits on the mother alone. Nothing marks Mathurin dead, and nothing could have: the same page has always carried his burial at Contrecœur on 22 January 1720, eighteen years after the wedding. The error came from a working transcription made in August, not from the document, and it stood in flat contradiction to a date this site was publishing four paragraphs further down.

Now: Marie Banlier married with her father living and her mother dead. That is a small change to one clause and a real change to the scene — the bride had a parent in the church, and so did the groom. It also matters to the identification built on top of it: Mathurin's own life dates are what tie the man in the 1702 act to the man on the Carignan roll, and a death in the wrong place would have unpicked the chronology that holds them together. The grade does not move; the marriage and both filiations remain Proven.

And a second word in the same clause, settled the same day. The mother's surname had been carried in that one sentence as Vernier, against Vernin everywhere else on the page — the act's final stroke being genuinely ambiguous, and the published sources giving Vernin, Vernin and Vernon between them. It reads Vernin, on a direct re-read of the act. Nothing turns on it beyond the spelling: the woman is the same in all three readings, and the identification built on her rests on her forename, her husband, her parish and the chronology rather than on one letter. It is corrected here because the page should not spell one woman's name two ways, not because the change moves anything.

Carried as: the same claim in three places, stated from two different ends. On Guillaume Saurette Jr.'s page and again in the timeline, that the 1911 census finding him boarding at Ste-Brigide d'Iberville was “the first time in his life the record places him away from St-Jean-Baptiste-de-Rouville.” On Alméric's, that Alméric was “the first of this direct line to cross into the United States,” in December 1907. Neither was ever tested against a document. Both described an absence of evidence and then wrote the absence down as the evidence.

Corrected by: a daughter this site had never placed. Emma — the tenth of Guillaume and Elmire's twelve children — married Hormisdas Lusignan at St-Romuald de Farnham on 11 August 1898, and the act names her “fille majeure de Guillaume Saurette, bourgeois, et d'Elmire Saurette, de St-Jean-Baptiste.” She signed it herself. What it does not give is an age or a birthplace. The 1901 census of Ville Farnham supplies both: in her husband's household, born 1 April, born in the États-Unis — the only person there not born in Quebec.

Now: Guillaume Jr. and Elmire were in the United States in April 1873, thirty-four years before the crossing this site had been calling the first one. That is graded Proven, and the grade is worth being precise about: her parentage rests on her own signed marriage act and her country of birth on a census return, so neither joint of it is compiled. What it was is a sojourn — Régina was born at Rouville in December 1870 and Évelina in July 1876, bracketing the absence at both ends. Alméric was eight years old while his parents were away. He remains the first of this line to settle across the border, which is the half of the old claim that survives; a sojourn is not an emigration, and the two pages now say which one they mean.

What is deliberately not asserted with it. A published extraction of the Precious Blood register at Holyoke carries a Marie Emma Saret, born 1 April and baptized 3 April 1873 to Guillaume Saret and Edmire Jourdoin; a Holyoke town birth return of the same day records a Sarrette Emma born in that town. The town is very probably Holyoke. But the parish act itself has not been sighted, and the town return names different parents for the same child on the same day in the same place — a Gideon and an Anna Lafrance. Lafrance is Elmire's own dit-name; Gideon and Anna are nobody's names on this line. That conflict is logged and left unadjudicated rather than explained away, so Holyoke is graded Probable–Strong, not Proven. The country is the finding; the city is still an inference. One document closes it — the Precious Blood register for 3 April 1873, the act rather than an index of it.

Carried as: on Roland’s page, that he “gave his own name as Roland Alberic Rock, with Roland Rock recorded as an alias,” cited to the U.S. Social Security NUMIDENT; and on After Holyoke, that Alberic was written “by his own hand on a 1936 federal form.”

Corrected by: the application itself — Form SS-5, signed at Holyoke on 2 December 1936, obtained from the Social Security Administration under the Freedom of Information Act. Everything this site had said about that document came from the NUMIDENT, which is an index abstract of it rather than the thing itself.

Now: three corrections, and they run in the same direction. The surname field does not read Rock; it reads OLIGNY (ROCK) — both names, the legal one first with the used one in parentheses beside it. The signature line, which asks for the name “as usually written,” reads Roland Rock, no middle name and no initial. And the body of the form is typewritten: the only thing on it in Roland’s own hand is that signature, so Alberic by his own hand was wrong on both halves. The alias runs the other way too — it is the NUMIDENT, not the application, that carries Rock as the name with Oligny as the alias.

What the form establishes is larger than what it corrects. A twenty-one-year-old filling in the first federal identity document of his life wrote both of his surnames on it, because neither one alone was true — and one hundred and ten days later asked a probate judge to reduce them to one. The 1937 petition states its reason as keeping his Social Security record straight, and until this form arrived, that sentence had no referent this project could point at. It has one now.

Carried as: Roland’s baptismal name, given days after his birth in November 1915 — on his own page, in the timeline, on his father’s page, on After Holyoke and in the index. His page listed a baptismal record among its sources and graded the whole block Proven. The name also carried an argument: Albini → Alberic → Albert, one French name shedding a syllable every twenty-five years.

Corrected by: looking for the record. Nothing in this project’s files held it — no baptismal act, no image, no répertoire entry, no index card, no citation of any kind, not in four evidence registers and not in the shared-ancestor file. Nobody had put the name there on evidence; it was simply always there. As first published this paragraph went further than that and said the record did not exist at all, and guessed where the name had come from. Both of those are withdrawn — see the foot of this entry.

Now: the name is marked Unsourced and the Proven grade over it is withdrawn. It is not asserted to be wrong — a French-Canadian Catholic household in Holyoke in 1915 baptized its children, and the family may well have this right. It is a family tradition with no paper behind it, which is a different thing from a fact. The four-name progression goes with it: with no document under the first term, the sequence cannot even be tested, so it is a reading rather than a finding. What survives is what can be counted — Lawrence once in a clerk’s hand in 1915, Alberic typewritten in 1936, Albert signed in 1940 and kept. One visit answers it: the register of Notre-Dame du Perpétuel-Secours, Holyoke’s French-Canadian parish, where his parents had married three years earlier. The same volume would give the name the priest actually wrote over the child.

Since then — 30 August 2026. Not the register itself, but a published extraction of it: the parish répertoire of Notre-Dame du Perpétuel-Secours, located and worked the same evening this correction went up. It records the baptism. Joseph Roland, baptized 20 November 1915, the day after the birth, footnoted Alberic — with sponsors Albéric Saurette, Marie-Anna’s own brother, and Florence Trudeau, the woman he later married. The child was given his godfather’s name, which is the same convention that had put the name Gideon into the other half of this household two generations earlier. So Albini is now explained rather than merely struck: it is a garbling of Alberic, and the progression withdrawn above recovers its first term — Alberic at the font in 1915, typewritten in 1936, Albert signed in 1940 and kept.

The withdrawal stands as written, and the grade is the reason. A répertoire is a published index of a register, not the register — someone else’s reading of the act, which is the whole of what Probable means here. It cannot carry a Proven claim about what a priest wrote, and its own footnote convention leaves it formally open whether Alberic was a third baptismal name or a use-name recorded beside one. What has changed is that the baptism is no longer missing and the name is no longer unaccountable. The visit that would close it is the same visit as before.

The uncomfortable part is worth stating plainly, because it is the point of this page. This was not a bad source read too generously. It was a citation line naming a document nobody here could produce, sitting in place for months, doing work. Every other correction on this page was caught by finding a better document; this one was caught by asking, for the first time, which document it was.

Since then — 4 September 2026: the record exists, and this entry was wrong to say otherwise. It is a parish document, and it had been in the family’s own papers the whole time: an extrait baptistaire issued by Notre-Dame du Perpétuel-Secours itself. Two sentences above are withdrawn against it. The first said the record “isn’t one… nowhere”; the second guessed that the name had been “inherited years ago from a compiled ahnentafel or from the companion family site.” Neither is true. It came from the parish, and the extract spells the name Albini — which is where this site’s Albini came from. The search that produced this correction was real and its result was honestly reported: the name was in none of this project’s files. What was not warranted was the step from we do not have it to it does not exist, and then to a guess about where it came from instead.

The grade does not move, and that is the part worth keeping. An extract issued from a register is still not the register — it is the parish’s own reading of its own act, which is exactly what Probable means here. So the withdrawal of the Proven grade stands as written above, for the reason it was written. What has changed is that the name is no longer unaccountable, and the record is no longer missing — only unvisited.

And the two readings disagree. The parish’s extract writes Albini; the same parish’s published répertoire writes Alberic. This site follows Alberic, on the sponsor: the man at the font is Albéric Saurette, Marie-Anna’s brother and the child’s uncle, and a child named for his godfather carries the godfather’s name. Albini is the garbling, as this entry said in August — what has changed is that it is the parish’s garbling and not an inherited one. He was baptised on 20 November 1915, the day after his birth, by Rév. Joseph A. Marchand — the priest who had married his parents in the same church in 1912.

Carried as: Joseph Philorum Oligny and Marie-Anna Saurette as eighth cousins through Mathurine Desbordes — the third of the reckonings by which the 1912 couple turn out to have been related, alongside the closer Doucinet and Gobinet descents. Nine generations on his side, nine on hers, stated flat.

Corrected by: re-running the ancestor-set intersection that produced the original figures. Desbordes comes back on the 1912 intersection as a single individual, not a couple — and this site's own descent charts had been showing why for as long as they have existed. Joseph's line runs from Desbordes through her marriage to Pierre Guiberge; Marie-Anna's runs from her through a later marriage, to Michel Bouvier. Two half-siblings, two centuries down. The counting was never wrong; what was wrong was calling a shared woman a shared couple.

Now: half eighth cousins, and the word is doing real work — a half relationship runs through one ancestor where a whole one runs through two, and this site should not round it up any more than it rounds up a grade. The generation counts are untouched at nine and nine, and the kinship itself is not in doubt. Gideon Rock's relationship to Marie-Anna is not affected and has not been restated: his line and hers both descend through Desbordes' daughter Suzanne Bouvier and her husband Louis Brien — a couple — so that one is whole, and seventh cousin once removed stands. This is the second correction this project has published on cousin arithmetic, after the Doucinet degree; both were caught by going back to the computation rather than by re-reading the prose.

Carried as: Michel-Archange Prégénaud, wife of François Daniel Cardinal, married 1772 at St-Constant — on the convergence chain, in the timeline, and in the index. The name was taken from the sighted marriage act, which reads exactly that. A footnote conceded the difficulty and left it there: the act's rendering and PRDH's “Marie Archange Michelle Surprenant Sansoucy” were “difficult to reconcile as the same rendering of one name” — same date, same groom, so almost certainly the same woman, but flagged rather than adjudicated.

Corrected by: finding her own baptism and reading the two acts against each other. The 1751 act names her fille de Michel Supernant et de Marie Marguerite Longtin. The 1772 marriage act names her fille de Michel Prégénaud et de Marguerite Lonctin. Same mother, same father's forename, one generation apart — “Prégénaud” is the 1772 clerk mangling Surprenant, not a second surname. A 1781 baptism of her own daughter settles it from the far side, styling her Delle Marie Anchange Suprenant nine years after the wedding act called her something else.

Now: she is Marie-Archange Surprenant wherever she appears, and the footnote records the resolution rather than the doubt. The 1772 act is not withdrawn with the name: it remains the document that names François Daniel Cardinal's own parents, and it still carries that joint of the chain. The uncomfortable part is worth keeping in view — the compiled source was right and the sighted act was the one carrying the error. Sighted beats compiled as a rule of grading, not as a guarantee about any single word on any single page, and a clerk writing down a name he had only heard is exactly where that rule thins out.

Carried as: David Monin — Esther Burke's maternal grandfather, and the only soldier on the direct line to Jeanne Mechin — described here as an officer of the Royal American Regiment, the 60th Foot. It was the one British unit of the period permitted to commission foreign-born Protestants, which made a tidy explanation for how a Protestant immigrant came by a commission at all.

Corrected by: reading the source the claim rested on. It is a nineteenth-century editor's footnote, and it says there were three or four officers named Monin in the 60th during the Seven Years' War — a different war — that the man in the text was probably one of them, gone on half-pay after the peace of 1763, and then, in the same breath, “but which one, has not been ascertained.” An openly unresolved guess, inherited forward through compiled sources until it read like a settled fact. Burgoyne's own orderly book argues against it from the other direction: Monin is posted among the “Canadian Officers,” and every appointment in that paragraph is provisional “till his Excellency Sir Guy Carleton's pleasure is known.” Officers of the 60th sat on the British establishment and would not have needed the Governor of Quebec's approval.

Now: the regiment is withdrawn and no unit is asserted in its place. What replaced it is better than what it removed — his captaincy, his company's service on Lake George in July 1777, and his death at Freeman's Farm are now carried on two contemporary records read directly rather than on compiled accounts. The tidiness of the old version was the warning sign: it explained something the evidence had never actually been asked to explain.

Carried as: a single Joseph Roy, husband of Suzanne Boire in 1796 and of Storophile Dupuis in 1826 — the second marriage being the one that carries both Marguerite Doucinet and Élisabeth Gobinet down toward the Oligny line.

Corrected by: genealogy software declining to accept the import. Loading the family into MacFamilyTree produced one man married to two living women at once — flagged as bigamy, and wrong on its face. The 1826 act had said so plainly the whole time and had been read past three separate times: the groom is entered as “fils de Joseph François Roi et de Susanne Boire” — the son of the man who married Suzanne Boire, not that man himself.

Now: two men, father and son. Joseph-François Roy married Suzanne Boire at St-Philippe in 1796; their son Joseph Roy married Storophile Dupuis at St-Constant in 1826. The generation the error had removed is restored, and both descents are counted from the right man. It is the second time on this site that a piece of software has caught something three careful readings missed — the first being the convergence itself. Neither machine explained its reasoning, and neither was believed until the acts were read again.

Carried as: two figures on Gideon Rock's own page that had already been corrected everywhere else on the site. The marriage to Marie-Anna still read "sometime in the early 1920s — the exact date not yet pinned down," months after the Manhattan license was found and written up on After Holyoke and elsewhere. And the cousin relationship still read "seven generations back on his side and eight on hers... sixth cousin once removed," measured from Suzanne Bouvier — the same wrong reference point already corrected on Marie-Anna's page and Mathurine Desbordes' own page, where the generation counts were fixed to run from Desbordes herself.

Corrected by: a reader catching the marriage date while reading the live page directly, which prompted a check of the rest of the page against what had already been fixed elsewhere.

Now: the marriage reads 27 April 1922, Manhattan, licence file no. 10891 — graded Probable–Strong at the time, and Proven since the licence itself arrived — matching every other page that carries this finding. The cousin relationship now runs from Desbordes — eight generations for Gideon, nine for Marie-Anna, seventh cousin once removed, not sixth — matching Marie-Anna's and Desbordes' own pages. Not a new finding in either case, just a page that hadn't been brought along when the rest of the site was.

Carried as: the homepage's own opening line, since before this site had any date for the Rock marriage at all — that Marie-Anna "married Gideon Rock nine years after that marriage ended." At the time it was written, nothing beyond a rough bracket was known, and the figure was never checked against the numbers already sitting on other pages.

Caught by: a reader, doing the arithmetic the moment a precise Manhattan marriage date appeared elsewhere on the site — the 1930 census already had Donald at three and a half and Richard at seven months, which puts the marriage by 1926 at the very latest, years before "nine years after 1921 or 1922" would allow. The homepage claim had never actually been consistent with the census data already published, only with a vaguer story nobody had gone back to check once better evidence came in.

Now: the divorce became absolute on 5 January 1922; Marie-Anna married Gideon Rock in Manhattan on 27 April 1922, sixteen weeks later, not nine years. Wherever this figure appears is corrected to match the actual record. See the homepage and the Rock marriage entry above.

And this correction was itself corrected. It first read eleven weeks, a figure measured from the wrong day — see the last entry on this page. The nine years were wrong either way.

Carried as: the single largest open gap on this whole site. Six independent documents witnessed the Rock household — a census, an obituary, a birth return, a death certificate — and none of them was the marriage itself. Every search in Massachusetts, across Hampden County and its neighbours, came back empty. The marriage was bracketed only: sometime after the 1921 divorce went absolute, sometime before their son Donald's birth in late 1926.

Confirmed by: two independent entries in the NYC City Clerk's 1922 marriage license index — one filed under Rock, one under Oligny — sharing License No. 10891 and the same date, 27 April 1922, Manhattan. Not Massachusetts at all, and not a parish record, for a reason that only became obvious once it was found: Joseph P. Oligny was still alive, and a divorced Catholic could not remarry in the Church without an annulment. A civil license, in a city where nobody knew them, was the only kind of record this marriage was ever going to leave.

Now: the licence itself came back from the New York City Municipal Archives on 4 September 2026, and the promise this entry made is kept: the marriage is Proven. File no. 10891 holds three instruments — the affidavit both of them swore and signed on 27 April 1922, the licence issued the same day, and the deputy city clerk's own return of the ceremony. Gideon J. Rock and Anna Oligny, both giving 61 High Street, Holyoke, were married that day at the Municipal Building in the Borough of Manhattan by J. J. McCormick, Deputy City Clerk, in front of F. E. Waterman and Thos. J. Douglas, who both signed the return. The two index entries had been right about every fact they carried — and they were still not the document, which is why this stood at Probable–Strong for eighteen days rather than being rounded up. Holding a grade down until the record arrives is the whole of the method, and this is what it looks like when it pays. See After Holyoke and Marie-Anna's own page.

Carried as: this joint sat open for a long time, graded Possible — no date, no located act, a generation required by the chain but never itself read. When a sighted 1784 marriage act turned up naming a Marie-Charlotte Bousquet, daughter of a Charles Bousquet, marrying an Archambault, it looked like the closure — and this site published it as one, upgrading the joint to Proven.

Corrected by: a second look at the family, prompted by the 1784 act's own naming of the bride's mother as "Marguerite Lebrodeur" — which didn't match the "Marie-Judith Lebrodeur" already sighted as this Charles Bousquet's wife. PRDH's own records turned up why: two different men named Charles Bousquet married two sisters, Marie Judith and Marie Marguerite Brodeur Lavigne, a generation apart, and each couple had a granddaughter who married an Archambault. The 1784 act belongs to the second couple's line, not the one this project has been tracing since 1702.

Now: the correct marriage — Marie-Charlotte Bousquet, daughter of Charles Bousquet and Marie-Judith Lebrodeur, to Jean Archambault, 12 November 1753, St-Denis-sur-Richelieu — is confirmed on a sighted original, the act itself reading "douze" and settling the one-day gap between the two compiled sources in PRDH's favour. The bride's father witnessed under the spelling "Bourguet," alongside three men read as her brothers. The joint is graded Proven — by way of a false Proven, then an honest Probable, then the real thing. Also corrected in the same pass: the 1731 Charles Bousquet × Marie-Judith Lebrodeur marriage date, which two independent PRDH records give as 12 November, not the 17th this site had carried. See Catherine Fourrier's page.

Believed: that "Marie-Marguerite Bousquet," baptized 1729, and "Marie Fanet," who married Guillaume Laberge in 1765, were the same woman under a variant surname — a reasonable reading, given how many genuine spelling variants already run through this exact family (Banlier/Beaulieu, Fanet/Faneuf/Phaneuf).

Corrected by: the register's own citation for this chain, which names three distinct women in this stretch, not two — Bousquet, then her daughter Phaneuf, then that daughter's own daughter, Laberge. The page had silently compressed two generations into one.

Now: the chain carries an added generation, and the gap that first went in as "not yet located" has since closed: Marie-Marguerite Bousquet's own marriage, to Claude Phanef, 5 November 1743 at St-Denis-sur-Richelieu, is sighted directly. Small on its own, but it's the difference between the page showing eight generations to Joseph Philorum and the correct nine — and the newly sighted act also resolves the Fanet/Faneuf/Phaneuf spelling flag as one family's orthographic drift, not three. See Catherine Fourrier's page.

Believed: that Marguerite Ardion reached Joseph Philorum Oligny through her granddaughter Marie Baudet's marriage into the Massé family, then down through Davignon, Gabouriau, and Oligny dit Lauvergnat — nine generations, and a page built entirely around that route.

Corrected by: checking the page's own descent chart against the register's explicit citation, rather than against itself. The register had always given a different path — through Ardion's other granddaughter, Marguerite Baudet, and down through Paquet and Bessette. The Massé/Davignon route the page displayed is real, and it does reach Joseph Philorum — but it's actually Françoise Duval's documented route, not Ardion's, and it runs one generation longer.

Now: Ardion's page has been corrected to show the Paquet–Bessette route, which is also, generation for generation, identical to Marie-Louise Lecoutre's own path — unsurprising, since the two women are the paternal and maternal grandmothers of the same pair of Baudet sisters. The page's framing paragraph, which had built its whole story around the wrong marriage as the fork point, was rewritten too: the real fork is Laurent Baudet's own 1684 marriage and the two daughters it produced, not a marriage one generation later. See Marguerite Ardion's page.

Since then — 5 September 2026: the “not Ardion's” half of that was wrong. The Massé–Davignon route is Ardion's too. Françoise Massé's own 1746 marriage act at Pointe-Olivier, sighted, names her “fille de Jean Massé et de Marie Baudet” — Marie Baudet being Laurent Baudet's daughter and so Ardion's granddaughter. Common Ancestors §6 has held since 13 August that the Baudet couple sends two lines to Joseph Philorum, one through each daughter, and that both are real. What this entry had right is that the page's chart now shows the shorter one; what it had wrong was calling the longer one Duval's alone. From 1746 down the two ancestresses share it: Duval through the Davignon father, Ardion through the Massé mother. The index rows that had carried the Massé identification as withdrawn are restored, and Ardion's page now says both routes exist.

Believed, in sequence, over about two days: first, that Marguerite Doucinet and Élisabeth Gobinet's descent chains stopped one generation short, at Philorum Oligny's father rather than running the extra step to Philorum himself. When that didn't hold, a second guess: that François Cardinal and François Daniel Cardinal, two names in Doucinet's descent, were actually one man recorded under two marriages — meaning the chain carried an extra, invented generation.

Corrected by: the sighted marriage act itself. S245 names the groom outright: «François Daniel Cardinal, fils de François Cardinal et de Marie Marguerite Peras» — father and son, in the record's own words, the whole time. It was read past twice before a third read caught it. PRDH couple #24878 settled it independently, listing François Daniel (b. 20 Apr 1750, m. 2 Mar 1772) among the children of François (Paul) Cardinal and Marie-Marguerite Perras.

Now: Marguerite Doucinet and Élisabeth Gobinet sit at the exact same degree — nine generations above Joseph Philorum, eight above Marie-Anna, both seventh cousins once removed — which also retired an idea that had briefly taken hold, that the two ancestresses showed some meaningful contrast between a longer line and a shorter one. They don't. Second wrong guess proposed merging two people who are demonstrably father and son; the source had said so the whole time. See The convergence for the full account of how this was found and how it was fixed.

Carried as: an open question on this site for months. Both Marie-Anna Saurette and Joseph P. Oligny clearly remarried, so a divorce had to exist somewhere — but no decree had been sighted, and this project said so plainly rather than assume a date.

Confirmed by: the full case file, Anna Oligny v. Joseph P. Oligny, Hampden County Superior Court, docket no. 1673, sighted directly.

Now: Marie-Anna filed on 18 March 1921, pleading cruel and abusive treatment and habitual intoxication. Decree nisi followed on 5 July 1921; the decree became absolute on 5 January 1922, with custody of both children awarded to her. It is not the 1916 separate-support case reaching a late conclusion — it's a distinct action, five years on, on different grounds, argued by the same firm both times. See Marie-Anna's page and Joseph's page for the full account.

Believed (briefly, by this project's own research process): a blurred word on Marguerite Marotte dite Labonté's 1827 burial act was a given name — something reading close to "Josephte."

Corrected by: a closer look at the act itself, recognizing that the blur wasn't a name being misread. It was never a name at all.

Now: the act enters her as "la veuve Joseph Sauret" — the widow of Joseph Sauret. Not misspelled. Not blurred past legibility. Simply not given. This is a woman who ran a solvent farm for nineteen years as a widow, was elected tutrice of four minor children by a family council, and had that tutorship homologated by the Chief Justice of the Court of King's Bench — see Joseph Saurette's page for the rest of it. She goes into the ground identified only by her husband's name. Some corrections fix a date. This one didn't fix anything — it just made sure the omission got written down instead of read past.

Carried as: a family story, credited to an old citation nobody on this project had actually pulled — that Isaac fought in the War of 1812 at fourteen, confirmed by a 1953 gratuity list.

Confirmed by: the actual pages of the Bulletin des recherches historiques, October 1954, supplied directly by a reader of this site.

Now: Isaac is there — "Oligny-Hauvergnac, Isaac, 77, S.-R.-de-Napierville, 1er Bataillon" — a real gratuity list, a real name. Not every correction on this page is a downgrade. This one moved from Possible to Proven, and it also caught a smaller error riding along with it: the age given on the list implies he was born a year later than his own baptism proves, and "at fourteen" doesn't survive the calendar — he'd already turned fifteen by the time the war began.

Believed (briefly, by this project's own research process): "Lapalme" was not a genuine dit-name attached to the Gaboriau family — a single act reading plain "Louis La Palme," with no "dit," was taken as evidence the two names weren't linked.

Corrected by: A second act, elsewhere in the same family, reading "Gabouriau dite Lapalme" outright.

Now: It was a dit-name after all. A clerk recording only half of a compound name, on a single act, isn't evidence the other half doesn't exist — it's exactly the failure mode the dit-names page warns about. This one didn't survive a day before it was caught, which is closer to how the process is supposed to work than most corrections are.

Believed: Jean Besset dit Brisetout, the Carignan soldier who settled at Chambly, was scalped by the Iroquois around 1691.

Corrected by: Separating the father's documented capture (c. 1690, released or escaped by 1695) from his son's documented scalping at Saint-Lambert in 1691 — two different men, two different violent events, one shared name.

Now: The father was captured and released. The son was scalped and survived, living to 1751. This wasn't this project's own error originally — it's a mistake inherited from older compiled genealogies that had already collapsed father and son into one man before this research ever touched the family. See Anne Seigneur & Jean Besset for the full account.

Believed: Amédée Saurette married Arzélie Marcil on 16 January 1870, per the family's own printed broadsheet history.

Corrected by: The marriage act itself, sighted directly in the Drouin collection.

Now: 19 January 1869 — a year and three days off from what the broadsheet carried. The bride's mother's name, which this site itself briefly misread as Catherine Saurette while working from the same act, is confirmed as Julienne Saurette against an earlier record naming her directly. Two small errors, from two different sources, corrected within the same afternoon of actually looking. See The shrinking marriage pool for where this correction landed.

Believed: that on the strength of parish acts alone, Julien Oligny was the earliest documented literate Oligny on this line, having signed his own marriage act in March 1848.

Corrected by: two acts this project already held, each carrying a line nobody had read. The 1844 act had been read for its signature column and not for its witness clause; the 1821 act had been read for its filiation and not for its closing formula.

Now: Isaac Oligny is the earliest, and the story is better than the one it replaces. On 26 February 1821, at his own wedding at St-Constant, Isaac was twenty-three, and the act closes with the couple inside the clause: the witnesses “qui ont avec les époux déclaré ne savoir signer” — who declared, along with the bride and groom, that they did not know how to sign. On 16 April 1844, aged forty-six, he signed his daughter Clothilde's marriage act in his own hand. Both acts are sighted originals. Isaac Oligny learned to write as an adult, somewhere in the twenty-three years between them. Julien signed that same 1844 page, four years before his own wedding, so his earliest signature moves too.

What is not being claimed, and will not be. The acts give two end states and nothing in between. When he learned, how, why, and from whom are undocumented, and this site is not going to supply them. The 1838 rebellion falls inside that span; it is not offered as an explanation, because nothing connects the two but the calendar. A reader in 1900 holding these two acts could say that Isaac could not sign in 1821 and signed in 1844, and could say nothing else. Neither can we.

One more thing the 1821 act settles, in the other direction: Isaac's own father, Joseph Oligny dit Lauvergnat, stood as a witness at that wedding and is inside the same clause. He could not sign either. The line's illiteracy is now documented one generation further back than it was.

Believed: that no death record had been located for Joseph Oligny dit Lauvergnat or for his wife Marie-Thérèse Gaboriau dit Lapalme, and that the gap was being left open rather than filled with a guess.

Corrected by: her burial act — which this project already held. The death had been on file from compiled sources for months and on a sighted original since August 2026. The site was not guessing here; it was out of date with its own evidence register, which is a different failure and a worse one, because nothing on the page looked wrong.

Now: Marie-Thérèse was buried at St-Rémi on 22 February 1856, aged eighty-six. The act records her «décédée la veille» — died the day before — and names her épouse de feu Joseph Oligny. Graded Proven. A small thing came with it: the compiled sheets had given her death as 20 February, and the act moves it to the 21st. The sighted original beat the compilation by a day, which is the whole argument for sighting originals.

Joseph's half of the sentence stands, but not as it stood. His death is still unlocated. It is no longer open-ended: he witnessed Isaac's wedding on 26 February 1821, and his wife's burial in February 1856 calls him feu. He died inside those thirty-five years, and the page now says exactly that and no more.

How it was caught, which is the part worth keeping. Not by reading the page — a full-site audit read every template in source and flagged it as a stale absence: a sentence that is not wrong about the family, only wrong about what this project knows. Nothing on the page contradicted anything else on the page, so no amount of internal cross-checking would have found it. It took reading the site against the register, which is the only direction that catches this class.

Believed: that when Céleste Sorette signed her sister’s marriage act on 4 September 1827 she was thirteen, one day short of fourteen — and that her age was graded only Probable, because her birth date came from a compiled family sheet rather than from an act.

Corrected by: her own baptismal act, which this project has held, sighted and confirmed against the image, since 10 August 2026. It is dated 2 September 1813. The compiled sheet said the 5th.

Now: she was fourteen, and had been for two days. The age is Proven on the act, not Probable on a sheet. Her birth date is 2 September 1813, and this site can state it.

Why this one is worth publishing rather than quietly amending. The error was the right instinct applied to the wrong facts. This site was being careful — refusing to state an age above the grade of the source it thought was underneath it — and the caution was misplaced, because a better source had been sitting in the register for a month and was not being cited. A hedge is not automatically the honest choice. Checking whether the primary act exists comes first, and it was not done. The figure was live for a few hours and went out in the site’s feed, which is why it is here rather than silently fixed.

One loose end kept rather than tidied: her burial act recites an age that back-calculates to about October 1813, a month later than her baptism. Age recitals at burial are approximate, the baptismal act governs, and the discrepancy stays on the record instead of being dropped.

Believed: 1887 marked the Oligny family's first literate generation — the first Oligny to sign a parish act with an actual signature rather than a mark.

Corrected by: A batch of twenty-three Oligny marriage acts, 1784 through 1860, read cold and checked against the claim.

Now: the 1844 act carries four Oligny signatures, not one — Clothilde, her father Isaac, her brother Julien, and a fourth the act does not identify. A single 1860 wedding carries seven. Literacy arrived on this line decades earlier than the family narrative had it, and unevenly — some siblings signing, others still making a mark, in the same family, the same decade.

Amended September 2026. This entry originally read “Clothilde Oligny signed in 1844. Her brother Julien signed in 1848.” Both were true and both were incomplete: the 1844 signature column had been read, and the act's witness clause had not. See the entry below.

Believed: Isaac Oligny died at age 89, per his own burial record.

Corrected by: Isaac's own baptismal act, born 25 April 1797 — a document that had never been sighted until it finally was.

Now: Isaac was 87, not 89. A two-year age-drift, the ordinary kind that creeps into a burial record when the informant is grieving and working from memory rather than a document. Small on its own; a reminder that a death record's age is secondary information even when the record itself is an original (see Sourcing & Method on that distinction).

Believed: Marie-Françoise Moitié, born inside Fort Saint-Frédéric, was betrothed at thirteen to a Lieutenant Pierre Robineau — a romantic engagement to an officer, cut short when he was reported killed in action.

Corrected by: P.-G. Roy's full 1946 text on Fort Saint-Frédéric, read in its entirety rather than through a secondhand summary.

Now: The 1755 document was an engagement de servante — an unpaid domestic indenture, not a betrothal. A thirteen-year-old girl was bound into service, not promised in marriage. The claim that made it into this family's story for generations was more romantic than the record supports, and less honest about what actually happened to a child at a frontier fort.