How online trees invent ancestors

When we audited my family tree at the end of September 2026, it held 6,109 people and 652 direct ancestors. Six parent or spouse links were wrong and were removed. Twenty-one conflicts were flagged for review. A woman was born a year after her own wedding. A man was dead a year before he signed a contract. None of this was random. The errors fall into patterns, the patterns come from how online trees are built, and the patterns are the lesson.

Eight ways a tree goes wrong

Every example below is from my own tree, and every correction is backed by a record we have read. That is the point of listing them: not that the tree was bad, but that each kind of error has a signature you can learn to see.

Pattern	What the tree said	What the record says
1. Impossible dates	Marie Jeanne Manart born 1688	She married in February 1687 (St-Martin, Cambrai). Born before about 1667.
1. Impossible dates	Benjamin LeBorne × Marianne Perret married 1804; their son born 1803	The son’s baptism: born 11 June 1808. Right day, wrong year.
2. Wrong generation	Pouponne Darensbourg, daughter of Hubert (b. 1783) and Pouponne Perret (b. 1789)	Her son was born 1804. Hubert was her brother. Her children’s baptisms name Charles Darensbourg × Françoise Lavergne.
3. Death from nowhere	Benjamin LeBorne died 1829 (FamilySearch)	He signed at his son’s marriage contract, 7 July 1830. Inventory of Nov 1831 calls his wife a widow.
3. Death from nowhere	Felix Borne died 1933	Death index, two obituaries and the grave: 9 September 1926.
4. Father and son merged	“Birth & baptism” source for Antoine Benjamin Borne (b. 1808)	The attached record is his father’s baptism of 1779. Both were baptized Antoine Benjamin.
5. Sisters merged	“Marie Arcesse Perret, b. 1779” as Marianne Perret	Arcize Perret, wife of Maximilien Trosclair, is Marianne’s sister. Marianne is Marianne in every record.
6. Names borrowed from strangers	William Emanuel McElveen Sr	“Emanuel” belongs to a William of Bulloch County, Georgia. His pension file has no middle name.
7. Phantom parents	Michel LeBorne, son of Nicolas LeBorgne × Josette Darembourg	No record contains a Darembourg. WikiTree removed the pairing in 2020. Parents unknown.
8. Children who don’t exist	Two “Jean Jacques Borne, b. 1708” in Michel’s family	No such baptism at Notre-Dame-de-la-Chaussée. Marked for removal.

Ornate 1888 lithograph titled Family Record, with blank oval frames for names arranged above a boat full of people on a river, surrounded by scenes of childhood and old age — A printed family record of 1888, sold blank for the owner to fill in. The ovals were empty for a reason. An online tree is the same form with the ovals pre-filled by strangers. Chapman Bros., Family Record, lithograph, Chicago, 1888. Library of Congress.

Where each pattern comes from

Impossible dates happen because nobody checks a date against its neighbours. A birth year typed next to a marriage year never gets compared with it. The fix is mechanical: every parent at least fifteen at a child’s birth, every mother under fifty, no event after a death, no child before a marriage without a note explaining it. A spreadsheet can do most of this. The tree software mostly does not, or does and is ignored.

Wrong generation is what happens when a name is unusual and a searcher finds one. Pouponne Darensbourg was attached to a Hubert and a Pouponne Perret because those were the Darensbourgs the searcher found, and nobody noticed that a woman cannot be the daughter of a man born in 1783 and have a son in 1804. In a family that reuses a name across two generations, the only protection is a record that names the relationship, and her children’s baptisms name her parents outright.

Death dates from nowhere are the most insidious, because they look like facts. A year with no source is almost always somebody’s estimate that lost its “about” in the copying. Benjamin LeBorne’s “1829” was contradicted by a contract he signed in 1830 and settled by an inventory in 1831, found by reading an unindexed film page by page, as described in When the records burned.

Merges, of father with son or sister with sister, come from the French Catholic habit of naming a boy for his father and a girl for her aunt, combined with search engines that return both. Antoine Benjamin Borne the elder and Antoine Benjamin Borne the younger, baptized 1779 and 1808, had the elder’s baptism attached to the younger’s record as his “birth & baptism”. The two-identifying-points rule exists for exactly this.

Detail of a 1752 French map of lower Louisiana: the Mississippi winding past Nouvelle Orléans to the Gulf, with Lac Pontchartrain above and the settlement marked “les Allemands ou Carlstain” on the river above the city — The German Coast, “les Allemands”, on d’Anville’s map of 1752, a few bends above New Orleans. Most of the merged and misdated people in this post lived along this stretch of river, in a handful of families that shared a handful of names. J. B. B. d’Anville, Carte de la Louisiane, 1732, published 1752 (detail). Library of Congress.

Borrowed names, phantom parents and invented children are the signature of trees copying trees. Nobody found a record saying William McElveen’s middle name was Emanuel; somebody found another William Emanuel McElveen and took the name. Nobody found a Josette Darembourg; the pairing was on one site, got copied to another, and is now on FamilySearch where it will be copied again. The two Jean Jacques Bornes of 1708 appear in no register. They are not errors of reading. They are errors of trust.

An index is not a record

One more pattern deserves its own heading because it underlies several of the others. Most of what online trees cite are indexes: a line in a database that says a baptism or marriage of such a name happened in such a parish in such a year. The index is made by a volunteer or an algorithm reading the register, and it carries their reading errors, their guesses at a faded surname, and sometimes the wrong page. It also carries almost none of the content. The marriage index that led us to Claiborne Thibodaux’s 1883 wedding gives the date and the two names. The register entry, which we have not yet seen, will give the parents of both parties and the witnesses.

Felix Borne’s 1880 marriage index entry names his parent as “Benjamin W Borne”, which is wrong; his father was Antoine Benjamin, and the only Benjamin W. Borne in the family is Felix’s own son, Benjamin Walter, born eleven years earlier. An indexer’s slip, probably, and harmless once you know. But a tree built on that index line would have Felix as the son of a man who was in fact his child. Several of the errors in the table above began the same way, with an index trusted where a record was needed. The cure is always the same: open the image.

How to audit your own tree

Export it and work on a copy. Never edit the original file. Every change goes in a log with the date, the old value, the new value and the record that justified it.
Run the date checks. Parents at least fifteen; mothers under fifty; nothing after a death; no child before a marriage. Every failure is either a typo or a merged person, and both matter.
Open every source. For each fact, ask: have I seen this record, or have I seen a line that says a record exists? A tree citing “Louisiana Births and Christenings” is citing an index. The index can be wrong, and the image is usually one click further.
Look for duplicates and merges. Same name, same parish, overlapping dates: assume two people until a record says one.
Treat online trees as pointers. They are where leads come from. They are not evidence. A fact that appears in forty trees with no source is one unsourced fact, copied forty times.
Keep the uncertain ones uncertain. Benjamin LeBorne’s burial place is still unknown; the tree’s “Edgard” has no source, and it stays flagged rather than deleted or accepted.

Illuminated manuscript page of about 1218: a Tree of Jesse, with Jesse asleep at the foot, a king and the Virgin seated in the vine above him, and prophets in roundels at the sides — A Tree of Jesse from a psalter of about 1218: the original family tree, and an argument rather than a record. The genre has never quite shaken the habit. Psalter of Blanche of Castile, f. 15v, c. 1218. Bibliothèque nationale de France.

What the audit changed

Six links removed, twenty-one flagged, and a checklist for the live Ancestry tree so the corrections could be made there too. Since the audit, research has moved several items again: the audit itself still called Benjamin LeBorne’s 1829 death “unproven” when it is now disproved, and still gave Marie Jeanne Manart “about 1673” when the right value is “before 1667”. Those corrections are listed on the audit’s own page as known issues, dated, because a report that cannot be corrected is as dangerous as a tree that cannot be. The habit of correcting yourself in public applies to audits too.

Six hundred and fifty-two direct ancestors sounds like a lot. It is about what you would expect for a tree that reaches the 1600s in a few lines and the 1800s in most, and the arithmetic in how many ancestors you actually have says why the number can never be as large as the doubling suggests. What matters is not how many names are in the ovals. It is how many of them you have seen written down by someone who was there.

What the papers prove

Marie Jeanne Manart married in 1687 and so was born before about 1667 (St-Martin, Cambrai).
Antoine Benjamin Borne was born 11 June 1808 (baptism, Edgard).
Pouponne Darensbourg’s parents were Charles Darensbourg and Françoise Lavergne (her children’s baptisms, 1804, 1806).
Benjamin LeBorne was alive on 7 July 1830 and dead by 7 November 1831 (marriage contract; Ascension inventory).
Felix Borne died 9 September 1926.
William McElveen Sr had no recorded middle name (pension file S.21880; Sumter estate papers).

What isn’t proven yet

Benjamin LeBorne’s burial place.
Michel LeBorne’s parents (the phantom pairing is removed; nothing replaces it).
Whether the two “Jean Jacques Borne, 1708” entries are copying errors or a real child in another parish. No baptism has been found; removal is pending.

Sources

Borne Ancestry Audit, 29 Sep 2026, with known issues added 5 and 17 Oct 2026 (White Glove Genealogy, client report).
Archdiocese of New Orleans, St. John the Baptist (Edgard) registers: baptism of Antoine Binjamin, 26 Jul 1808 (B3, 119); marriage of Benjamin Le Borne and Marianne Perret, 1 May 1804 (M1, 120v–121); marriage of Maximilien Troxler and Arcize Perret, 19 Mar 1804.
Sacramental Records of the Roman Catholic Church of the Archdiocese of New Orleans, vol. 8 (SLC B17, 75 and 128) and vols. 9–10.
Marriage contract of Antoine Benjamin Borne and Marie Augustine Latour, 7 Jul 1830, exhibit in suit No. 1906, First District Court, Orleans Parish (1848); FamilySearch DGS 007670423, image 1323.
Ascension Parish, Court of Probates, Inventories and Sales 1825–1832, pp. 110–111 (19 Nov 1831); Ancestry collection 9067.
Archives départementales du Nord, Cambrai, St-Martin, marriages, 4 Feb 1687; Valenciennes, Notre-Dame-de-la-Chaussée, baptisms 1694–1712.
William McElveen, pension file S.21880, NARA M804; Sumter District estate papers, 1837–1841.
Orleans Parish death index and obituaries of Felix Borne, Sep 1926 (see the first post in this series).