Research Methodology
Every case study on this site exists because a record was ambiguous, a published account was mistaken, or a name was shared by more people than the evidence could separate. These pages are grouped not by family or by country but by the problem — because the problem is the transferable part.
A genealogical conclusion is only as good as the test it survived — and no conclusion is permanent. New evidence arrives, and the work is re-examined against it. These pages show that process, not only its results.
Most published genealogy presents finished answers. What it rarely shows is the reasoning that produced them: which sources were weighed against which, what was searched and found empty, and what the researcher believed before the evidence changed her mind. That reasoning is the part another researcher can actually use.
So the work here is organised by the kind of problem it solves. A study of a seventeenth-century Norman widow and a study of a nineteenth-century Brooklyn mat maker have nothing in common except the thing that makes them hard — and that thing is the method.
The Board for Certification of Genealogists puts it plainly: a statement about an ancestor, even one meeting the Genealogical Proof Standard, is not absolute or everlasting. When new evidence surfaces, the statement is re-examined to determine whether it remains valid. That re-examination is ordinary professional practice rather than an admission of failure, and several of the studies below are the record of it.
Too many men, one name
The commonest failure in genealogy is not a missing record. It is a record that fits more than one man, accepted because it fits the one you were looking for.
Spelling never separates two men of one name — clerks and curés spelled the same surname three ways on a single page. What separates them is something that cannot drift: a wage figure, a wife’s name, an occupation traced across decades, a parent named in a statutory register.
The Voyageur Years
Three men of one direct line carried the same given name across three generations, with cousins carrying it beside them. One North West Company ledger account survives, and it names no age, no parish and no father. One candidate is excluded on direct evidence; two remain; two wage years carry no evidence in either direction — which the study states rather than resolves.
Read the case studyThe Invisible Voyageur
Two men of one family, both called Paul Guilbault, in the same volumes in the same year. The clerks spelled both several ways. The wage column separates them completely: five hundred livres settling at 617.14, against three hundred and fifty settling at ninety-six.
Read the case studyOne Name, Three Men
A son vanishes into one of Scotland’s commonest surnames. The proof turns on a peculiarity of Scottish law — a statutory marriage names both parties’ parents — converting a shared surname into a documented identity, and excluding the Brooklyn watchman who shared the name and nothing else.
Read the case studyThe Brooklyn Mat Maker
Seven years and dozens of John Kennys in one city. What finally distinguished them was not a name at all but a career: an occupational progression traced through directories, used as a unique identifier where names could not be.
Read the case studyPierre Morin dit Champagne
Two men of one name served in the same regiment, listed in different companies seven pages apart on the 1668 muster roll. One became a founding ancestor; the other disappeared. A research note documenting exactly what separates them.
Read the case studyThe Cousin Who Wasn’t
A family tree claimed kinship with a signer of the Declaration of Independence. The bridge was one surname appearing on both sides of the Atlantic. Following each family to its documented origin split one name into three — and the kinship collapsed. A shared surname at the colonial-to-European transition is the most dangerous coincidence in genealogy.
Read the postOccupational Tracking: When Name Searches Fail
When your ancestor shares one of the commonest surnames in a city of 800,000, name-based genealogy hits a wall. Career progression, read across directories and censuses, becomes the identifier the name cannot be.
Read the postFour Words That Solved a Mystery
“Kenny, Elizabeth, wid. Richard.” One city-directory line, four words, and seven years of failed research resolved — because a widow’s entry names the husband no other source would.
Read the postCorrecting the published record
An error that reaches print acquires a life of its own. Databases copy monographs, trees copy databases, and a birth year sourced to nothing outlives everyone who could have checked it.
Go back to the original. A published account is derivative evidence and is cited as such; where a register and a database disagree, the register governs, and the discrepancy is named rather than smoothed away. The same discipline applies to conclusions published here: when new evidence arrives, they are re-examined against it, and the revision is shown.
Resolving Conflicting Age Evidence
A woman states her age under oath in 1672; she had given a different age at confirmation in 1664; a 1666 census gives a third. The published monograph followed the census. Why a self-stated age under oath outweighs a third-party estimate — and how a ten-year error survived in print.
Read the case studyThe Death That Never Was
For three centuries the databases recorded a death in 1704. Two daughters’ marriage contracts — a father “absent due to illness” in 1700, a mother styled widow by 1701 — proved the 1704 burial belonged to another man entirely.
Read the case study520 Livres for a Murdered Husband
An inherited account misdated the act by a year, miscast it as a post-suit settlement, and named the wrong parties. Read folio by folio, the deed is the clearest surviving proof that colonial law gave a habitant widow standing against a nobleman.
Read the case studyBorn in the Pays d’en Haut
Two parish registers thirty-four years apart are the whole documentary life of a man born in the interior to an Ojibwe mother. This study is the account of why the fur trade records once thought to fill the gap between them do not belong to him — and what remains when they are taken out.
Read the case studyThe Free Man of the Prairies
Two errors in the public record, neither sourced to a single primary document, had corrupted every generation above and below a Columbia River couple. Correcting them took four archive collections, two sacramental registers naming the same father on consecutive days, and a retrieved engagement contract signed with an X.
Read the case studyOne Parcel, Three Households
A notary’s hand produced a surname some readers transcribed as “Soudain.” Two later Crown records bearing the man’s own signature said otherwise. When a transcription and an autograph disagree, the autograph decides.
Read the case studyWhat absence proves
“No record found” is not a result. It becomes one only when you can say precisely what was searched, how far, and what would have been there if the answer were otherwise.
A negative search is only as wide as the range actually browsed, and its extent must be stated whenever it is cited. Equally: a gap in the record is not evidence of anything on its own. A man home between engagements appears in his parish register like everyone else, and a career can run for twenty years without the registers noticing.
When the Record Doesn’t Exist
How to turn “no record found” into evidence. A missing 1887 Chicago birth certificate, reconstructed from baptism records, cemetery cards, and a documented negative search — and why the documentation of the search is the part that matters.
Read the postWhen the War Burned the Records
In much of the Philippines the record you most want was destroyed in 1945. How to reconstruct a family from what survived — church salvage books, notarial registers, delayed registrations, and the U.S. records of relatives who emigrated.
Read the postThe Tide That Turned Both Ways
A family crossed the Atlantic twice and no migration record documents either journey — because the United Kingdom kept no inward passenger lists before 1878. Eighteen sources, three crossings, zero migration records, and a proof assembled entirely from what the absence implies.
Read the case studyPhilibert Couillaud dit Roquebrune
No birth record, no marriage record, no death record, and a documentary void of fourteen and a half years. What survives is a mark on a notarial obligation — and a dit name that became a surname carried by hundreds of thousands.
Read the case studyNames the records resisted
Indigenous names written by French priests, nicknames a register could not hold, dit names that swallowed the original surname. When the name itself is unstable, ordinary searching fails before it starts.
Reconstruct the variant set before searching, not after. An Algonquian name passing through the ears and pens of three priests produces spellings no index will connect — and the ecclesiastical 8 ligature alone indexes as 8, ou, v, w and 3. Where a published transcription and its own manuscript photograph disagree, go to the manuscript.
The War Chief’s Wife
Eleven primary records, eleven spellings, three priests, nineteen years. Identifying Geneviève Abitakijikokwe required examining a published transcription that contradicted its own manuscript photograph, and returning to the original register to resolve it.
Read the case studyThe Woman at the Grave
A woman stood witness at a two-year-old’s burial in 1801 and no one had identified her. Tracing one Algonquian name across five colonial spellings — Mador, Mabre, Missinebi8e, mesepik8e, Messinabikwe — revealed an Ottawa woman from Michilimackinac who had buried her own daughter in the same cemetery.
Read the case studyThe Abitakijikokwe Discovery
For two hundred years she was recorded only as sauvagesse. Systematic work across five Quebec parishes recovered her Ojibwe name from a single marriage record — and with it the eight-ligature problem that has to be solved before such a name can be searched at all.
Read the case studyThe Daughter Remembered as Tonang
A family kept its matriarch only as a nickname that no register could hold. Triangulating the grandparents named across three generations of Kalibo baptisms makes the documented case that “Tonang Quimpo” was Petrona Quimpo.
Read the case studyFinding an Italian Ancestor in French-Canadian Research
A dit name — Lepiedmontois, “the Piedmontese” — revealed a soldier from Racconigi hiding among ten thousand French settlers. What a dit name conceals, and what it can be made to reveal.
Read the postHidden Protestants
Among the first women who settled Quebec were Huguenots who converted to survive. Tracing a buried heritage through temple registers and abjuration records — a religious identity the Catholic registers were designed not to preserve.
Read the postReading fur trade records
Company account books, engagement contracts and mission registers document lives the Quebec parish system never described. They are also the records most often over-read — and the group where new evidence has most often required earlier conclusions to be re-examined.
An account book gives you a name and a number — no age, no parish, no father’s name. Figures are the discriminator and spelling is noise; an index date describes the volume, not the man; duplicate volumes are not independent witnesses; and a page that foots tells you your reading is right, while one that does not tells you something has been misread.
The North West Company: A Genealogist’s Guide
What the NWC was, where its records are held, which indexes to search, and how to reach the plates. Including the limitation nobody warns you about, and the microfilm substitution card that can make a volume look searched when it never was.
Read the guideFinding a Gabriel Guilbault in the NWC Records
Three volumes, a cross-post balance, an account marked settled. Then the parish registers, tested properly, showed the man found was not the man expected. The search method is sound and unchanged; it is the sixth step that decides an identification, and this post is what that step looks like.
Read the postThe Munnick Annotations
How to read the Catholic Church Records of the Pacific Northwest: the B-, M-, S- and A- annotation system, transcription and annotation as separate evidentiary layers, tracking a name through phonetic variants — and where Munnick can be wrong, which makes the resource more useful rather than less.
Read the postFollowing the Canoe Routes
A man appears in Quebec records, disappears for years, then resurfaces with a family that seems to have materialised from nowhere. The explanation is geographic — and how long a journey took is often the fact that settles an identification.
Read the postMarriage à la façon du pays
Unions between French-Canadian traders and Indigenous women were social and economic institutions, and the parish system recorded them — when it recorded them at all — only in retrospect. Where to find these families, and what a register does to a marriage it never witnessed.
Read the post“Voyageur et maintenant agriculteur”
One phrase in a 1798 baptism, usually read as a man wearing two hats. The adverb in the middle says otherwise. Five occupational designations across thirty-five years — including a change of trade datable to the day by a word squeezed above the line.
Read the postDNA as corroboration
Genetic evidence confirms or refutes a documentary hypothesis. It does not generate one, and a segment does not name an ancestor.
Documentary evidence carries the identification and DNA corroborates it. Triangulation proves a shared segment and a shared deep origin; it does not, on its own, prove a named common ancestor — and where it doesn’t, the honest answer is to leave the question open and say why.
The Donaghmoyne Network
Four couples married in one County Monaghan parish between 1841 and 1858; their descendants emigrated to five American destinations. A reconstructed paternal kit validated across three platforms, triangulation across thirteen confirmed matches on chromosome 2, and a Y-DNA signature.
Explore the networkThe Segment That Pointed Two Ways
A chromosome painting called one stretch Indigenous American — a hypothesis, not an ancestor. Triangulation across three platforms confirmed the segment was real and paternal, and led two ways. Both arms documented Ojibwe; the named common ancestor between them remains, honestly, the open question.
Read the case studyTwo Families, One Story
Three marriages bound two families across two continents — but were they always one family? Turning from parish records to the DNA of living descendants narrowed a century-old question to a single missing generation.
Read the postOne Parish, Five Destinations
Distinct clusters of matches pointing to relatives scattered across five American cities, all tracing back to one small Irish parish. How do you prove that families who emigrated decades apart were connected before they left?
Read the postThe Mathematics of French-Canadian Cousinhood
Why two people of French-Canadian descent are almost always ninth or tenth cousins, and what pedigree collapse and the founder effect do to a match list before you have read a single record.
Read the postWhen One Ancestor Appears Twice
How the same woman becomes an eighth great-grandmother twice. Pedigree collapse traced through a Fille du Roi whose descendants married each other eighty-five years later — and what it means for interpreting your matches.
Read the postReading a document as an object
Before a record is evidence it is a physical thing — bound in an order, filmed by someone making decisions, indexed by someone else, and displayed by a database that may render it wrongly.
The register governs over the database, and the plate governs over the transcription. A finding aid’s dates, spellings and standardised surnames are editorial decisions rather than statements in the record — and what a filmer chose to omit, or an archivist noted as missing, is part of the evidence too.
The Bertault–Banne Trial
A sixty-three-leaf criminal file, bound partly in reverse, its chronology reckoned in weekdays rather than dates. The dossier had to be rebuilt from the datelines and catchwords inside it before it could be read at all — and it corrects a forename, a birth year and a murder weapon that decades of secondary accounts had wrong.
Read the case studyAround the Font
The 1666 census presents a household as one entry among the habitants. Ten recovered baptism registers told a different story — who stood as godparents, and who officiated — reframing the household from patron-client to quasi-familial.
Read the case studyThe 1926 Irish Census: A Genealogist’s Guide
Sealed for one hundred years and released in April 2026 — nearly three million individuals. What the census contains, how to search it effectively, its known data limitations, and the citation conventions to use.
Read the guideThe 1863 Lake Map
The first time individual Prince Edward Island tenant farmers were recorded by name on a commercial map. What a cadastral survey can prove that a parish register cannot — and what it cannot prove on its own.
Read the postHidden in Plain Sight
FamilySearch Full-Text Search reads the text inside digitised documents rather than indexed names. One query surfaced a multi-generational Philippine paper trail that had been invisible for decades — and changed what is possible in a records environment shaped by wartime destruction.
Read the postWhen Newspapers Tell the Whole Story
No birth record, no marriage record, no official death record. What to do when the vital records fail entirely, and how far a newspaper run can be pushed as substitute evidence.
Read the postResearch conducted to these standards
Documentary research and narrative family history for clients who want conclusions that will hold — with the reasoning shown, the sources cited to the plate, and the open questions stated as open.
Explore services Begin a research inquiryStoryline Genealogy
From Research to Story