Your ancestor’s marriage record names her mother as Johanna. Her death certificate calls the same woman Anna.

An online family tree offers a third name.

Which should you enter in your own tree?

The temptation is understandable: choose the earliest record, trust the official certificate or accept the name appearing most often.

But genealogical research is not a vote. Three websites may repeat one mistaken transcription. An original document may contain information supplied by someone who never knew the family.

A disagreement is not a reason to abandon the research. It is a reason to examine how each answer reached the page.

State the question precisely

Begin with a question small enough to investigate.

“Who was my ancestor?” is too broad. “Who was the mother of the Maria Jansen who married Pieter Smit in Delft in 1854?” gives you a person, an event and a relationship to test.

Record the competing answers without choosing between them yet.

Then check something more fundamental: do the records concern the same Maria?

Compare spouses, occupations, residences, relatives and movements. Two documents cannot meaningfully contradict each other about one person until you have established that they probably describe that person.

An address or a repeated family network may help make that connection.

Preserve what each record actually says

Open the complete record image whenever possible. Read the headings, surrounding entries and any continuation pages.

Transcribe the disputed passage before interpreting it. Preserve the spelling and mark uncertain letters explicitly. If a name appears to be “Anna,” do not transcribe “Johanna” because that is the answer you expect.

Keep three things separate:

• Observation: The mother’s name appears as “Anna de Vries.”
• Interpretation: This differs from “Johanna de Vries” in the marriage record.
• Possible explanation: A shortened name, reporting error or different person could account for the difference.

The possible explanation is a hypothesis—not yet a finding.

Save the archive reference, collection, volume, entry or page number, image identifier and link. Someone else should be able to find the material and check your reading.

Identify the voices behind the statements

A document has a creator, but the creator may not be the source of every fact.

A registrar wrote the certificate. A declarant supplied information. Some details may have been copied from supporting documents. A later indexer selected and transcribed particular fields.

Ask who supplied the disputed information and what opportunity that person had to know it.

For Dutch civil records, pay attention to roles such as aangever—declarant—and getuige—witness. Do not assume that everyone named in the record supplied the family details.

The same principle applies worldwide. A census return might reflect answers given by one household member. A death registration might rely on a neighbour. A marriage file might include an extract from an earlier birth record.

When the informant or information source cannot be established, say so. Unidentified knowledge is not automatically worthless, but it should not be treated as direct knowledge.

Compare statements, not entire documents

Avoid labelling a whole certificate “reliable” or “unreliable.”

A death record may provide well-supported information about the death while giving uncertain information about the deceased’s parents. A marriage record may accurately name a spouse but contain an estimated age.

A comparison table helps you evaluate the particular claim.

The following example is fictional and illustrates the method.

Source	Mother named	Information origin	What needs checking?
Maria’s marriage record	Johanna de Vries	Not yet established	Examine supporting documents
Maria’s death record	Anna de Vries	Declarant identified as a neighbour	Investigate the neighbour’s knowledge
Online family tree	Anna de Vries	Cites the death index	Trace the claim to the original
Candidate birth record	Johanna de Vries	Father recorded as declarant	Establish that this child became the bride

Notice that the candidate birth record does not settle the question by itself. You must connect the child in that record to the adult you are researching.

Count information origins, not websites

Suppose Anna appears in a death index, two databases and six online trees.

That looks persuasive—until you discover that every entry ultimately depends on the same death certificate.

You have found repeated access to one statement, not nine independent confirmations.

Likewise, a marriage attachment containing an extract of a birth record can be valuable, but its birth information derives from that earlier record. The extract and the birth entry should not be counted as two independent reports of the birth.

Trace the chain wherever possible:

Historical statement → copy or extract → index → online tree

Independent agreement comes from evidence whose relevant information did not simply travel along the same copying chain.

Check whether the conflict is real

Some apparent disagreements concern different events or different levels of detail.

A child could be born on 12 March, registered on 13 March and baptised on 16 March. Those dates do not necessarily conflict.

One record might name a village, another its municipality and a third a nearby city better known abroad. That could reflect geography rather than error—but you must investigate the historical boundaries.

An age-derived birth year is also different from a recorded birth date. Someone described as 63 at death did not necessarily have their exact birth date recorded.

Before deciding that a clerk made a mistake, ask whether you are comparing like with like.

Test competing explanations

For the fictional Anna–Johanna disagreement, several explanations remain possible:

• The names refer to one woman who used both forms.
• The death informant supplied an incorrect name.
• One passage has been misread.
• The records concern different women.
• A family relationship is more complicated than assumed.

Choose research that can distinguish between these possibilities.

A birth record may test the parents’ names. A population register may connect the child to the later household. The mother’s own marriage and death records may show how her name was recorded elsewhere.

If you suspect two people with the same name, reconstruct both. Researching only the preferred candidate can hide evidence against your conclusion.

Keep a search log, including unsuccessful searches. Record the collection, place, period, variants and access limitations. “Not found” means much more when the reader knows what was searched.

Explain the conclusion—and the disagreement

A useful conclusion does more than announce the preferred answer.

It identifies the person, cites the relevant sources, explains their connections and addresses the conflicting statement.

For our fictional case, a provisional note might read:

The evidence currently favours Johanna de Vries as Maria’s mother. The candidate birth record names Johanna, and household records connect that child to the woman who later married Pieter Smit. The death record names Anna, but its declarant was a neighbour whose knowledge of Maria’s parents is unestablished. A reporting error is possible, although not demonstrated. Maria’s marriage attachments should be examined before the conclusion is finalised.

In real research, attach precise citations to the statements in that paragraph.

The Board for Certification of Genealogists’ standards emphasise focused research, documentation, analysis and resolution of conflicts before drawing conclusions. A written explanation makes that reasoning available for others to assess. BCG: Ethics and Standards.

You do not need to hide uncertainty to produce useful research. You need to make its boundaries clear.

When the evidence does not resolve the question

Sometimes the necessary register has not survived. An informant remains unidentified. Two candidates still fit.

Do not force a choice.

Write what is established, what remains disputed and which explanations are still possible. Keep an uncertain parent-child relationship out of the confirmed structure of your tree, or label it clearly according to your software’s capabilities.

Then choose a specific next action:

Search the marriage attachments for the civil marriage in Delft in 1854, looking for the bride’s birth extract and any documents concerning her parents.

That is more useful than “do more research.”

A practical Research Desk exercise

Choose one disagreement already present in your notes.

1. Write a precise research question.
2. Preserve and transcribe each conflicting statement.
3. Check that the records concern the same person.
4. Identify the creator, informant and any copying chain.
5. Compare genuinely independent evidence.
6. Test at least one alternative explanation.
7. Write a cited conclusion—or an explicit statement that the question remains unresolved.
8. Record the next research action.

A contradiction is often where the most valuable work begins.

It asks us to slow down, question a familiar answer and distinguish what the records say from what we hope they mean.

The goal is not a family tree without disagreements. It is a family history whose conclusions can withstand careful examination.

How do you resolve conflicting genealogy evidence?

Preserve and transcribe each conflicting statement, confirm that the records concern the same person, and identify who supplied the information. Trace copying between sources, compare independent evidence and test alternative explanations. Write a cited conclusion that addresses the disagreement, states remaining uncertainty and identifies the next research action.

Sources and further reading

• Board for Certification of Genealogists: Ethics and Standards — research planning, documentation and reasoning from evidence.
• BCG: Skillbuilding—Evidence Correlation — bringing evidence together.

All illustrations in this article are AI-generated and are not reproductions of the records discussed.