The Evidence Problem in Genealogy

How a forensic definition of evidence changes one question: Who are the biological parents of Earl Mack Douglas?

Research Journal Note

I’m putting this here as a snapshot of my process. Everyone researches differently. This is not a sermon. It’s not a universal method. It’s me writing out what I’m doing, how I’m organizing it, and how I’m trying to keep my own brain from poisoning the case.

---

---

Genealogy has a dirty little secret

Most family trees are not built on evidence. They’re built on agreement.

A record agrees with a hunch. A hint agrees with a desire. A name agrees with a name. Everyone nods. Everyone clicks “Save.” Case closed.

That is not research. That is a group hallucination with citations.

Science would call it contamination. Forensics would call it mishandling.

And the dead would call it what it is.

A second erasure.

Here’s the exact checklist I use so I don’t accidentally invent Earl’s parents.

Most family trees aren’t built on evidence. They’re built on agreement.

Evidence is not a screenshot. Evidence is not a story I want to be true. Evidence is not “it’s in a tree so it must be real.”

Evidence is what survives testing.

Evidence is what can be dragged into a bright room, under harsh light, in front of someone who wants it to be wrong, and it still holds its shape.

So this is the rude thing I’m doing in my own work. I’m defining evidence the scientific and forensic way, then applying it to one case file that will not stop staring back.

My great grandfather, Earl Mack Douglas, born January 23, 1896, Duluth, Minnesota.

And the record says: No parents listed.

Not unknown. Not unclear. Not “possibly.” Blank.

That is not a gap. That is a scene.

No parents listed is not a gap. It’s a scene.

A logical definition of evidence and proof

In strict logic, evidence is the pile of premises I slam onto the table like I’m emptying my pockets after a night I barely remember. A census line. A baptism note. A DNA segment. A date scribbled by a clerk who wanted to go home. Each one is a statement I’m asking the universe to accept as true.

They don’t convict the conclusion. They circle it. They tighten the air. They make it harder for my story to keep breathing.

Proof is the moment the door locks.

Proof is the formal chain, the steel rails of inference, where every step clicks into the next and the conclusion is no longer a guess, no longer a mood, no longer a narrative I’m sweet talking into existence. If the premises are true and the reasoning is valid, the conclusion cannot escape.

That’s the difference.

Evidence is smoke in the hallway. Proof is the fire marshal pointing to the exact room it started in, then showing the burn pattern, then daring anyone to deny it.

Evidence is smoke in the hallway. Proof is the room it started in.

Evidence is not a record because it exists. Evidence is a record that survives testing.

---

Proof language vs case language

One of the first problems I keep catching in myself is vocabulary. Genealogists love the word “proof” because it sounds like a gavel. It sounds like certainty. It sounds like sleep.

Science does not start with proof. It starts with a hypothesis. Forensics does not start with proof. It starts with a question and an evidence bag.

So the Earl version of this looks like hypothesis writing, not name claiming.

Instead of “Earl’s parents were probably X and Y,” I write:

“Earl Mack Douglas’s biological mother was a woman living in Duluth in late 1895 and early 1896 whose documentary footprint intersects Earl’s earliest custody trail, and whose genetic network fits what the maternal side should look like.”

Instead of “Douglas must be the father because that’s the name,” I write:

“The surname Douglas attaches to Earl at a specific time through a specific person or institution, and that attachment could be biological, step, foster, or administrative.”

That framing keeps the case attackable. Which is the point.

A conclusion that cannot survive attack is not a conclusion. It is a wish wearing a necktie.

A conclusion that can’t survive attack is a wish wearing a necktie.

Evidence discriminates, or it’s confetti

Here’s the forensic definition that matters to me:

Evidence separates one explanation from another.

If a piece of “evidence” fits five different stories equally well, it is weak. It’s set dressing. It’s the kind of thing that feels like progress while the case stays unsolved.

In Earl’s case, a few temptations keep showing up:

• A random “Douglas” in Duluth in 1896 is not a win. Duluth is full of working men, common names, and disappearing people.

• A single census entry with a rounded age is not a win.

• A Find A Grave memorial is not a win.

• A tree that claims parents with no sources is not a win. It’s a bedtime story typed in public.

Those things can still be useful. They can be leads. Doorways. Manageable rabbit holes.

But they don’t get promoted to “conclusion” just because they exist.

The question I keep returning to is simple:

What does this record rule out about Earl’s parents?

If the answer is “nothing,” it stays what it is. A note. Not a conclusion.

Every record is a witness

Forensics doesn’t ask, “Is this document real?”

It asks, “How trustworthy is this testimony?”

Records are not holy. Records are human. Humans lie, forget, guess, round numbers, protect reputations, avoid shame, and rewrite history with a smile.

So in this file, every record gets treated like a witness. Four questions keep repeating in my notes:

1) Who created it?
A clerk. A minister. An enumerator. A hospital ledger keeper. A county official. An undertaker. A neighbor with a pen.
If Earl’s birth record is blank, that blank was made by a human hand, inside a system, for a reason.

2) When was it created relative to the event?
A record created near the birth event has less memory rot. A record created decades later can be grief plus guessing.

3) How would they know?
Direct knowledge or hearsay. Eyewitness or family polish.

4) What happens if they are wrong?
Some records punish error. Some reward it.
A census can be a drive by interview. A death certificate can be rushed and secondhand for birth details. An obituary can be public relations for the dead.

None of this means “throw it out.” It means rank it.

A record is a witness. It gets treated like one.

How I Interrogate a Record: The Art of Scrutiny

I don’t ask if a record exists. I ask if it can survive pressure.

If it can’t survive scrutiny, it doesn’t get to shape the story.Evidence isn’t what supports my theory. Evidence is what can destroy the wrong ones.

1) What exactly is the claim?
Not the story. The claim.

2) Is it original, derivative, or an index?
An index points. An original speaks.

3) Who is the informant, and how would they know?
Eyewitness or hearsay.

4) How close is it to the event?
Days is better than decades.

5) What incentive did the creator have?
To tell the truth, to move a case forward, to protect someone, to get paid, to go home.

6) What does this rule out?
If it rules out nothing, it’s not evidence. It’s decoration.

7) What would falsify it?
What new record would make this collapse cleanly?

Verdict: strong, moderate, weak, or lead only.

EXHIBIT A: Census line (household snippet)

1905 Wisconsin Census: Not “unknown.” Not “unclear.” Adopted son. The system found room for a label, not a mother.

Names and roles shown: Lewis, Julia L. (Head), [Margaret] (Sister), Douglas, Earl (Adopted son).

1. Claim: At enumeration, Earl Douglas is living in this household and is described as an adopted son.

2. Original or derivative: This is an image of the enumerator’s sheet, closer to “original” than any index, but still a human transcription.

3. Informant: Unknown. Could be Julia, Margaret, a neighbor, or the enumerator guessing fast.

4. Timing: Strong for “who lived with whom that day.” Weak for birth facts. This is not a birth record.

5. Incentive: “Adopted” can mean legal adoption, informal placement, foster, or respectability language. It can also mean “don’t ask.”

6. Rules out: It does not prove parents. It does suggest a custody story and a surname attachment inside a household.

7. Falsifier: A guardianship file, adoption decree, county relief record, or earlier placement record that contradicts the relationship or timeline.
   

   Verdict: Moderate evidence for custody and household context. Lead only for biology.

Chain of custody, because Earl’s evidence has been handled

In the lab, chain of custody is non negotiable. In forensics, it’s a courtroom lifeline. In genealogy, it’s the difference between evidence and a rumor that learned how to print.

This is the rule that keeps saving me:

An index is not evidence. An index is a direction to evidence.

Earl edition:

If a database says “Earl Mack Douglas, born 1896, parents unknown,” that isn’t the exhibit. That’s the receptionist pointing at the file cabinet.

So my notes keep tracking:

• Did I see the original image or only a transcript?

• Is this original, clerk copy, abstract, or compiled derivative?

• Is it complete or cropped?

• Where does it live, and how can someone else retrieve it?

When chain of custody is unclear, the record feels handled. Touched. Smudged. Misread. Reposted. Retyped. Reimagined.

And that’s how parents stay ghosts.

When chain of custody is missing, the evidence feels handled with bare hands.

Controls, because bias is always in the room

Science uses controls because humans are a biased instrument.

We do not just observe reality. We negotiate with it.

In this file, the control is competing hypotheses. Earl’s parents don’t get treated like a single answer. They get treated like suspects in separate rooms.

The Earl hypothesis list:

• H1: Earl’s mother was local to Duluth and shows up in city systems: churches, hospitals, charity, directories, ward networks.

• H2: Earl’s mother was transient and shows up in port city systems: boarding houses, short term work, sickness, charity intake, court trouble, placements.

• H3: Earl’s mother is detectable through DNA clustering even if paper refuses to name her.

• H4: Earl was placed informally or through an institution and renamed, and “Douglas” is administrative, not biological.

• H5: Earl’s father is detectable through DNA clustering even if paper refuses to name him.

• H6: The blank parent fields are not missing info. They are policy, stigma, or deliberate omission.

Then the uncomfortable part begins.

The evidence gets asked to discriminate.

A candidate mother who cannot be placed in Duluth during the conception and birth window weakens. A paternal cluster anchored to a family with no plausible path into Duluth weakens. If “Douglas” appears only after placement or custody change, the entire story shifts.

---

Replication, Earl edition

If science has a spine, it’s replication. Not because scientists are pure. Because they’re not.

So when I write a claim about Earl’s parents, I keep circling a test.

Could someone else pull the same records, follow the same path, see the same conflicts, understand why I ruled out other candidates, and end up in the same place?

Whenever the argument starts leaning on “trust me,” it stops feeling like evidence and starts feeling like confidence.

Earl deserves better than confidence. He deserves receipts.

Falsifiability, or the case turns into fog

Good evidence can break a theory. That’s the point.

If nothing could change my mind, then I’m not researching. I’m defending a belief system with footnotes.

So the file keeps a list of things that would break my favorite version of Earl’s story:

• A baptism record naming parents that do not match leading candidates.

• A guardianship or placement record naming a mother or documenting a surrender.

• County home or poor relief intake linking a woman and infant Earl by name or timeline.

• DNA triangulation that anchors Earl’s paternal side to a different ancestral couple than the current working cluster.

• Proof that “Douglas” first appears as foster, step, or institutional label, not biological surname.

If a conclusion can survive anything, it usually means it’s vague enough to be immortal.

If nothing could change my mind, I’m not doing research. I’m doing belief maintenance.

GPS, translated into forensics, with Earl on the table

Genealogy already has a rigorous framework: the Genealogical Proof Standard. The problem is how easy it is to treat it like a bumper sticker.

So in this file, the GPS reads like protocol.

A conclusion earns the word “proven” only after five things happen:

• Reasonably exhaustive research. Not a hint. Not a quick search. Wide enough that silence starts meaning something.

• Complete, accurate citations. Every claim gets a trail back to the source. Receipts.

• Analysis and correlation. Records get weighed, compared, lined up, and forced to speak to each other.

• Conflict resolution. Contradictions get confronted, explained, or used to kill the wrong theory.

• A sound, coherent conclusion. Not a pile of documents, an argument that can survive cross examination.

That’s GPS.

Not a single smoking gun.

A full autopsy.

The GPS isn’t a slogan. It reads like protocol when the case is hard enough.

No parents listed is not a mystery. It’s an evidence problem.

This is the heart of Earl’s file.

A child appears in a record with parent fields blank.

My nervous system wants a name. My heart wants a mother. My imagination wants a plot twist.

Science wants constraints.

So the question gets tightened:

Which candidate woman most plausibly gave birth to Earl Mack Douglas in Duluth in January 1896, given time, place, networks, institutional pipelines, and documentary and genetic constraints?

And separately:

Which paternal family network most plausibly contributed Earl’s paternal DNA, given clustering, triangulation, and geographic plausibility?

That framing keeps it testable. It keeps it falsifiable. It keeps it real.

The next move is never “chase names.” Names are late stage rewards.

The earlier move is chasing systems:

Hospitals. Maternity wards. Charity homes. County relief. Guardianship. Boarding houses. Church baptisms and sponsors. Newspaper notices. Cemetery registers. Undertaker books.

And DNA clusters, when they exist, defining the genetic neighborhood like a map of smoke.

The haystack shrinks. The needle has fewer places to hide.

The quiet twist

Sometimes the problem is not missing evidence.

Sometimes the system produced evidence designed to erase a person while preserving the institution.

Mothers disappear in paperwork because the world did not love them equally. Poor people vanish between enumerations because survival does not leave clean lines. Children get renamed because institutions treat them like inventory. Women get erased because shame was cheaper than support.

And that means the blank isn’t neutral.

It isn’t just “unknown.” It’s pressure made visible.

A clerk’s clean box. A culture’s dirty secret.

In Earl’s case, the blank parent fields are not only missing data.

They are the shape of what it cost to be his mother.

Silence is not always absence. Sometimes it’s design.

Toolkit, Earl specific

This is the template I keep coming back to so the case doesn’t turn into vibes and wishful thinking. Every candidate mother and every candidate father network gets its own card. Same structure. Same pressure. Same light.

Earl Evidence Card

• Claim

• Competing hypotheses

• Sources used (image, derivative, index)

• Chain of custody notes

• Informant

• Date created vs event

• What this discriminates (what it rules out)

• Conflicts

• Resolution

• Falsifiers

• Conclusion strength (possible, likely, highly likely, demonstrated)

Genetic add on

DNA doesn’t care about family stories, but it still has noise, traps, and rules.

• Match set used

• Cluster and why it matters

• Shared DNA and segment usefulness

• Triangulation status

• Common ancestor status and sourcing

• Alternative explanations (endogamy, pedigree collapse, NPE, pileup regions)

• What DNA discriminates

• DNA falsifiers

DNA is trace evidence. It doesn’t solve the case alone. It tells me where to dig and what stories can’t survive.

Final words, Earl edition

Evidence is not the thing I found.

Evidence is the thing I found that survives the test.

Earl Mack Douglas is not a blank box in a database. He is a human being who got processed by a city and a system that kept breathing while it erased the people who made him.

A family tree is not a monument. It’s a set of claims under cross examination. If Earl is going to be named honestly, it won’t be with confidence. It will be with documentation that can survive the light.

Because the truth is not what I want.

The truth is what remains after I try to break it.

Genealogy is a justice system for the dead. Stop sentencing people on hearsay and a shared first name.

And if I do this right, the file won’t end with a name that feels good.

It’ll end with a name that can’t run.

---

---

Receipts

• Board for Certification of Genealogists. Genealogy Standards (Second Edition Revised, published 2019 with additional revisions in 2021). Nashville, TN: Ancestry, 2019. bcgcertification.org+1

• Jones, Thomas W. Mastering Genealogical Proof. Arlington, VA: National Genealogical Society, 2013. National Genealogical Society+1

• Mills, Elizabeth Shown. Evidence Explained: Citing History Sources from Artifacts to Cyberspace. 4th ed. Baltimore: Genealogical Publishing Company, 2024. Genealogical.com+1

• Wisconsin, U.S., State Censuses, 1855–1905. Database with images. Ancestry. Accessed [December 19, 2024. Citing Wisconsin State Census, 1905, State Historical Society of Wisconsin (Madison).

• National Research Council. Strengthening Forensic Science in the United States: A Path Forward. Washington, DC: National Academies Press, 2009.

• Popper, Karl. The Logic of Scientific Discovery. 2nd ed. London and New York: Routledge, 2002.

• Legal Information Institute, Cornell Law School. “Federal Rules of Evidence, as amended to December 1, 2024.” Accessed December 21, 2025. Legal Information Institute+1

• Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993).