Centimorgans in the Dark
A Manifesto for Anyone Trying to Identify Unknown Great-Great-Grandparents
I can lie to myself with a number.
94 cM.
67 cM.
31 cM.
Those numbers look clean. Clinical. Responsible. They look like progress.
They are not progress.
They are pressure.
If I don’t handle that pressure correctly, I will build fiction with math and call it genealogy.
Four generations back is where sentiment dies.
This is where documentation thins out into rumor.
This is where poverty erases paper.
This is where fathers go unnamed and mothers get folded into margins.
This is where a clerk writes what fits the form and moves on.
This is where people start misusing centimorgans.
And this is where I have to decide whether I’m serious.
Four Generations Back Is Physics
I am not dealing with nostalgia. I am dealing with recombination.
Every generation cuts DNA in half. Not poetically. Mechanically.
Chromosomes swap segments. Blocks fracture. Pieces disappear.
By the time DNA passes from parent to grandparent to great-grandparent to great-great-grandparent, it has been reshuffled four times.
What began as a coherent genome becomes fragments.
A great-great-grandparent contributes roughly 6.25 percent of my DNA.
That 6.25 percent does not sit neatly anywhere.
It survives in pieces. Fragments.
Some pieces are large enough to recognize.
Some never make it to me at all.
Third cousins share, on average, around 75 cM.
But averages conceal violence.
Some share 120 cM.
Some share 40 cM.
Some share 15 cM.
Some share nothing.
Nothing.
That’s not malfunction.
That’s inheritance doing exactly what it’s supposed to do.
If I expect clarity at this depth, I’m misunderstanding biology.
Centimorgans don’t tell me who my ancestors are.
They tell me how much survived.
The First Lie I’m Tempted to Tell
“That 92 cM match must be the line.”
It’s the oldest lie in genetic genealogy.
I see 92 cM and I feel relief. Finally. Something solid.
But 92 cM doesn’t say, here is your great-great-grandparent.
It says, here is a corridor.
Inside that corridor are multiple possible relationships. Third cousin. Half second cousin once removed. Second cousin twice removed. Equivalent variations.
Range is not identity.
If I collapse possibility into certainty, I’m not solving a mystery.
I’m narrating one.
Narratives are comforting.
Structure is uncomfortable.
If I want the truth, I have to choose discomfort.
The Cliff Below 20
Then the numbers fall.
19 cM.
14 cM.
11 cM.
8 cM.
It feels like the signal died.
It didn’t die.
It diluted.
By fourth cousin level, about half of genealogical cousins share no detectable autosomal DNA.
The rest often share less than 20 cM.
Below 20 cM, probability shifts.
Small segments aren’t meaningless. They’re fragile. They require reinforcement.
If I treat every 12 cM match as proof, I fabricate ancestry.
If I treat every 12 cM match as noise, I might miss a pattern.
What I have to do is read clusters, not fragments.
At this depth, individuals mislead.
Networks reveal.
Cluster or Illusion
One match is noise.
Three matches who match each other and trace back to the same county in 1840 is a signal.
If I share 102 cM with one person, 76 cM with another, 58 cM with a third, and they all intersect in the same geographic corridor, that’s not coincidence.
That’s inheritance leaving fingerprints.
The question isn’t who is this match.
The question is what community do these matches belong to.
If I chase a person instead of a pattern, I will misplace my ancestor inside the wrong branch and defend it with passion.
Passion is not evidence.
Pattern is.
Segment Size Is the Discipline
Total cM seduces me.
Longest segment disciplines me.
A single 35 cM contiguous segment rarely survives by accident.
Five scattered 7 cM fragments might.
So I have to prioritize:
Longest segments above 25 to 30 cM.
Totals above 50 cM.
Matches who also match each other.
Small fragments don’t build cases.
Large surviving segments carry weight.
This is not romantic work.
It’s structural.
The Sabotage of Endogamy
If my ancestors came from a small island, a closed religious community, or a tightly intermarried immigrant settlement, cM totals inflate.
I may share 120 cM not because I’m close, but because I’m connected multiple ways through distant pathways.
If I don’t test for that, I’ll overestimate proximity.
At GGGP depth, humility isn’t optional.
Overconfidence is how lineage gets fabricated.
DNA Gives Range. Time Narrows It.
If I share 94 cM with someone whose ancestors were born in 1820, that suggests a shared ancestor in GGGP territory.
But suggestion isn’t confirmation.
I have to check chronology.
Are the generations aligned.
Do migration routes intersect.
Is the timing plausible.
DNA says possible.
Time says probable or impossible.
Without time, centimorgans float in abstraction.
Build Forward, Not Backward
This is the part I resist.
I want to name the unknown couple.
I want the reveal.
What I need to do instead is slower and harder.
Identify the ancestral couple shared by the cluster.
Build their descendants forward. Every child. Every branch. Every migration.
Then look for where my unknown ancestor could logically fit.
The DNA defines the community.
The documents define the individual.
If I reverse that order, I invent ancestry and call it discovery.
I don’t get to choose my ancestors.
I only get to eliminate the wrong ones.
The Quiet Part
There’s a moment in this work that no one talks about.
It’s not excitement. It’s not revelation.
It’s the realization that some branches will never yield enough surviving DNA to confirm what happened.
That some truths may be statistically narrowed but never ceremonially named.
That the blank line might not fill in with a clean baptism record and a smiling photograph.
That it might fill in with probability so strong that nothing else survives.
And I have to decide whether that’s enough.
That’s the quiet part.
That’s where ego dies.
Negative Evidence Matters
If documented descendants of a proposed couple don’t match at expected levels, that weakens the hypothesis.
If my strongest matches cluster in one migration corridor and my theory places my ancestors somewhere else, that tension isn’t inconvenient.
It’s directional.
Centimorgans eliminate as effectively as they suggest.
Absence across multiple lines is data.
Discomfort isn’t disproof.
Pattern is.
Probability Versus Certainty
Centimorgans are valid.
But only within boundaries.
They are valid as relationship range estimators.
Cluster identifiers.
Hypothesis filters.
Elimination tools.
They are not absolute proof.
They are not emotional confirmation.
They are not replacements for documentation.
If I blur that line, I become my own unreliable narrator.
The Blank Line Is Not Neutral
When I open Earl’s tree and see that space where parents should be, I feel the weight of it.
No Parents Listed.
Not argued.
Not debated.
Just absent.
Something broke four generations back.
A record wasn’t created.
A father wasn’t named.
A mother was erased.
A migration swallowed context.
A clerk wrote what fit the form and moved on.
Centimorgans are what survived that silence.
Fragments that slipped through recombination and landed in living bodies.
If I misuse those fragments, I extend the silence.
If I inflate them, I create fiction.
If I ignore them, I concede the field.
So this is the standard I hold.
I won’t worship a number.
I won’t mistake probability for proof.
I won’t let emotional hunger outrun structural evidence.
I’ll let centimorgans constrain me.
I’ll let clusters narrow the field.
I’ll let elimination remove the pretty theories first.
Because if Earl’s unknown great-great-grandparents are going to be named, they’ll be named through discipline, not desire.
Final Truth
Four generations back, DNA doesn’t shout.
It whispers.
It leaves fragments.
It leaves patterns.
It leaves probability weighted by biology.
Centimorgans aren’t the answer.
They’re the boundary.
If I follow structure, the field narrows.
If I chase numbers emotionally, the field expands into chaos.
When a name finally replaces that blank line in Earl’s tree, it won’t be because I wanted it to be true.
It’ll be because every other possibility collapsed under pressure.
I’m not here to decorate the fog.
I’m here to narrow it until the truth is the only thing left standing.
More Proof
Receipts
Blaine T. Bettinger, “The Shared cM Project Version 4.0,” The Genetic Genealogist, March 2020, https://thegeneticgenealogist.com/wp-content/uploads/2020/03/Shared-cM-Project-Version-4.pdf.
Jonny Perl, “Shared cM Project 4.0 Tool,” DNA Painter, last modified March 26, 2020, https://dnapainter.com/tools/sharedcmv4.
International Society of Genetic Genealogy (ISOGG), “Autosomal DNA Statistics,” ISOGG Wiki, last modified 2023, https://isogg.org/wiki/Autosomal_DNA_statistics.
Blaine T. Bettinger, The Family Tree Guide to DNA Testing and Genetic Genealogy (Cincinnati: Family Tree Books, 2016).
Roberta Estes, “Concepts: Your Matches on the Same Segment Are Not Necessarily Related to Each Other,” DNA-Explained, December 13, 2022, https://dna-explained.com/2022/12/13/concepts-your-matches-on-the-same-segment-are-not-necessarily-related-to-each-other/.
Jonny Perl, “Shared cM Project 4.0 Tool,” DNA Painter, relationship probability tables and caveats regarding endogamy and pedigree collapse, https://dnapainter.com/tools/sharedcmv4.
Board for Certification of Genealogists, Genealogy Standards, 2nd ed. (Nashville, TN: Ancestry, 2019).
AncestryDNA, “Shared DNA and Matches,” Ancestry Support, accessed February 14, 2026, https://support.ancestry.com.