Do Koreans even exist? part 1. Blood

Do Koreans even exist?

What a silly question.

Of course they do. In fact there are approximately 85 million of them.

• South Korea: ~51,500,000 (60.6%)
• North Korea: ~26,200,000 (30.8%)
• United States: ~2,600,000 (3.1%)
• China: ~2,100,000 (2.5%)
• ROW: ~2,600,000 (3.0%)

In fact, of the 193 member states in the United Nations (plus two non-member observer states, Vatican City and Palestine), according to the Korean Ministry of Foreign Affairs, Koreans exist in 182 of them.

Living. Breathing. Some even flag-waving. 

Real world human beings that are Korean.

In fact, I'm one myself.

Then why would I even question whether Koreans exist?

It's not that I question whether there are people in this world that identify as Korean,
it's that I question whether the term 'Korean' itself is real in the first place.

For that matter, I question whether any category we attach to people, be it race, ethnicity, tribe, or clan, is real.

As we explored in the previous post:

'People' is a completely made up construct.

and,

'Lineage', and 'Legitimacy'' is like drawing a line across a flowing river.

The only two prerequisites of becoming a member of a People are:

"I believe I am."

and,

"Yes, I believe you are."

So in the next series of posts I plan to trace the origins of this fictional concept called, "Koreans."

First we will look at "Korean" DNA (the biological reality).

Then "Korean" language (the cognitive software).

Then finally, "Korean" archaeology and history the material and political reality).

To see if "Koreans" actually exist.

- ה -

1. Korean DNA

I'm going to burst your bubble before we go into hundreds of words of technical jargon.

There is no such thing as Korean DNA.

In fact, there is no such thing as any <fill in People here> DNA.

There are a mixture of different mutations of DNA, and it is that pattern that is commonly associated with a <fill in People here> DNA.

So instead of Korean DNA = X,

you get, Korean populations have a mixture of = X, Y, Z.

The problem is;

'People' A = X, Y, Z

'People' B = A, B, X

'People' C = A, D, Z

So the pattern could be sort of unique to a 'People',

but individually we are one of A, B, D, X, Y, Z. 

So if we take the example above,

I could be tempted to say;

'People' A = Y, since Y only occurs in 'People' A.

'People' B = B, since B only occurs in 'People' B.

'People' C = D, since D only occurs in 'People' C.

But what it Y, B, and D only occurs in <1% of each 'People'?

So, sorry.

There is no such thing as Korean DNA.

- ה -

Short primer: what the hell are genetic markers?

The human genome consists of 23 pairs of chromosomes: one set inherited from the father, one from the mother. Each chromosome is a single DNA molecule built from millions of nucleotide base pairs: Adenine, Cytosine, Guanine, and Thymine (ACGT).

parent to child,

parent to child,

These DNA get handed down through the generations.

But once in a while the DNA makes a copy error.

A 'G' gets replaced with a 'T',

A 'C' gets replaced with an 'A'.

A mutation occurs.

• Most mutations do nothing at all.
• Some are harmful and cause conditions like sickle cell anemia, where one single letter in the gene that makes hemoglobin is swapped.
• Rarely, the typo does something useful. Koreans carry some fun examples: a mutation that makes our armpits smell less (a mutation in the ABCC11 gene), while mostly missing the mutations that let adults keep drinking milk without diarrhea (a mutation in the MCM6 gene).

Fascinating stuff: smell less, fart more.

But for our purposes, we are interested in mutations that act as something else entirely: 

genetic markers.

A genetic marker is simply a recognizable variation in DNA that can be inherited. If a mutation occurred in one man thousands of years ago and his descendants inherited it, that mutation becomes a signpost marking one branch of the human family tree.

So let's trace someone's male ancestry:

XYZ → XYZ → (mutation happens) → XY'Z → XY'Z → (mutation happens) → XY'Z'

that Y' and Z' are genetic markers.

And when enough of these markers accumulate along the same branch, geneticists can group the descendants together into something called a:

haplogroup.

Think of a haplogroup as an address on the human family tree.

For example (the infamous "Korean" gene):

O → O1 → O1b → O1b2

Each step represents another branch created by another inherited mutation.

Everyone in O1b2 shares the mutations that define O1b2, but they also inherited the older mutations defining O1b, O1, and O.

Unfortunately, geneticists occasionally redraw the tree and rename the branches as new information becomes available.

So the exact same lineage might be called one thing in a study from 2005 and something completely different in a study from 2025.

Because apparently this wasn't confusing enough already.

This tracking works on both sides of the aisle.

• Paternal markers ride on the Y-chromosome, making them testable in males.
• Maternal markers ride on mitochondrial DNA, making them testable in everyone, but passed down the family tree exclusively by daughters.

But there is a glaring limitation here.

Your father's father's father's line is not your complete ancestry. 

Neither is your mother's mother's mother's line.

Those are merely two threads on the outermost edges of a massive tapestry. 

You have thousands of other ancestors who contributed to who you are, completely invisible to those two tests.

And this creates an extremely important distinction.

If 40% of Korean men carry a particular Y-chromosome haplogroup, that does not mean:

Koreans are 40% that haplogroup.

It means:

40% of the Korean men sampled have one direct paternal line that belongs to that haplogroup.

The same man could carry one paternal marker, a completely different maternal marker, while the overwhelming majority of his ancestry came through thousands of other people invisible to both.

So:

Haplogroup frequency ≠ ancestry percentage.

To see the rest, geneticists have to look at the rest of the code.

Autosomal DNA.

Remember those 23 pairs of chromosomes we mentioned at the beginning? The sex chromosomes (X and Y) make up just one pair. The other 22 pairs are your autosomes. They hold the vast bulk of your genetic code.

But unlike the clean, straight-line inheritance of Y-DNA and mitochondrial DNA, autosomal DNA does something delightfully messy.

Before you pass your DNA to a child, your body takes the chromosomes you got from your mother and the chromosomes you got from your father, and it physically swaps chunks of them. This process is called recombination.

It’s like taking two decks of cards; one red, one blue, cutting them, mashing them together, and handing half that new, chaotic deck to your kid.

This means you get exactly 50% of your autosomal DNA from your father and 50% from your mother. But when you look at your four grandparents, the math gets wobbly. You don't inherit exactly 25% from each of them. Because of the random shuffle, you might get 28% from your maternal grandmother and only 22% from your maternal grandfather.

Go back another generation, and the shuffle happens again. And again.

Because it keeps getting sliced in half and mixed together, by the time you get 6 or 7 generations back, the chunks of DNA from specific ancestors have been diced so many times that some of them get completely lost in the shuffle. You could have a legitimate, documented great-great-great-great-grandfather whose genetic contribution to you is mathematically zero.

DNA Type	What it traces	The Catch
Y-DNA	Direct paternal line only	Ignores almost your entire family tree
mtDNA	Direct maternal line only	Ignores almost your entire family tree
Autosomal DNA	The enormous mixture of recent ancestry	Individual ancestral signals fade as you go backward

So, if you want to know what region of the world your father's father's father came from 10,000 years ago, you look at the Y-chromosome.

But if you want to find your second cousin, or figure out what percentage of your recent ancestry is from a specific region over the last few hundred years, you look at the autosomal DNA. It holds the messy, blended record of the collective group of people who actually made you.

- ה -

The genetic markers found in Koreans

Before I show you the general genetic makeup of Koreans, I need to tell you something.

Korea eats its dead

Quite literally. Korean soils are often acidic, and acidic burial environments are bad for bone and DNA preservation. Researchers studying ancient Korean genomes specifically identify Korea's acidic soils as one reason we have comparatively few ancient Korean genomes.

Egypt gives us mummies.

Siberia gives us frozen corpses.

Europe gives us mountains of ancient DNA.

Korea gives geneticists: "Good luck."

We can test the DNA of living Koreans.

But if they've been buried...mostly untestable rubbish.

People have been living in and around the Korean peninsula for ages, and have consistently buried their bodies, but what's left of them reveal very little DNA that is testable.

Testable DNA is only found under extraordinary circumstances: a limestone cave for instance. Unfortunately, Koreans have been burying their dead in the ground for centuries.

Y-DNA

Since Korea eats its dead, let's start with the easiest evidence we have:

living Korean men.

The Y-chromosome is almost absurdly convenient for tracing ancestry.

A father gives his Y-chromosome to his son.

Who gives it to his son.

Who gives it to his son.

Unlike autosomal DNA, it does not get chopped up and shuffled with the mother's DNA every generation. For our purposes, it travels down one paternal line largely intact, collecting the occasional mutation along the way.

So if we collect the Y-chromosomes of a bunch of modern Korean men and sort them by those mutations, distinct family branches start appearing.

• O2-M122: 44.27%
• O1b2-M176: 31.42%
• C2-M217: 12.25%
• N-M231: 4.55%
• O1a-M119: 2.17%
• D1a2a-M55: 1.58%
• Q-M242: 1.38%
• R-M207: 0.4%
• C1a1-M8: 0.2%
• O1b1-M95: 0%
• Other: 1.78%

Three enormous branches immediately jump out:

1. O2-M122: 44.27%
2. O1b2-M176: 31.42%
3. C2-M217: 12.25%

Together, those three account for roughly 88% of Korean paternal lineages.

Add N, at around 5%, and four branches account for more than 92%.

At first glance, this looks promising.

Maybe we've finally found:

Korean DNA.

Not even close.

• O2 is found throughout East Asia and is particularly common among Han (漢) Chinese (formed 25,000 to 30,000 years ago).
• O1b2 is found most heavily among Koreans and Japanese (formed 5,000 to 28,000 years ago).
• C2 stretches across enormous parts of northern Eurasia and is particularly common among Mongolian, Tungusic, and Siberian populations (formed 40,000 to 45,000 years ago).
• N reaches from East Asia across Siberia and eventually all the way into northern Europe (formed 35,000 to 40,000 years ago).

And the smaller Korean lineages make things even messier.

So what looks like one neat "Korean" pattern is actually a particular mixture of ancient paternal lines whose histories began thousands, and sometimes tens of thousands, of years before anyone had invented a thing called:

Korean.

So let's follow each of those branches backward.

O2-M122: ~44.27%

Let's start with the big one.

Nearly half of Korean men carry O2-M122 somewhere along their direct paternal line.

And if you look immediately to the west, you might think you've found the answer.

O2-M122 is even more common among Han Chinese. One large East Asian study found it in roughly 52% of northern Han and 54% of southern Han men. It is also common among Tibetans, Vietnamese, Hmong-Mien populations, and numerous other peoples throughout East and Southeast Asia.

So:

O2 = Chinese?

Nope. The lineage is far too old.

A classic phylogeographic study estimated major O2-M122 branches and their northward expansion at roughly 25,000 to 30,000 years ago.

For context:

• Agriculture hadn't been invented.
• China hadn't been invented.
• The Han (漢) hadn't been invented.
• Korea (韓) hadn't been invented.
• There wasn't even a Yellow River civilization to migrate from.
• There were just groups of hunter-gatherers moving around East Asia.

One of those men carried a mutation called O2-M122.

• He had sons.
• They had sons.
• Their descendants multiplied and spread throughout East Asia.
• Much, much later, some of those descendants became:
• Han Chinese (漢).
• Koreans (韓).
• Japanese.
• Tibetans.
• Vietnamese.
• And countless other 'Peoples'.

So when we say:

44.27% of Korean paternal lineages are O2-M122

we absolutely do not mean:

Koreans are 44.27% Chinese.

The lineage came first.

The Chinese Duck came tens of thousands of years later.

O1b2-M176: ~31.42%

This one gets much more interesting.

O1b2-M176 is found at particularly high frequencies in Koreans and Japanese.

It exists elsewhere, but Korea and Japan are where it really stands out.

Finally. Maybe this one is:

Korean DNA?

Still nope.

But this time we're getting closer to something historically useful.

Studies of Korean and Japanese O1b2 lineages have found evidence of major expansions roughly 6,000 to 10,000 years ago, with particularly high diversity among Korean samples suggesting an important prehistoric expansion somewhere in or around the Korean Peninsula.

That places us in the Neolithic.

• Thousands of years before Gojoseon.
• Thousands of years before Japan.
• Thousands of years before anyone called themselves Korean or Japanese.

And we now have something even better than modern DNA.

Ancient DNA.

O1b2 has been found directly in Yayoi-period remains from Japan, and recent genome studies increasingly connect much of the continental ancestry entering Japan during the Yayoi and Kofun periods with populations coming through the Korean Peninsula.

So O1b2 may eventually become one of our most useful paternal clues for understanding the prehistoric relationship between populations on the Korean Peninsula and the Japanese archipelago.

But it still doesn't belong to either modern People.

O1b2 ≠ Korean.

O1b2 ≠ Japanese.

It belonged to people who lived before either category existed.

Some of their descendants eventually became Koreans.

Some became Japanese.

And presumably plenty became something else entirely.

C2-M217: ~12.25%

Now we head north.

C2-M217 is widespread across northern Eurasia and is especially prominent among many:

• Mongolian,
• Tungusic,
• Siberian,
• and Central Asian populations.

So this one is extremely tempting.

Korea was invaded by northern peoples over and over again.

• Khitan.
• Jurchen.
• Mongols.
• Manchus.

And about 12% of Korean men carry C2.

Case closed?

12% Mongolian?

No.

Not remotely.

C2 is ancient.

The earliest currently known ancient individual carrying C2-M217 dates to roughly 19,500 years ago in the Amur River basin, and major C2 branches were already differentiating across northern Eurasia thousands of years before anything resembling a Mongol, Jurchen, Manchu, or Korean existed.

That doesn't mean later invasions contributed nothing.

They almost certainly moved actual human beings around, and actual human beings occasionally have children.

But you cannot look at a Korean man's C2 chromosome and announce:

Aha! Genghis Khan.

His paternal line may have been sitting somewhere around Manchuria, the Amur, Siberia, Mongolia, northern China, or the Korean Peninsula thousands upon thousands of years before Genghis Khan's parents had the decency to conceive him.

C2 tells us:

northern Eurasian paternal ancestry.

It does not tell us:

Mongol.

Again. The marker came first. The Duck came later.

N-M231: ~4.55%

And then there is N.

Only around 5% of Korean paternal lineages belong to N-M231.

Small enough to ignore?

Not if you're trying to understand where the spaghetti goes.

Because N does something spectacular.

It keeps going.

• North through East Asia.
• Into Siberia.
• Across northern Eurasia.
• And eventually all the way into Finland.

N-derived paternal lineages are extraordinarily important among several Uralic-speaking populations of northern Europe, while branches of the same enormous family are also found among Siberians, northern Chinese populations, Mongolic and Tungusic populations, Koreans, Japanese, and others.

One major phylogeographic study reconstructed an early northward dispersal beginning in East Asia around 21,000 years ago, reaching Siberia roughly 12,000 to 14,000 years ago, followed by later movements westward toward northern Europe.

Which gives us something wonderfully stupid.

A Korean man carrying N-M231 and a Finnish man carrying N-M231 share a paternal branch.

Therefore:

"Koreans are Finnish?"

Obviously not.

"Finns are Korean?"

Also no.

Go backward far enough and you simply reach some prehistoric guy in Eurasia whose descendants wandered in radically different directions.

One branch eventually ended up speaking Korean.

Another ended up speaking Finnish.

Neither of them inherited a little instruction embedded in the Y-chromosome saying what 'People' they were supposed to become.

And we've only looked at four branches.

But already the problem should be obvious.

• The largest Korean paternal lineage points deep into East Asia.
• The second connects Korea intimately with prehistoric populations that also contributed heavily to Japan.
• The third extends deep into northern Eurasia.
• The fourth eventually stretches all the way from East Asia to northern Europe.

All four exist in other 'Peoples'.

All four predate the Peoples carrying them.

And somehow, thousands of years later, particular descendants of all four ended up standing on the same peninsula saying: "We are Korean."

That's not Korean DNA.

That's human history.

///

A short not on the origins of O1b2:

O (M175): ~35,000–40,000 years ago / East or Southeast Asia

│

├── O1: ~30,000–34,000 years ago / Southern China or Southeast Asia

│    │

│    ├── O1a (M119): ~25,000–30,000 years ago / Southern Coastal China

│    │      (Later common among indigenous Taiwanese and Austronesian-speaking peoples)

│    │

│    └── O1b (M268): ~28,000–30,000 years ago / Southern China or Southeast Asia

│         │

│         ├── O1b1 (M95): ~15,000–20,000 years ago / Southeast Asia or South China

│         │      (Later strongly associated with Austroasiatic-speaking populations)

│         │

│         └── O1b2 (M176 / SRY465): ~25,000–30,000 years ago / East Asia

│                  (Later became especially common in Korea and Japan)

│

└── O2 (M122): ~30,000–35,000 years ago / East Asia

         (Later became the dominant paternal lineage across much of China)

Then where does O come from?

Haplogroup NO (M214), which originated roughly 40,000 to 45,000 years ago in Eurasia, shortly after early humans migrated out of Africa. Haplogroup NO was carried by a population of Ice Age hunter-gatherers. Around 40,000 years ago, this group experienced a massive split, resulting in two distinct lineages that took wildly different paths:

• Haplogroup O (M175): The Asian Branch. As we've covered, this group stayed in East and Southeast Asia, eventually developing agriculture and booming into the massive populations of China, Japan, Korea, and Polynesia.
• Haplogroup N (M231): The Siberian/Uralic Branch. The brothers of Haplogroup O headed strictly north into the freezing climates of Siberia. Over tens of thousands of years, they migrated westward across the Arctic Circle. Today, Haplogroup N is the dominant genetic marker of the Finnish people (up to 60%), Estonians, Yakuts, and various Siberian and Uralic tribes.

NO traces its lineage to:

↑

K2: ~40,000-45,000 years ago / Southeast Asia,

↑

K: ~45,000-50,000 years ago / South Asia, Southeast Asia, or the Middle East,

↑

CT: ~65,000-70,000 years ago / East Africa, and the first male to leave Africa (leave DNA trace),

↑

BT: ~70,000-80,000 years ago / somewhere in East Africa,

↑

Adam (A0): ~230,000-300,000 years ago / somewhere in West or Central Africa.

Hey Hitler,

meet your grandfather.

///

mtDNA

The maternal side works almost exactly the same way.

Except instead of following:

father → son → son → son,

we follow:

mother → daughter → daughter → daughter.

• Your mitochondrial DNA comes from your mother.
• She got hers from her mother.
• Who got hers from her mother.

Men carry mtDNA too, but they don't normally pass it on.

So if Y-DNA gives us one thin paternal thread through history, mtDNA gives us one thin maternal thread.

And when we collect the mitochondrial DNA of a bunch of Koreans, another set of ancient family branches appears.

• D: 32.4%
• B: 14.6%
• M7: 10.3%
• G: 10.3%
• F: 9.7%
• A: 8.1%
• Other: ~14.6%

The single largest branch was D4, a subclade of D, at roughly 23.8% by itself.

Already this looks very different from the father's side.

With Y-DNA, three branches accounted for almost 88% of Korean paternal lines.

The mother's side is considerably messier.

No single maternal lineage dominates.

And something else immediately jumps out.

Some of these lineages are most common toward the north.

Others are most common toward the south.

One study grouped roughly 60% of Korean maternal lineages with haplogroups especially common in northern East Asia and southeastern Siberia, while more than 30% belonged to lineages more strongly associated with southern East Asia.

So apparently Korean mothers didn't get the memo about being a homogeneous People either.

Let's follow the biggest branches backward.

D: ~32.4%

This is the monster.

Nearly one-third of Korean maternal lines belong somewhere within haplogroup D.

Most of those belong to D4, which alone accounts for roughly 23.8% of Korean mtDNA in the study we're using. D4 is widespread across northern East Asia and appears at high frequencies among Koreans, Manchurians, northern Chinese populations, Mongolians, Siberian populations, Japanese, and others.

So:

D = Korean?

Obviously not.

D = Manchurian?

Nope.

D = Siberian?

Still no.

The branch is ancient.

Modern estimates place the major expansion of D4 somewhere around 20,000 to 30,000 years ago, depending on the method used.

Once again:

• No Korea.
• No China.
• No Manchuria.
• No Mongols.
• No Japanese.

Just human beings surviving the last Ice Age somewhere in eastern Eurasia.

One woman carried the mutations that eventually defined D4.

• She had daughters.
• They had daughters.
• Their descendants spread.
• Much later, some of those descendants became:
• Korean.
• Manchu.
• Chinese.
• Japanese.
• Siberian.
• And plenty of other things.

The marker came first.

The Ducks came later.

B: ~14.6%

Now we head south.

Haplogroup B is ancient and widespread across southern China, Southeast Asia, Taiwan, the Pacific, and parts of East Asia.

It is also found among Koreans at a surprisingly substantial frequency: 

roughly 15%.

Unlike D, which has a particularly strong northern distribution, B is much more strongly associated with the southern side of East Asian population history.

And B is old.

Very old.

Major B lineages were already diversifying tens of thousands of years ago, long before rice farming, China, Korea, Taiwan, Austronesians, or anything else we might be tempted to attach to them. Some estimates put the deeper B family at roughly 40,000 to 50,000 years old.

So:

B = Southeast Asian?

Not really.

That's where many of its descendants became common.

It isn't what the original woman was.

And B does something particularly fun.

Branches of the same enormous maternal family eventually turn up across:

• southern China,
• Southeast Asia,
• Korea,
• Japan,
• the Pacific,
• and even among Indigenous peoples of the Americas.

Not because Koreans are Polynesian.

Or Native American.

Or Vietnamese.

But because if you follow your mother far enough backward, modern People boundaries start becoming completely useless.

M7: ~10.3%

M7 is one of my favorites.

Because unlike D, which spreads heavily across northern East Asia, M7 has much stronger roots toward southern and coastal East Asia.

Today it appears across:

• China,
• Vietnam,
• Korea,
• Japan,
• Southeast Asia,
• and a scattering of populations farther north.

The deeper M7 lineage may be roughly 35,000 to 50,000 years old, depending on which branch and dating method you use.

But the really interesting part is what happens when M7 starts branching.

• M7b becomes particularly important in China.
• M7c stretches deeply into southern and Southeast Asian populations.

And:

• M7a becomes strongly associated with Korea and Japan.
• Some M7a-related branches are found overwhelmingly in the Korean Peninsula and Japanese archipelago.

Aha! Finally.

Korean DNA?

You know the answer by now.

M7a was already branching thousands of years before Korean and Japanese were meaningful identities.

What it gives us is not a Korean marker.

It gives us a clue that prehistoric populations on the Korean Peninsula and Japanese archipelago were connected long before either society started drawing modern circles around themselves.

G: ~10.3%

Then we swing north again.

Haplogroup G is especially associated with northeastern Asia and southeastern Siberia.

It appears among:

• Koreans,
• Manchurians,
• Mongolians,
• Siberian peoples,
• northern Chinese populations,
• and populations extending westward toward Central Asia.

Some major G branches appear to have differentiated roughly 15,000 to 30,000 years ago, with individual subbranches sometimes considerably older or younger.

So again:

G = Siberian?

No.

It is a maternal branch whose descendants became common among several populations we now associate with northern Asia.

• Some of them eventually walked south.
• Some stayed north.
• Some ended up Korean.

F: ~9.7%

Then south again.

Haplogroup F is especially common throughout southern China and Southeast Asia.

Its broader ancestry is ancient, with the deeper F lineage stretching back tens of thousands of years. Important East Asian subbranches underwent major expansions well before the Neolithic.

And yet nearly one in ten Korean maternal lines belongs to F.

So standing next to: D. G. A. 

all pointing heavily toward northern East Asia,

we have: B. M7. F.

with much stronger southern distributions.

North. South. North. South.

Apparently the Korean maternal family tree has no interest whatsoever in cooperating with a simple origin story.

A: ~8.1%

And finally, A.

Haplogroup A is widespread across northeastern Asia and Siberia.

It appears among Koreans, northern Chinese populations, Mongolians, Siberian peoples, Japanese, and numerous Indigenous populations of the Americas.

Major East Asian A lineages were already expanding somewhere around 15,000 to 25,000 years ago.

And once again we can play the stupid game:

Korean woman with haplogroup A.

Native American woman with a distant branch of haplogroup A.

Therefore:

Koreans are Native American?

No.

Native Americans are Korean?

No.

Their maternal lines simply meet somewhere so far back in northeastern Eurasia that neither category means anything anymore.

• At some point one branch stayed in Asia.
• Another eventually crossed into the Americas.
• Thousands of generations later, their descendants invented entirely different Peoples.

So let's stop for a second.

The paternal side gave us:

O2. O1b2. C2. N.

The maternal side gives us:

D. B. M7. G. F. A.

And those maternal lines don't even point consistently in the same direction.

• Some have strong northern distributions.
• Some have strong southern distributions.
• Some connect particularly closely with Japan.
• Some stretch into Siberia.
• Some eventually stretch into Southeast Asia and the Pacific.
• Some have relatives whose descendants crossed into the Americas.

One study estimated the Korean maternal pool at roughly 60% northern-associated lineages and more than 30% southern-associated lineages. It also found Koreans clustering genetically with Manchurians, Japanese, Mongolians, and northern Han Chinese when maternal markers were compared.

And here's where things get really interesting.

The paternal and maternal stories are not identical.

The same study found the maternal Korean population looking predominantly northern, while the Y-chromosome showed a substantially stronger southern contribution. The authors even estimated a larger southern contribution through paternal than maternal lineages.

Meaning:

the men tell one version of the story.

the women tell another.

And neither tells us:

Korean DNA.

They tell us that different human populations moved into, out of, and around the Korean Peninsula at different times, through different routes, under different circumstances.

Then their descendants mixed. And thousands of years later, we gave the mixture a name:

Korean.

///

The origins of D4:

D4: ~25,000–27,000 years ago / northern China, Korea, and Japan

↑

D: ~35,000–48,000 years ago / Central Asia, Siberia, and even Native Americans

↑

M: ~60,000 years ago / The Middle East and South Asia

↑

L3: ~65,000–70,000 years ago / East Africa, and the first female to leave Africa (leave DNA trace)

↑

Eve (L0): ~150,000–200,000 years ago / somewhere in East Africa or Southern Africa

Hey Hitler,

meet your grandmother:

Surprisingly, Mitochondrial Eve seems to have lived roughly in the same place as Y-Chromosomal Adam. Today both A0 and L0 are seen most commonly among the Khoisan (San) Peoples of Sub-Saharan Africa, with some tribes reaching frequencies of 70-100%.

///

Autosomal DNA

Now we stop following two lonely threads.

• Y-DNA follows one father-to-father line.
• mtDNA follows one mother-to-mother line.
• Autosomal DNA looks at the gigantic scrambled mess in between.

Instead of asking:

"Which paternal or maternal branch do you belong to?"

we can now ask:

"When I compare hundreds of thousands of mutations across your genome, which other people do you most resemble?"

This is where population genetics starts looking less like a family tree and more like a map.

• Take a few hundred Koreans.
• Take a few hundred Japanese.
• Take northern Han Chinese.
• Southern Han Chinese.
• Mongolians.
• Tibetans.
• Vietnamese.

Then compare hundreds of thousands of points across their autosomal DNA.

Nobody's DNA contains a little tag saying:

Korean.

Instead, geneticists measure how similar the patterns are.

One common technique is called Principal Component Analysis, or PCA.

The math is not important.

What matters is the output.

You turn people into dots.

People whose DNA patterns are more similar appear closer together.

People whose patterns are less similar appear farther apart.

And when you do that with East Asians, something interesting happens.

• Koreans form a cluster.
• Japanese form another cluster.
• Han Chinese form another.
• Mongolians another.
• Zoom out to include Europeans and Africans and these East Asian populations sit extremely close together.
• Zoom back in far enough and the differences become visible again. 

A 2018 genome-wide study using 100 Koreans and hundreds of thousands of autosomal markers could distinguish Korean, Japanese, and Han Chinese clusters, with their genetic relationships broadly following geography.

So...

Korean DNA?

Still no.

What we've found is:

a Korean-shaped cluster.

A population of people who, because they have been reproducing largely with one another for a long time, tend to share slightly more genetic similarity with one another than with the people next door.

That is very different from finding some biological substance called:

Korean.

And autosomal DNA lets us do something even more interesting.

We can compare modern Koreans not just against living people, but against dead people.

Or at least the few dead people whose DNA Korea didn't eat.

Recent ancient-DNA studies repeatedly place a major part of Korean ancestry around prehistoric populations living in Northeast China and the West Liao River region.

One 2023 study found that modern Koreans could actually be modeled using Bronze Age West Liao River farmers as a single ancestral source.

Use a more complicated two-source model, and the same Koreans could be modeled as roughly:

85% West Liao River Bronze Age-related

15% Taiwan Hanben-related

Now this looks promising again.

85% West Liao River!

Liaoning!

We've found them!

The Proto-Koreans!

Nope.

Because, of course, the West Liao River people were already mixtures too.

A recent ancient-genome model estimated those Bronze Age West Liao River populations themselves at roughly:

• 48% Yellow River-related ancestry
• 52% Ancient Northeast Asian-related ancestry.

So our neat:

Korean

↓

becomes:

West Liao River

↓

which becomes:

Yellow River + Ancient Northeast Asian

↓

and then those populations themselves have

older ancestors,

↓

who had

older ancestors,

↓

who had

older ancestors...

↓

And we end up in

Africa.

And there is another important warning. 85% / 15% is not the equivalent of:

"Your DNA is objectively 85% this and 15% that."

Autosomal ancestry models depend upon which ancient populations survive, which samples geneticists possess, and which populations they choose as reference points.

Change the reference populations, and the percentages can change.

In fact, the same 2023 study found that Koreans could also be statistically modeled using the West Liao River population alone. No Korean DNA needed.

That's not a contradiction.

It means these are models.

The geneticist is basically asking:

"Given the dead people whose DNA I happen to have, which combination best reproduces the DNA of the living people standing in front of me?"

The dead don't come labeled either.

A skeleton does not say:

Hello. I am 52% Ancient Northeast Asian.

We give it that description because its DNA resembles other ancient samples that we have grouped together.

And those labels become increasingly ridiculous when projected backward as modern Peoples.

• The Yellow River-related ancestors weren't Han Chinese.
• The West Liao River ancestors weren't Korean.
• The Ancient Northeast Asians weren't Manchu.
• The Taiwan Hanben-related ancestors weren't Taiwanese citizens.

They were populations of human beings who existed at particular times and places.

Their descendants moved.

Mixed.

Split.

Moved again.

And eventually some particular mixture became sufficiently isolated and internally connected that, when we plot its descendants on a genetic chart today, they form a recognizable little cloud of dots.

We label that cloud: 

Koreans.

So Y-DNA gave us a dozen ancient paternal branches.

mtDNA gave us a completely different collection of ancient maternal branches.

And now autosomal DNA gives us the giant mixture of everybody between them.

All three tell slightly different stories.

None gives us:

Korean DNA.

What they give us is something considerably more interesting:

the genetic history of the people who eventually became Korean.

- ה -

This is the best (and only) map I could find showing the movements of Y-chromosomal DNA. The terminology used is a little antiquated, but the names change as the phylogeny was revised, so it is mostly accurate.

• (old) C3 → (now) C2-M217
• (old) O3 → (now) O2-M122
• (old) O2b → (now) O1b2-M176
• (old) O → (now) O-M175
• (old) N → (now) N-M231
• (old) D → (now) D-M174

What the map does show is how different haplogroups started in Africa, broke into different child haplogroups, and finally ended up near what is now Korea. A broad paint stroke, not a detailed diagram.

The DNA make-up of modern People in East Asia looks roughly like this:

Koreans

O2 ~43%

O1b2 ~32%

C2 ~15%

N ~4%

Other ~6%

Japanese

D1a2a ~32%

O1b2 ~32%

O2 ~20%

C2 ~6%

C1a1 ~5%

Other ~5%

Manchus

O2 ~40%

C2 ~20%

O1b2 ~15–30%

N ~5–15%

Other ~5%

Tungusic:

C2 ~50–65%

N ~20–40%

O and others generally much lower

but enormous regional variation

Mongols
C2 ~55–60%
O2 ~17–20%
N ~5–10%
R1a ~5–6%
Q ~3%
O1b2 ~0–1%
Other ~10%

Han Chinese

O2 ~53%

O1a ~12%

O1b1/other O1b ~10%

C2 ~8%

N ~6%

Other ~11%

Looking at this together, a few haplogroups stand out:

O2
Han ~53% > Korean ~43% ≈ Manchu ~40% > Japanese ~20% > Mongol ~17% > Tungusic low

Age of Mutation: ~30,000–35,000 years ago (Yellow River Basin, China)

* Starting around 400 BCE and continuing for centuries. O2 is the signature of the Sino-Tibetan expansion. The arrival of O2 in Korea corresponds with the eastward expansion of the Yan State, the introduction of the Jeomtodae (rolled-rim) pottery, the establishment of the Han Dynasty commanderies (like Lelang), and later migrations during the Three Kingdoms period.

O1b2
Korean ~32% ≈ Japanese ~32% > Manchu variable > Han ~1% > Mongol ~0% ≈ Tungusic negligible
Age of Mutation: ~25,000–30,000 years ago (Southern Manchuria)

* ~1500 BCE. While the mutation is ancient, the population remained relatively small until they adopted millet and rice agriculture. They expanded rapidly southward into the Korean Peninsula, establishing the Mumun culture.

C2
Tungusic ~65% ≈ Mongol ~58% >> Manchu ~20% > Korean ~15% > Han ~8% > Japanese ~6%
Age of Mutation: ~35,000–40,000 years ago (Central/Northeast Asia)

* This is the oldest continuous genetic layer in Korea. These populations arrived during the Late Paleolithic and were the primary inhabitants during the Chulmun pottery era.

N
Tungusic ~40% > Manchu up to ~15% > Mongol ~5–10% ≈ Han ~6% > Korean ~4% > Japanese ~1%
Age of Mutation: ~35,000–40,000 years ago (East Asia / Siberia)

* Arrived in Korea roughly alongside C2 during the Paleolithic.

D1a2a and C1a1
Japanese ~32% / ~5% respectively > everybody else essentially negligible
Age of Mutation: ~35,000–40,000 years ago (Japan)

* During the Last Glacial Maximum, the ancestors carrying D1a2a and C1a1 undoubtedly walked through the Korean Peninsula to cross the land bridges into Japan. However, after the ice melted and sea levels rose (~10,000 BCE), Japan was isolated. The D1a2a and C1a1 populations trapped in Japan became the Jomon people and thrived in isolation.

Meanwhile, the D1a2a and C1a1 populations left behind on the Asian mainland and the Korean Peninsula were completely wiped out, assimilated, or genetically overwritten by the massive Bronze Age (O1b2) and Iron Age (O2) migrations. They survived only in the protective isolation of the Japanese archipelago.

At a very high altitude, we're beginning to see a few different (very rough) axes:

Continental East Asia: O2

Korea–Japan corridor: O1b2

Inner/Northern Asia: C2

Northern Eurasian/Siberian world: N-M231

Japanese archipelago: D1a2a + C1a1

So based on all this a rough timeline might look like this:

~40,000–20,000 BP: Paleolithic Northeast Asians

People are already living on the Korean Peninsula and throughout Manchuria, northern China, Siberia, and the Japanese archipelago.

This is during the Ice Age. Sea levels are far lower than today and, during the Last Glacial Maximum (~c. 26,500-19,000 BCE), much of the Yellow and Bohai Seas is exposed as dry land. What we now call the Korean Peninsula is therefore not really a peninsula at all, but the eastern edge of a much larger continental landscape.

Japan is also much closer to the continent. Hokkaido is connected northward through Sakhalin, while the Korea Strait is much narrower than it is today.

At this depth, there are no Koreans, Japanese, Manchus, Mongols, Koreanic speakers, or Japonic speakers. There are simply different populations of Northeast Asian hunter-gatherers moving through the landscape, hunting animals, gathering plants, making stone tools, and occasionally leaving behind traces from which we can reconstruct fragments of their lives.

We do not know what languages they spoke.

DNA:

• C2 is already part of the northern Northeast Asian world. A C2-M217 male is directly attested in the Amur region by around 19,500 years ago.
• The ancestral branches that would eventually produce O2, O1b2, and N already exist somewhere in eastern Eurasia.
• The paternal ancestors of the later Japanese D1a2a and C1a1 populations are also somewhere in this broader East Asian world.
• We cannot yet say which of these lineages were actually present on the Korean Peninsula, or in what proportions.

~20,000–10,000 BP: the Ice Age ends and the map begins to change

The Last Glacial Maximum passes. Temperatures rise. Glaciers melt. Sea levels begin climbing.

The enormous plain between China and Korea slowly disappears beneath the Yellow and Bohai Seas. The Korea Strait widens. Japan becomes increasingly separated from the continent.

That matters because populations that had once occupied a much more connected landscape now begin following increasingly different histories.

The ancestors of the Jōmon become progressively isolated within the Japanese archipelago. Korea does not undergo the same kind of isolation. Even as water closes around three sides of it, the peninsula remains open to Liaodong, Manchuria, and continental Northeast Asia through the north.

This may be the beginning of one of the deepest demographic differences between later Korea and Japan (haplogroup D).

Japan increasingly becomes an island world.

Korea remains a continental crossroads.

DNA:

• C2 remains established across northern Northeast Asia.
• N is diversifying across Siberia and northern Eurasia.
• O2 and O1b2 already exist, although we do not know where their surviving branches were geographically concentrated.
• The ancestry that will later characterize the Jōmon becomes increasingly isolated in the Japanese archipelago.

~10,000–5000 BCE: Early Holocene hunter-gatherers

By the beginning of the Holocene, the world is considerably warmer and increasingly recognizable.

Forests spread. Coastlines approach their modern positions. Newly created bays, tidal flats, estuaries, rivers, and shallow coastal waters become extraordinarily productive environments for human settlement.

Different regional lifeways become increasingly visible.

In Japan, Jōmon populations develop highly successful hunting, fishing, gathering, and increasingly sedentary communities. Pottery appears remarkably early, eventually giving us the cord-marked vessels from which the name Jōmon derives.

Similar changes occur on the Korean Peninsula. Coastal and riverine communities exploit shellfish, fish, mammals, nuts, and plants, leaving behind shell middens, settlements, and some of the earliest pottery traditions in Korea. Over time these traditions develop into what archaeologists broadly call the Chulmun or Jeulmun period.

Meanwhile, something very different is beginning on the mainland.

In northern China, people increasingly cultivate millet. In the Yangtze basin and farther south, other populations begin cultivating rice.

For the first time, Korea sits between three increasingly distinct demographic worlds: Jōmon populations to the east, hunter-gatherer and fishing communities on the peninsula itself, and growing agricultural societies to the west.

There is still no defensible way to call any of their languages Koreanic or Japonic.

DNA:

• Japanese Jōmon populations become strongly associated with the paternal lineages D1a2a and C1a1.
• The Korean Peninsula probably contains genetically varied hunter-gatherer populations with connections both northward into Northeast Asia and across the Korea Strait.
• Mainland populations carrying O2, O1b2, C2, N, and other lineages continue to differentiate and expand.
• O1b2 undergoes a major Holocene expansion during roughly this broad period, eventually producing its extraordinary concentration in Korea and Japan. Where that expansion occurred remains uncertain.

~5000–1500 BCE: Korea begins becoming a mixture we can recognize

The agricultural revolution does not stop at the Yellow Sea.

Millet cultivation spreads eastward through northern China and Liaodong and eventually onto the Korean Peninsula. New crops arrive together with new technologies, settlement patterns, and probably new people.

But the people already living in Korea do not simply disappear.

The Chulmun world changes gradually. Hunting, fishing, gathering, and cultivation coexist in different proportions from place to place. Communities interact with Liaodong, the West Liao River region, the Chinese mainland, the Russian Far East, and the Japanese archipelago.

Ancient DNA gives us our first tantalizing glimpse of just how messy this world may have been. Neolithic individuals from Korea are not genetically uniform. Some show strong continental Northeast Asian affinities. Others, particularly along the southern coast and islands, show substantial Jōmon-related ancestry.

The Korean Peninsula is already beginning to look less like the home of one ancient population and more like a meeting place between several.

Toward the end of this long period, the Chulmun world gives way to the Mumun. Agriculture becomes increasingly important, settlements become larger, social organization changes, and connections with Liaodong and the continent intensify.

We still cannot confidently identify the languages being spoken. Some models place ancestral Koreanic on or near the peninsula by this point. Others do not. Japonic may also have been present somewhere in this broader region.

For now, the safest answer remains: we don't know.

DNA:

• Ancient Korean genomes show that Neolithic populations on the peninsula were already genetically heterogeneous.
• Some individuals show strong mainland Northeast Asian / West Liao-related ancestry.
• Some southern coastal individuals show substantial, and occasionally very high, Jōmon-related ancestry.
• Continental farming expansions probably increase ancestry associated with northern East Asian populations, including populations rich in O2 paternal lines.
• O1b2 is expanding somewhere in this broader Northeast Asian world and will eventually become extraordinarily common in both Korea and Japan.
• C2 and N remain part of the northern population landscape, although we cannot yet assign meaningful Korean frequencies to them.

The later Korean population is beginning to make intuitive sense:

older peninsular populations + northern Northeast Asian populations + continental agricultural populations + whatever population history produced the enormous Korean concentration of O1b2.

~1500–300 BCE: Bronze Age reshuffling

With the Mumun period, the pace accelerates.

Agriculture becomes increasingly central to life on the peninsula. Millet remains important, while rice cultivation expands. Settlements grow. Social hierarchy becomes more visible. Dolmens proliferate. Bronze objects appear, followed eventually by iron.

The peninsula is deeply connected to Liaodong and the wider mainland, while maritime networks connect southern Korea to northern Kyushu and the Japanese archipelago.

This is also the period in which language finally begins to enter the story, although frustratingly indirectly.

Some linguistic models place Japonic-speaking populations on the Korean Peninsula during part of this period. Koreanic may have existed farther north before subsequently expanding southward. Other models place Koreanic on the peninsula substantially earlier.

The details remain disputed.

What is much harder to dispute is what happens across the Korea Strait.

Beginning around the first millennium BCE, people from the Korean Peninsula begin moving into northern Kyushu in substantial numbers. They bring continental agriculture, technologies, settlement practices, and ancestry.

The Yayoi transformation begins.

These migrants mix with the Jōmon populations already living in Japan. Over the following centuries, their descendants spread eastward through the archipelago.

For the first time, we can begin to see why modern Koreans and Japanese look simultaneously related and profoundly different.

Both populations carry enormous amounts of O1b2, substantial O2, and smaller amounts of C2.

But Japan also retains the much older Jōmon-associated D1a2a and C1a1 paternal layer that is almost absent from Korea.

DNA:

• Bronze Age Korea appears increasingly shifted toward continental Northeast Asian ancestry, while retaining older peninsular ancestry.
• O2 probably increases with continuing mainland population movements.
• O1b2 is becoming one of the major paternal lineages associated with the Korea–Japan population corridor.
• C2 continues to connect Korea to populations farther north in Manchuria and Northeast Asia.
• Large-scale migration from the Korean Peninsula into Japan introduces continental ancestry that will become a major component of later Japanese populations.
• Jōmon-related ancestry remains dominant in some Japanese populations but becomes increasingly mixed with continental ancestry during the Yayoi period.

~300 BCE–700 CE: the historical blender

And then, suddenly, the people acquire names.

Gojoseon. Buyeo. Goguryeo. Okjeo. Dongye. Samhan. Gaya. Baekje. Silla.

The Han Empire destroys Gojoseon and establishes commanderies in northern Korea. Buyeo occupies parts of Manchuria. Goguryeo expands across the Yalu. Mahan, Jinhan, and Byeonhan develop in the south. Baekje, Silla, and Gaya emerge from this increasingly crowded political landscape.

But these names should not fool us into imagining sealed populations.

A Goguryeo person does not possess "Goguryeo DNA."

Neither does a Baekje person possess "Baekje DNA."

People move across borders. Kingdoms conquer neighboring populations and absorb them. Refugees flee wars. Elites marry across political boundaries. Farmers relocate. Soldiers settle conquered territory. States collapse and their populations are incorporated into whatever replaces them.

Languages move too.

Koreanic languages spread across most or all of the peninsula. Whatever Japonic languages may once have survived there gradually disappear. Political identity changes repeatedly while biological ancestry continues flowing underneath it.

Ancient genomes from the Three Kingdoms period show that this mixing has not entirely finished. Individuals from Gaya, for example, can differ substantially in the amount of Jōmon-related ancestry they carry even while belonging to the same historical society.

Yet they are already genetically close to modern Koreans.

By the end of this period, the dozens of older population layers have become increasingly difficult to separate from one another.

DNA:

• Three Kingdoms-period Koreans already show substantial genetic continuity with modern Koreans.
• Individuals within the same historical population can nevertheless carry noticeably different proportions of older Jōmon-related and continental Northeast Asian ancestry.
• The paternal population probably already contains the major lineages that dominate Korea today: O2, O1b2, C2, and N.
• Those haplogroups no longer correspond to separate recognizable peoples. Their paternal ancestors may have entered the population through migrations separated by thousands or tens of thousands of years.

By the time we reach Silla, Baekje, and Goguryeo, asking whether Koreans came from C2 people, O2 people, or O1b2 people has probably become the wrong question.

They came from all of them.

Different populations had entered the peninsula at different times, from different directions, carrying different genes, technologies, subsistence strategies, and probably different languages.

Then they lived together.

They mixed.

They absorbed one another.

And eventually, somewhere in that process, their descendants became Korean.

Not perfect, but at least it's an imperfect skeleton of a story.

- ה -

So if bones, modern and old, don't give us Koreans, let's look at something else.

Language.

///

The few fragments of DNA that Korea didn't eat:

We know that Korean soil eats DNA. But we also mentioned that we found some bodies that preserved their genes. Unfortunately, there aren't a lot of published studies. Korean soil must have been hungry.

Ando c. 6300–3000 BCE

(여수 안도패총 (麗水 安島貝塚))

Y-DNA: O2a2a1a1b2

mtDNA: unresolved

Yŏndaedo c. 5500 BCE

(통영 연대도 패총 (統營 煙臺島 貝塚))

Y-DNA: C1a1 (previously called C-M8)

mtDNA: D4e2d

Changhang c. 4700 BCE

(부산 가덕도 장항유적)

Y-DNA: D1a2a1

mtDNA: D4b2a

Yokchido c. 2500–1500 BCE

(통영 욕지도 패총 (統營 欲知島 貝塚))

Y-DNA: -

mtDNA: M7a

Taejungni 768–542 BCE

(서산 대죽리 유적 / 대죽리 패총)

Y-DNA: C2-M217 (C2a-L1373)

mtDNA: unresolved

Gimhae / Gaya 300–500 CE

(김해 대성동 고분군 + 김해 유하리 패총)

Y-DNA: D1a2a1, O1b2a1a2a1b

mtDNA: D4e2, D4e2a, B4c1a1a1a, M10a1b, F1b1a1a1, D4a1, D5a2a1a1

Imdang-Joyeong / Silla sphere 4th–6th c. CE

(경산 임당동과 조영동 고분군)

Y-DNA: O1b2a

mtDNA: highly diverse mtDNA

Not enough samples to say anything definitive.

But based on the available published articles:

Y-DNA:

• The oldest DNA discovered is O2 (heavily represented in Han (漢) Chinese.
• The second oldest is C1a1 (the genetic signature of the Jōmon people; and found among the Ainu)
• The third is D1 (another Jōmon marker, and the most prevalent among the Ainu).
• The fourth was C2 (which later became associated with the "Genghis Khan" gene).
• Finally, O1b2 first appears in the currently published Korean ancient-DNA record, around 300–500 CE, in Gaya of all places.

mtDNA:

• D4 appeared much earlier in 5500 BCE, on an island off the southern coast of Korea.

It seems our women are ancient.

Our men, not so much (or Korean soil especially likes eating Korean men).

///

- ה -