Skip to main content
The method, not the marketing

What DNA can actually tell you

Ancestry testing is sold as a percentage pie chart. What it really produces is far narrower, far stranger, and far more interesting.

What a haplogroup is

DNA copies itself imperfectly. Every so often a copying error sticks, gets passed to the next generation, and stays there — harmless, permanent, and inherited by everyone descended from that person along that line.

A haplogroup is just the set of people who share such a marker. It is a branch on a tree, and the tree is the shape of human descent. Nothing about it is a nationality, a percentage, or a personality.

Because the markers accumulate, branches nest inside branches. E-PF2546 sits inside E-M81, which sits inside E-M96. The deeper you go, the smaller and more recent the group.

Two threads out of thirty-two

Y-DNA passes from father to son. Mitochondrial DNA passes from a mother to all her children, but onward only through her daughters. Each traces one unbroken line and says nothing about anyone else.

Thirty-two ancestors, two visible lines A row of 32 squares representing a person's ancestors five generations back. Only the first square, the all-paternal line traced by Y-DNA, and the last square, the all-maternal line traced by mitochondrial DNA, are highlighted. The remaining 30 are unmarked, because neither test describes them. Your 32 ancestors, five generations back Y-DNA father’s father’s father… mtDNA …mother’s mother’s mother 30 ancestors, invisible to both
This is the whole limitation, in one row. Go back five generations and you have 32 ancestors. Y-DNA describes one of them. mtDNA describes one more. Go back ten generations and it is two out of 1,024. Everything else — including most of where anyone actually comes from — is outside what these two tests can see.

That is not a flaw. A single unbroken line is exactly what makes these markers useful: because they do not get shuffled each generation, they preserve a readable record over tens of thousands of years. The mistake is reading two threads as a whole cloth. More on what that rules out.

The molecular clock

Where the dates come from, and why they wobble

Nobody digs up a date. To estimate when a lineage began, you count the mutations separating two sequences and divide by how fast mutations are thought to accumulate. That rate is itself estimated — and different studies choose differently.

The consequence is that respectable papers can disagree by thousands of years about the same lineage, using the same data.

Two published age ranges for E-M81 Solé-Morata and colleagues in 2017 estimate the E-M81 common ancestor at 2,000 to 3,000 years ago. D'Atanasio and colleagues in 2018 estimate 2,000 to 4,200 years ago. The ranges overlap but the second is far wider. When did E-M81 begin? Two answers. Solé-Morata 2017 2,000–3,000 years ago D’Atanasio 2018 2,000–4,200 years ago 2,000 4,200 7,000 years ago
Bar length is the uncertainty, not the answer. The wider estimate spans the difference between a Roman-era expansion and a Bronze Age one. When this site prints a single number, it is a midpoint — and the bar is what the evidence supports.

What ancient DNA changed

Before

Prehistory was inferred backwards from living people. You measured today's diversity and reasoned about what must have produced it — powerful, but always an inference.

After

Sequencing the dead tests those inferences directly. Several long-standing stories about the Maghreb changed once genomes came out of the ground rather than out of a cheek swab.

It also has a hard limit here: DNA degrades in heat, and the Maghreb is warm. North African ancient genomes remain scarce, so conclusions rest on few individuals from few sites — and the picture has shifted more than once in the last decade. See the papers.

Last updated: 10 September 2026