Skip to content
Sephardim.com

A documentary register

Blog

The Avar, the Priest, and the Levite: Why Your Y-DNA and Your Ancestry Can Tell Different Stories

· Deric


Sometime between 650 and 800 CE, a man was buried in an Avar-period cemetery at Mödling, just outside Vienna. His grave looked like those of his neighbors, and his DNA looked like theirs too. Hobbyist ancestry calculators put him at over 99% local European, with his closest matches among late Avars and Sarmatian-era people from Hungary. By every autosomal measure, he was an ordinary Central European of his time.

He was not an outsider, either. The cemetery data show him embedded in a large local family: a brother (MGS068) buried in the same cemetery, a daughter (MGS126), and a web of second- and third-degree relatives around them.

His Y chromosome says something else. The sample, catalogued as MGS069 in Wang et al. 2025 (Nature), belongs to J1-Z18271. That's a paternal branch about 2,700–2,800 years old, and today it is overwhelmingly Jewish. A 2025 preprint on the lineages of the Cohanim (Jewish priestly families) found that 9 of Z18271's 11 main sub-branches are entirely or mostly Jewish. They are spread across Ashkenazi, Sephardi, Yemenite, Bukharan, Iraqi and Syrian communities, and most are linked to Cohen status. Its parent branch has been found in Bronze and Iron Age remains from Megiddo in Israel and Sidon in Lebanon.

One honest limit: MGS069's own Y call stops at Z18271 itself. It isn't resolved to one of the eleven sub-branches, and two of those eleven are not Jewish. So the most we can say from his sample alone is that he carried a line from a clade that is today overwhelmingly Jewish.

But he isn't the only one. MGS071, a second-degree relative buried in the same cemetery, is resolved — to J-S17446, a sub-branch whose modern carriers are overwhelmingly Jewish: Ashkenazi families from Poland, Lithuania, Ukraine, Russia and Moldova, a Judeo-Moroccan speaker, a Hebrew-speaking Egyptian — and one man from Guerrero, Mexico. The genetic relationship code doesn't say whether MGS069 and MGS071 are related through their fathers or their mothers. If it's the paternal side, MGS069 carries S17446 too; his lower-coverage sample simply didn't reach those markers.

So how do men from a clade like this end up with no Levantine ancestry at all?

Two kinds of inheritance

Your autosomal DNA is a shuffled mix of all your ancestors, and each one's share roughly halves every generation. A great-great-grandfather contributes about 6%. By seven or eight generations back, a single ancestor is down to about 1% and effectively invisible.

The Y chromosome works differently. It passes from father to son almost unchanged, and nothing dilutes it. One Levantine man in Roman times whose sons, grandsons and great-grandsons all married local women would produce, 15 or 20 generations later, a man who is fully European autosomally. That man would still carry the same Y chromosome.

That's almost certainly MGS069's story. The calculator models show essentially no Levantine signal — 0.1% Natufian in one of them, which is noise. That rules out a Jewish grandfather or great-grandfather and points to an ancestor far back, most plausibly in Roman Pannonia.

Jews in Roman Pannonia

This isn't far-fetched. Mödling sits in what was the Roman province of Pannonia, a frontier region full of soldiers, merchants and settlers from across the empire, including the Levant. In 2008, archaeologists at Halbturn, only a short distance from Mödling, found a tiny gold-plated amulet in the grave of a one- or two-year-old child from the 3rd century CE. Inside was a scroll bearing the Shema, "Hear, O Israel," written in Greek letters. It's the earliest evidence of Jewish life in what is now Austria. Nobody has tested that child's DNA, but the amulet shows that people connected to Judaism were living in this landscape centuries before MGS069 was born.

A recently released sample from the Levant, I37241, from the Akbari et al. 2026 dataset, is placed by YFull in the same Z18271 clade. Like MGS069, it is resolved only to Z18271 itself, so we can't yet say whether the two share a lower sub-branch or how recently their paternal lines meet. Its public metadata says only "Israel"; online reports that it comes from a Roman-period grave in Judea or Galilee are still unconfirmed. If the dating holds and deeper testing ever puts both men in the same sub-branch, they would be connected by a paternal ancestor who lived around Roman times — a single male line running from the Levant to the Vienna Basin. That's a hypothesis worth watching, not yet a finding.

This happens all the time

MGS069 is an early example of a very common pattern:

  • Spain and Latin America. Centuries after the forced conversions of 1391–1492, Catholic Spaniards, Portuguese and Latin Americans sometimes carry Sephardic Y lineages with little or no Jewish autosomal DNA. Z18271 itself has examples: a man from Lima, Peru, in the 1000 Genomes Project (HG01571) sits in its J-ZS232 branch, and the tester from Guerrero in J-S17446 is another. Because Z18271 is only about 2,800 years old, membership in it means something — unlike a haplogroup whose nearest Jewish relative lived five thousand years ago.
  • Italy and the Mediterranean. Two of Z18271's sub-branches have so far been found only in non-Jewish Southern Europeans and Near Easterners. Their ancestors may have been assimilated Roman-era Jews, Christian converts, or Israelite-era lines that were never part of Judaism.
  • Eastern Europe. Poles, Ukrainians and Hungarians sometimes carry specifically Ashkenazi branches after baptisms, adoptions or children born outside marriage. These events are more recent, so a few percent of Jewish autosomal DNA often still shows.

And the reverse: the Levites

The opposite also happens. A community can be deeply Jewish autosomally while one of its core paternal lines began outside the Levant.

The best-known case is R1a-Y2619, carried by a large share of Ashkenazi Levites. Because R1a is common in Eastern Europe, people long assumed it came from Slavic or Khazar converts. The data says otherwise. Behar et al. (2017) found the Levite branch "well nested within a plethora of phylogenetically close Middle Eastern sister-clades," with its closest relatives among Iranian Azeris, a man from Kerman, a Yazidi and one Iberian sample. The Ashkenazi cluster traces back about 1,743 years (95% range 1,334–2,200) to a single ancestor, and it split from those Middle Eastern relatives about 3,143 years ago (2,620–3,682).

One idea, discussed by Levite-lineage researchers rather than in the paper itself, is that the founder was a Persian or Mesopotamian man who joined the Jewish people around the Babylonian Exile — perhaps one of the temple servants (Nethinim) who returned with the exiles. His descendants became fully Jewish in every autosomal sense, and some later took on Levite status. Their Y chromosome still carries the signature of an ancient Middle Eastern ancestor. It's an attractive hypothesis, and it is a hypothesis.

What a haplogroup can and can't tell you

A Y-DNA haplogroup traces one line out of thousands. It tells you where one great-great-great-…-grandfather's line came from, not who you are, what you believed, or where most of your ancestors lived. Here the two kinds of evidence answer different questions:

  • MGS069 was a Central European of the Avar era whose distant paternal ancestor very likely came from the Levant.
  • The Ashkenazi Levites are a Jewish community whose paternal founder very likely came from the Middle East beyond the Levant — most plausibly Iran.

And the question that decides whether a haplogroup tells you anything at all is how recently your branch meets the nearest Jewish branch. For Z18271 that's within the last 2,800 years, which is why this story is worth telling. For most people's haplogroups, it's five thousand years or more — and then the Y chromosome is telling you about the Stone Age, not about your family.

Sources