The origin of the Ashkenazi Jews in early medieval Europe

Last year’s The time and place of European admixture in Ashkenazi Jewish history is very close to the last word on the genetics of the ethnogenesis of Ashkenazi Jews. Here’s the author summary:

The Ashkenazi Jewish population has resided in Europe for much of its 1000-year existence. However, its ethnic and geographic origins are controversial, due to the scarcity of reliable historical records. Previous genetic studies have found links to Middle-Eastern and European ancestries, but the admixture history has not been studied in detail yet, partly due to technical difficulties in disentangling signals from multiple admixture events. Here, we present an in-depth analysis of the sources of European gene flow and the time of admixture events by using multiple new and existing methods and extensive simulations. Our results suggest a model of at least two events of European admixture. One event slightly pre-dated a late medieval founder event and was likely from a Southern European source. Another event post-dated the founder event and likely occurred in Eastern Europe. These results, as well as the methods introduced, will be highly valuable for geneticists and other researchers interested in Ashkenazi Jewish origins.

Roughly the Ashkenazi Jews are a half and half mix of a Middle Eastern population and various European groups. The majority of the European ancestry is “Southern European,” probably something like Italian. But, a minority of the European ancestry is like “Eastern European.” Additionally, the former admixture pre-dated the bottleneck, and probably dates to ~1000 A.D., while the latter event post-dates the bottleneck.

For years I had thought that Isaac Bashevis Singer’s excellent novel The Slave was interesting but implausible. The reason being that Ashkenazi Jews and their gentile neighbors did not mix by this time, as the European ancestry in Ashkenazi Jews dates to the Roman period.

These results reject that model…the tract-length evidence is persuasive to me that admixture with Slavs did occur.  Some Italian groups are more north shifted, but the most parsimonious explanation while the Eastern European like ancestry came in later is that it tracks Jewish migration into Germany and Poland-Lithuania later.

The dating of admixture is something I’m less sure of. At 625 to 1,250 years before the present, it puts the emergence of the Ashkenazi community firmly in the Christian era. I don’t want to get into too many details, but from what I have read the Church and local authorities frowned on Jews owning Christians slaves, and tried to suppress instances where Christian slave women became concubines to Jewish men or even Judaized.

I had long assumed that these records reflect elite paranoia. If the dates of admixture are right they may reflect a real concern and a phenomenon (the Y and mtDNA evidence strongly point to the likelihood that the pattern was generally partnerships between Jewish men and gentile women).

And to be frank they tell us less about Jews than they do about the nature of “Christian Europe” in the early medieval period. There is one school of Reform Protestant which takes a dim view of how deeply Christian medieval Europe ever was. I think these results support the thesis that Christianity was an elite religion whose grasp upon the masses was more tenuous and illusory than we might imagine. There is also the reality that the feudal Christian state never had totalitarian authority over the population.

In theory Jewish assimilation of Christians to their identity, Judaizing, could be a capital crime. But if these results are correct it was quite common in the formation of the early Ashkenazi community before it moved north and then east. This decentralization and relative weakness of the early medieval Church and state, the superficially of mass Christianity, might also explain how vast regions of France defected from orthodox Christianity for decades in the 12th century during the ascendancy of the Cathars.

On a final note, I decided to do a little probing on the Middle Eastern forebears of the Askhenazi. The paper says that Levantine populations are the most likely source, which is entirely expected. But I wanted more detail, so I used the Human Origins Array dataset. You can see on the PCA above that the Ashkenazi Jews are shifted toward the European (Basque) population away from Middle Easterners, but if you project the line outward it lands on Christian and Muslim Lebanese. Haber et al. last year showed that there was continuity between the modern Lebanese and Caananites, and the Jews were likely originally a form of Canaanite. Curiously, Palestinian samples in the data are strongly shifted away for the Lebanese, toward groups like Saudis.

I understand it’s a hot potato politically, but if I didn’t have a dog in this fight I’d say that the contention that Palestine and Jordan (look at the Jordanian sample positions) underwent some population turnover is likely true (though I’d be curious about the data on Palestinian Christians).


57 thoughts on “The origin of the Ashkenazi Jews in early medieval Europe

  1. @Dx

    I posted about this last year: Glad to see others have seen the pattern.

    It seems obvious that connections to the west, broadly speaking, persisted longer in southeast Ashkenaz than in northeast Ashkenaz. That said, Western Ashkenazim are sort of distinctive PCA-wise, more distant from either major Eastern Ashkenazi cluster than the two are from each other.

    From what I’ve seen in Gedmatch results (and in one of the more recent Behar papers), the lack of Eastern Euro admixture places Western Ashkenazim between Eastern Ashkenazim, on one hand, and the more southerly Euro-Sephardim and Italkim. Either this means that German Jews can be modeled as almost purely Levantine + Italian, or something’s off about the Ashkenazi paradigm discussed above, if we need to account for late classical/early medieval German/French ancestry.

  2. Canon 68 of the Fourth Lateran Council in 1215 required special dress for Jews and Muslims “to prevent sexual intercourse, which has occasionally occurred by mistake.” Presumably any girls who mistakenly cavorted with Jews were expected to go ahead and marry them, so I’m not sure slavery or elitist-only Christianity are necessary to explain the genetic data.

  3. “Curiously, Palestinian samples in the data are strongly shifted away for the Lebanese, toward groups like Saudis”

    In fairness, the Ain Ghazal Bronze Age samples are also shifted towards Saudi compared to the Sidon ones due to different Levant/Iran ratios. Then you have an apparently slightly greater impact of a steppe-carrying element in the Lebanese compared to the Palestinians/Jordanians. It’s likely that Lebanese Christians (and Samaritans) represent more conservative elements but clearly various forces are at play here.

    As for the European input in Askhenazi Jews, in general I think the “Eastern European” input might be potentially slightly overrated compared to others via various methods perhaps because it’s much more recent. From what I’ve generally seen, my guess is that after the predominant Southeast/Italian European portion, a Northwestern European one (meaning likely north of Iberia and Italy) seems most important. There’s also a certain Iberian and North African looking element, perhaps connected with the Sephardim, and a small East Asian looking one that, whether it can be attributed to something like slight Khazar admixture or not, seems unlikely to come from an Eastern European source since, small as it is, it’s comparatively greater in Ashkenazim than any non northern Russian Eastern Europeans.

  4. The folks at seem to think that Iberia plays an underrated role in the Ashkenazi story, and that many of the (relatively limited in overall share) European Y-lineages among Ashkenazim are Iberian in origin. Carmi et al’s discussion of IBD tracts seems to suggest the possibility.

    Re: East Asian admixture, this point gets neglected all the time. I would love to see a deep analysis of the 1-2% East Asian share among Ashkenazim (IIRC, it’s absent from Western Ashkenazim, which makes sense). It seems a bit elevated, particularly in a southerly direction, when I run myself through most models (meanwhile, 23andMe gives its typical “<0.1%" figure). Kevin Brook could attest: there are actually better-supported hypotheses than a Khazar origin for it. mtDNA suggests a (broadly speaking) Chinese connection, which is pretty wild.

  5. Hey Razib,
    I know this is stupid question, but what’s the best way–for a person for whom the results of a personal DNA test are gospel–to reconcile the fact that when Ashkenazi Jews take a personal DNA test their results indicate “100% European” with the conclusions of academic papers that indicate otherwise? Is there anyway to explain the answer here beyond doubt for someone who distrusts all the peer-reviewed literature?

  6. It seems obvious that connections to the west, broadly speaking, persisted longer in southeast Ashkenaz than in northeast Ashkenaz. That said, Western Ashkenazim are sort of distinctive PCA-wise, more distant from either major Eastern Ashkenazi cluster than the two are from each other.

    Lithuania did stand apart historically as it followed the suit of Spain and Portugal in expelling the Jews in the 1490s. Only 10 years later, Lithuania reversed the course and people started moving back, but there seems to have been very few takers. A better-researched (for this early time frame) Pinsk, Lithuania’s 3rd most important Jewish town, has been settled by just 3 extended families in 1506, and for the following two generations received no new in-migration. It looks like Lithuania’s Jews were taxed at a much higher rate than in Poland, which might have been one of the causes. Only in the second half of XVI c., as the economy improved, did the migration take off. But the old-settler families kept their commanding positions in the economy, likely to a demographic advantage.

    Case in point, Risch 2003 analyzes clinical genetic data, including their earlier discovery of TSD 1277 mutation
    The mutation’s epicenter, with 1.1% pop freq, is in Lithuania, and it’s nearly absent in non-Litvaks. The haplotype coalescence is approx 350 years (as much as 500 years). The population expansion rate is as high as doubling per generation. Risch concludes that it is linked to the early-settler founder effects in Lithuania, coupled with hereditary reproductive advantage of their descendants. The paper also documents mutations specific for Galicia and Hungary, but not as specific as the Lithuanian example.

    (Where are the PCAs, btw?)

Comments are closed.