Y6923: a season of breakthroughs

Last time I wrote about E-Y6923/Y6938, the Y-chromosome lineage carried by one in every 17 or 18 Ashkenazi Jewish males, signifying one shared early medieval ancestor, I had no idea when we’d be able to report back with any new findings.

As it turns out, not all that long. An early trickle of new data from Family Tree DNA’s Y-700 customers, read against some previously inscrutable data, has helped to shed light on several of the unknowns I spelled out in January. Among them:

Where did Y6923 originate, and what are its closest relatives?

As of the beginning of this year, as far as we could assume, the common ancestor of all Y6923 men, Ashkenazi and non-Ashkenazi, could be dated to roughly 400 CE, give or take a few hundred years—that is, the tail end of the Roman era in Western Europe.

But what about Y6923’s deeper origins? Unfortunately, the common ancestor of Y6923 and its closest relative, Y4972—found among Romaniote Jews, Mediterranean Europeans, Russians, Armenians, and Gulf Arabs—dates back to 3500 BCE, give or take, that is, right before the first pharaohs of Egypt and the first city-states of Sumer.

One day, population movements in the proto-historic Near East will be interpretable using contemporary Y-chromosomes, but until then, what about the gap between 3500 BCE and 400 CE? In the received history of Canaan, Israel, and the Jewish Diaspora, most of this period is surprisingly well-accounted for—but in the phylogeny of E-Y6923, this span of eras constitutes a yawning chasm.

Ladies and gentlemen, allow me to announce the first step in the closing of the gap. FTDNA has identified a marker downstream of Y4971 (Y6923’s most recent identified ancestor, until now), and upstream of Y6923, called Y6926. Someone who tests positive for Y6926, but negative for Y6923, could be said to be the ‘closest cousin’ of all Y6923 carriers.

Unfortunately, only one individual has been positively identified as Y6926+ and Y6923-, and he hasn’t included any information (even a name) to hint at his origins.

But! In examining the Y-STR data of potential distant Y6923 cousins in the E-M35 project on FTDNA, I caught a real break: it turns out that our unidentified Y6926+ Y6923- individual clusters—apparently within just a few hundred years—with a group of Emirati and Omani men from the same extended clan, along with a few other unidentified individuals.

An informed correspondent backed my assessment: this group of Gulf Arab men is very likely to be Y6926+ (we already know that they’re Y6923-; if they were Y6923+, we would have known by now). They are our closest cousins, and assuming that Latin American Y6923 is of converso origin, Y6923’s closest non-Jewish connection.

First of all, this link strengthens the case for a Middle Eastern origin of Y6923. Inferring beyond that is tough at this stage, but based on a preliminary account, Y6926 essentially ‘splits the difference’ between Y4971 and Y6923 in terms of distinguishing mutations, which suggests a common ancestor for Y6926’s Y6923+ and Y6923- descendants somewhere very roughly around 1600-1500 BCE, or in Levantine archaeological terms, the end of the Middle Bronze Age.

This was a time of upheaval in the region, and if we assume a Levantine, rather than Arabian, origin for Y6926, our tree would seem to correspond with the theory that it was also a time of proto-Arab migrations out of the Levant and into North and East Arabia. Again, this part is highly speculative.

How about Y6923’s big internal split?

In my last post, I wrote about the emerging picture of a split within Y6923 along pretty clean lines: Y6938 (now considered interchangeable with Y6940) characterizing all Ashkenazi members, and Y102667 being the signature of non-Ashkenazi members. We now have a few more data points, and in the broad sense, they appear to strengthen this paradigm.

First, one new individual of Tunisian Jewish origin has appeared on the YFull tree as Y102667+. His direct patriline, however, is said to be from somewhere in Turkey, having arrived in Tunisia in the early 19th century. This new cousin’s surname is attested in two different places: among Kurds in Turkey, and in elongated form, among Maghrebi Jews.

I am doubtful, however, about a Kurdish/Mesopotamian/Eastern Anatolian link: first of all, I don’t think Kurds had standard surnames 200 years ago, second, I don’t believe Jews were ever known to take on Kurdish clan names, and third, Jewish communities that far east were not part of the Sephardic network that would have linked, say, the Jewish communities of Istanbul, Izmir, and Tunis.

In addition, FTDNA has turned up a Y6923+ individual of Algerian (most likely Algerian Jewish) origin in its Y Haplotree; the individual does not appear on the list of Y6938/Y6940 downstream members (Y102667 doesn’t appear on the tree).

Finally, the E-M35 project has identified a Mexican and a Peruvian who are both estimated to be Y6923+, but Y6938/Y6940-, which would seem to make a pattern out of the one Puerto Rican individual who’s been known for a few years to be Y6923+ (now confirmed Y102667+). Given that all other known Y6923+ Y6938/Y6940- / Y102667+ individuals are Sephardic or North African Jews, a converso origin for these Latin American patrilines seems more likely than not.

While none of these new downstream findings changes the possibility space of the Y6938 / Y102667 split much, I’m inclined to say it slightly strengthens the case for an explicitly Sephardic origin of Y102667, and an identification of the split with the early branching of the late classical Western Roman Jewish community into proto-Ashkenazim and proto-Sephardim. That said, North African Y102667 might not have come from Spain, leaving other, less immediately legible possibilities open.

What we’re waiting on

  • More confirmed Y6926+ Y6923- individuals, for better ethnogeographical and chronological insights
  • Identification of sibling branches to Y6923, under Y6926
  • Confirmation of the Algerian individual’s Jewishness
  • Confirmation of the Peruvian, Mexican, and Algerian individual as Y102667+

2 thoughts on “Y6923: a season of breakthroughs

  1. Robert Sommers

    Is this 6926 showing up on Yfull? I see two branches upstream of 6923 and downstream of Y4971, Y4971* and Y4972. Y4971* is a lone Iraqi. On FT 4971 has three downstream branches, 4972, 36952 and 6926. is there a Yfull corollary?

    • whitmanic

      So far YFull doesn’t seem to have gotten its hand on any Y6926+ Y6923- individuals. As it stands, they consider Y6926 to be coextensive with Y6923, along with a few dozen other mutations.

Leave a Reply

Your email address will not be published. Required fields are marked *