For geneticists, like me, genetic testing has always been a game of trust. We can test people’s DNA to learn about their ancestries, relatives, disease predispositions and crimes. We geneticists do genetic testing after providing counseling that explores the process and outcome. In turn, we request consent to proceed.
One of my common questions is, “Suppose that the test result does not meet your expectations?” There are many ways that their expectations may not be met. They can learn about biological mothers and fathers who did not rear them. This discovery can change their relationships with their parents of rearing and other family members. They can learn that they had Jewish, European, African or other ancestries that they were not expecting. This discovery can challenge their self-identity. They can discover that they are at risk for life-threatening illnesses that they were neither expecting nor prepared for, or that the life-threatening illnesses that they were always expecting will not develop. They can also learn new factoids that are simply untrue. Learning the unexpected, especially when it is untrue, can breach trust.
The most recent factoid that some 23andMe genetic testing users learned was that their Y chromosomes (called “haplogroup R-M198”) may descend from a single Khazarian male who lived during the first millennium of the Common Era. What gave this factoid weight is the well-established fact that the R-M198 lineage is present among 50% of Ashkenazi Jewish Levite males. As reported in these pages, the 23andMe disclosure was picked up by members of the “alt-right” with one member tweeting: “Science questions everything. Khazar origin is a valid hypothesis with expanding evidence.” Following the tweets, 23andMe retracted these reports, apologized to its customers and withdrew any references to Khazars from its website.
So why did a fact get embellished by a factoid that was tweeted by the “alt-right”? The simple answer is that the threshold for acceptable scientific evidence was lowered.
The Khazars were a local Turkic people along a trade route between the Caucasus and Volga who may have converted during the first millennium. Other peoples who lived along trade routes in classical antiquity and after may also have converted to Judaism, so conversion itself is not particularly noteworthy. Their purported conversion to Judaism was magnified into a major founding event for the Jews by Arthur Koestler in his 1976 book, “The Thirteenth Tribe.” Koestler popularized an earlier thesis that most Ashkenazi Jews were not descended from Middle Eastern and European progenitors, but rather from Khazars who moved into current Russia, Ukraine and Poland after their kingdom was overrun by the Russians in 965 C.E. The story of the few converted Khazars became the story of the majority of contemporary Jews. Koestler, himself Jewish, was well intentioned and thought that a Khazarian origin for Ashkenazi Jews would have diminished anti-Semitism: The progenitors of contemporary Jews could simply not have been present at the killing of Jesus Christ. His intentions were ill-founded. From its inception in the 19th century, Khazarian origin theory promoted anti-Semitism, precisely because it denies or plays down a Semitic origin for Ashkenazi Jews. Jews are converts, not Jews. Or as tweeted, “”Zion is a castle of lies.”
So is there the expanding evidence that the tweet promised? Re-examining the claims of the Khazars’ conversion, Shaul Stampfer decided that many were pseudepigraphic and the remainder was of questionable reliability. Stampfer, a historian at the Hebrew University, stated in an academic monograph, “Many of the most reliable contemporary texts that mention Khazars say nothing about their conversion, nor is there any archaeological evidence for it. This leads to the conclusion that such a conversion never took place.” For the unknowing (like me), pseudoepigraphy refers to spurious or pseudonymous writings that were ascribed to biblical patriarchs or prophets, but were composed much later.
And what about science questioning everything? In recent years, Eran Elhaik has presented two formulations for a Khazarian origin for Ashkenazi Jews. In his first formulation, Elhaik, a geneticist, presented a case for Caucasian ancestry for Ashkenazi Jews using data that were provided by Doron Behar and Karl Skorecki. Elhaik observed genetic similarity with contemporary Armenians and Georgians, populations that he argued represented proxies for Khazars — so proxies of proxies. Elhaik called his paper “The Missing Link in European Ancestry….” His findings were amplified by Ofer Aderet in an article in Ha’aretz that was called “The Jewish People’s Ultimate Treasure Hunt.” Few geneticists believed that there was a missing link or a treasure hunt. Noah Rosenberg, Behar and their collaborators tested Elhaik’s hypothesis by sampling a broader group of Caucasian populations and populations in general. They found that Ashkenazi Jews carried no particular genetic similarity to groups from the South Caucasus. In their 2013 publication, they went on to state, “It cannot be claimed that evidence of Ashkenazi Jewish similarity to Armenians and Georgians reflects a South Caucasus origin for Ashkenazi Jews without also claiming that the same South Caucasus ancestry underlies both Middle Eastern Jews and a large number of non-Jewish populations both from the Middle East and from Mediterranean Europe.”
In his second formulation, Elhaik partnered with Paul Wexler, a linguist. In this formulation of Khazarian and Ashkenazi Jewish origins, Iranian Jews mixed with Slavs along a trade route. Their issue converted the Khazars and Sorbs, created the Yiddish language, then populated Eastern Europe along with Jews from Central Europe. The Yiddish language was not a derivative of German, but rather of Slavic origin. The first Yiddish speakers were not European Jews living in Germany, but rather Slavic-speaking Sorbs who were converted by Iranian Jewish traders and then relexified their language with German and Hebrew words. To prove his point, Elhaik employed a method that he developed, called “Geographic Population Structure.” Remarkably this method placed the Iranian progenitors of Ashkenazi Jews in northeast Turkey in villages with names like “Iskenaz,” “Eskenez,” “Ashanas” and “Aschuz.”
Typically, scientists like simple hypotheses that are easy to prove, and Elhaik and Wexler’s formulation was not simple. More rigorously, their methods were criticized by a Czech-English-Russian research team led by Pavel Flegontov and not known for its prior work on Jewish population genetics. This team published a paper in the same journal in which Elhaik published his two formulations and stated: “In our view, there are major conceptual problems with both the genetic and linguistic parts of the work. We argue that GPS is a provenancing tool suited to inferring the geographic region where a modern and recently unadmixed genome is most likely to arise, but is hardly suitable for admixed populations and for tracing ancestry up to 1,000 years before present, as its authors have previously claimed.” They were similarly critical of the linguistic methods, stating, “”All methods of historical linguistics concur that Yiddish is a Germanic language, with no reliable evidence for Slavic, Iranian or Turkic substrata.”
So science may question, but science does not overturn existing observations, such as a shared origin for Ashkenazi and Sephardic Jews, without very compelling evidence. The population genetics of many groups would have to be revised based on Elhaik’s work, and a recent study from Shai Carmi and his collaborators confirmed that such a revision would not have to take place. Carmi, a geneticist at the Hebrew University, found that the major source of European ancestry among Ashkenazi Jews was southern European, accounting for 60%-80% of the European ancestry or 30%-40% of their total ancestry. They went on to time two key events: the founding of Jewish populations in Europe about 25-50 generations ago, and the subsequent founding of Ashkenazi Jews about 30 generations ago.
Given the controversy swirling around new formulations for Khazarian origins of Ashkenazi Jews, 23andMe’s disclosure based on lowering the threshold for scientific evidence is puzzling. 23andMe is a high-powered research company. In recent years, it recruited top-notch statistical and population geneticists, some of whom were collaborators and colleagues of mine. 23andMe’s founder, Anne Wojcicki, is the daughter of a Jewish woman who was married to an Ashkenazi Jewish man. She hardly seems someone who would be prone to promoting anti-Semitic theories. Seemingly, the sensational trumped common sense.
Unfortunately, marketing the sensational has always been a feature of 23andMe’s strategy that has gained it not only bad press, but also injunctions from regulatory bodies. Geneticists and consumers perceived their genetic test for dry or wet earwax types as frivolous: Pinky fingers will do the trick at no cost. A 2006 report from the Government Accountability Office based on its own direct to consumer testing at 23andMe and three other companies found that the company provided misleading test results based on deceptive marketing and other questionable practices. At a public meeting in Toronto in 2008, where I questioned the analytical validity, clinical validity and clinical utility of 23andMe’s genetic tests, Johanna Mountain, the company’s scientific director, responded that their tests had analytical validity (meaning that they get the same results the second time), but she punted on the issues of clinical validity (how well do they predict disease) and clinical utility (how useful they are for preventing or treating disease). In 2008, 23andMe was one of several direct to consumer genetic testing companies that received cease and desist letters from California and New York. In 2013, the U.S. Food and Drug Administration forced 23andMe to stop marketing and selling its genetic tests. Recently, it received clearance from the agency. So 23andMe has learned to play by the rules, but still tries to push the envelope. Hopefully it will learn from this experience.
In 2010, when Skorecki and I published our articles about the genetic origins of the Jewish Diaspora groups, historian Shlomo Sand told Michael Balter at Science magazine that Hitler would have been proud of the studies. I asked Balter to remove Hitler from the discussion, and I see from my own recent download of his article that the reference to Hitler is gone. Yet, the fact remains that a genetic test can reveal someone’s Jewish identity. Factoids that are meant to mask facts are also damaging by attempting to delegitimize the origins of people.
Many can contribute to preventing these harms. Scientific investigators and their reviewers can consider whether their paradigm-shifting, Copernican discoveries will be delegitimizing and should be recast before being published. Purveyors of genetic tests can reflect before they disclose the sensational. And consumers should beware.
Harry Ostrer is a visiting fellow at the Netherlands Institute for Advanced Study, in Amsterdam, and a professor of pathology and pediatrics at Albert Einstein College of Medicine, in the Bronx.
How 23andMe Fell For Anti-Semitic ‘Khazar’ Canard