The Sephardim of France: Descendants of Sefarad, or the Maghreb of 1962?

Roughly two thirds of the Jews of France describe themselves as Sephardic. Sefarad is the medieval Hebrew name for the Iberian peninsula, and almost none of those families came from Iberia. They came from Algiers, Constantine, Tunis, Sousse, Fez and Casablanca, and most of them arrived between 1956 and 1967. This article asks what survives of the label when it is put through a calculator. It uses a pooled Global25 estimate of the present-day French Jewish population and, for the first time on this site, the ten usable genomes recovered from the mass graves of the 1348 pogrom at Tarrega in Catalonia, published in March 2026 and the only direct sample of the Jews of Sefarad that exists. The result has three parts, and only two of them are the ones anybody expects.

A word that changed continents

Sefarad appears once in the Hebrew Bible, in a single verse of Obadiah about exiles from Jerusalem, and nobody knows what place the prophet meant. By the eighth century Jewish writers had attached the name to the Iberian peninsula, and it stayed there. The Jews expelled from Castile and Aragon in 1492, and from Portugal in 1497, were therefore the Jews of Sefarad in the plainest possible sense: they were named after the country they had just been thrown out of.

They scattered in every direction. The largest share went east into Ottoman territory, to Salonica, Constantinople, Izmir and Safed, where they kept Castilian as a spoken language for four hundred and fifty years. A second stream crossed to North Africa, into Fez, Tetouan, Tlemcen, Algiers, Tunis and Oran. A third went to Italy, a fourth to Amsterdam and Hamburg by way of a generation or two of outward Catholicism, and a fifth, the same converso merchant families, settled quietly at Bordeaux, Bayonne and Saint-Esprit under royal letters patent that did not mention that they were Jews.

Then the word came loose from the place. Over the following centuries the Sephardic prayer rite spread across the Maghreb and much of the Middle East, partly carried by the exiles themselves and partly by the enormous rabbinic authority of the codes they produced. A liturgical order that had originated in Iberia became the standard order from Marrakesh to Baghdad. By the twentieth century, and especially in Israeli administrative usage after 1948, sepharadi had stopped meaning descended from the Jews of Spain and had come to mean not Ashkenazi. A Jew from Djerba, whose community may predate the Roman conquest of Africa, and a Jew from Sanaa, whose ancestors certainly never saw Europe, were both filed under the same heading.

France inherited that vocabulary, and then inherited a population to attach it to.

What happened to French Jewry between 1956 and 1967

Before the war France had something on the order of three hundred thousand Jews, overwhelmingly Ashkenazi: the old Alsatian and Lorraine communities, the Comtadin remnant of the papal enclave around Carpentras, the Portuguese families of Bordeaux and Bayonne, and on top of all of them a very large immigration from Poland, Russia and Romania between 1881 and 1939. Around seventy-six thousand were deported, of whom roughly seventy-three thousand five hundred did not return, about a quarter of the community. Between 1945 and 1948 some eighty thousand survivors and displaced persons arrived from central and eastern Europe and brought the numbers roughly back to where they had been.

Then the French empire in North Africa ended. Morocco and Tunisia became independent in 1956, Algeria in 1962. Something in the region of three hundred thousand North African Jews left for France across the following decade, and the Algerian component was not immigration at all in the legal sense: the Cremieux decree of 1870 had made the Jews of Algeria French citizens, so when about a hundred and ten thousand of them out of a community of a hundred and thirty to a hundred and sixty thousand crossed the Mediterranean in 1962 they crossed as repatriates. They arrived with French passports, French schooling, and in many cases French war records.

The community that resulted is the largest in Europe and the third largest in the world, most recently estimated at between four hundred and forty thousand and five hundred thousand people on the narrow definition. Somewhere between sixty and seventy per cent of it is routinely described as Sephardic. What that means in practice is Maghrebi: Algerian first, then Tunisian, then Moroccan, with a smaller Egyptian element that left after 1956.

The irony is worth stating before the measurements start. The one French Jewish community that genuinely was Iberian by descent, and that used the word Sephardic in its original sense, was the Portuguese nation of Bordeaux and Bayonne. It numbered a few thousand people, it was culturally distinct and proud of it, and it dissolved into the general French Jewish population across the nineteenth and twentieth centuries. The name it carried is now borne by half a million people who reached France from the other side of the Mediterranean.

What the uniparental markers already said

Two observations framed the question long before genome-wide data existed, and they point the same way.

On the paternal side, the Maghrebi Jewish communities do not look like their Muslim neighbours. The lineage that dominates North African Muslim populations is E1b1b1b1a, better known by the marker M81, the Berber-modal haplogroup, which runs above half in many Moroccan and Algerian samples and reaches extraordinary frequencies in some Berber-speaking villages. It is a minority lineage among Maghrebi Jews, who instead carry the standard Jewish diaspora set: J1, J2 and the eastern branches of E1b1b, in proportions much closer to those of Jewish communities elsewhere in the Mediterranean than to those of the countries they lived in.

On the maternal side, Behar and colleagues showed in 2008 that several North African Jewish communities descend from a startlingly small number of founding women. The Libyan, Djerban and to a lesser extent Tunisian maternal pools are dominated by a handful of lineages, the pattern produced by a small founding group followed by centuries of endogamy. Moroccan Jewish maternal diversity is much higher. A community formed by mass conversion of local women looks like the local women; these do not.

So the uniparental evidence already argues against the Berber conversion story and in favour of small founding populations that stayed small and married inward. What it cannot do is say when those founders arrived, or from where. That is the autosomal question.

Data and method

Everything below uses Global25 scaled coordinates, with population averages merged from Davidski's published files and the Moriopoulos 2026 collection. Ancestry proportions come from non-negative least squares with a sum-to-one constraint imposed by appending a heavily weighted row of ones, with the resulting weights normalised. Distances are scaled Euclidean distances multiplied by one thousand, the convention used throughout this site.

There is no population in Global25 called French Jews, and there cannot be, because French Jewry is not a breeding population with a common history: it is three or four communities that have been sharing a country for sixty years. The target used here is therefore a construct, and it is worth being blunt about that. It is a weighted average of the components in roughly the proportions the demographic literature gives: sixty per cent Maghrebi, split thirty-three per cent Algerian, fifteen Tunisian and twelve Moroccan in line with the size of the three migrations, thirty-two per cent Ashkenazi using the eight-individual French Ashkenazi average rather than the pooled diaspora one, and eight per cent standing in for the Egyptian, Turkish, Levantine and residual old-French element, for which the Sephardim of Turkey are the least bad available proxy.

That construct is not fragile. Shifting the balance to fifty per cent Maghrebi and forty Ashkenazi moves it 2.8 units. Shifting it the other way, to seventy and twenty-five, moves it 2.8 units. Splitting the Maghrebi share equally between the three countries instead of by migration size moves it 1.7. The one substitution that matters is replacing the French Ashkenazi average with the pooled 1,209-individual Ashkenazi average, which moves it 6.2 units, and the reason for that is discussed later. Every one of those numbers is smaller than the smallest gap this article rests on.

The ancient anchor is new. Cristina Santos and colleagues published sixteen individuals from the Roquetes necropolis at Tarrega, in Lleida province, in March 2026. The site contains six communal graves holding at least sixty-nine bodies, many with lethal trauma to the skull and face, dated to the middle of the fourteenth century and attributed to the pogrom that destroyed the town's Jewish community during the Black Death in 1348. Eleven individuals yielded enough data for full analysis. Their qpAdm modelling resolves the population as roughly sixty-nine per cent Canaanite and thirty-one per cent medieval Iberian non-Jewish and non-Islamic, they carry the paternal and maternal lineages characteristic of the Jewish diaspora, and their runs of homozygosity indicate background endogamy without close kinship, notably less inbred than the medieval Jewish communities sampled at Erfurt and at Norwich. Ten of them are in the Global25 release and are pooled here as a single average.

Two other medieval Jewish communities are used as comparison anchors. The Erfurt burials in Germany, dated to around 1300 and split by the excavators into an eastern and a western profile, of which the eleven-individual western group is used here. And the six genomes from a well at Chapelfield in Norwich, dated to the late twelfth century and published by Brace and colleagues in 2022, from a community that had been founded from Rouen after 1066 and that Hebrew sources call Tzarfati, which simply means French.

One panel does the work in every decomposition below, so that rows are directly comparable: a Bronze Age Canaanite source from Megiddo, a medieval Iberian Christian source, a Moroccan Shilha Berber source, and a sub-Saharan African source. The condition number of that matrix is 6.11 and the pole separations run from 167 units between the Canaanite and Iberian poles up to 758, all of them comfortably above the hundred-unit threshold below which we treat two-source separation as unreliable on this site. A deep seven-source panel of Levantine, Anatolian and Iranian Neolithic, Yamnaya steppe, western hunter-gatherer, Iberomaurusian and a sub-Saharan pole, with a condition number of 12.1, reproduces the same ordering; the proximal panel is used because it resolves recent admixture much better.

Where the Jews of France sit

The simplest measurement sets up everything else.

The five nearest populations to the pooled French Jewish estimate are Algerian Jews at 10.3, Turkish Jews at 12.5, the Sephardim of Turkey at 13.3, Moroccan Jews at 14.6 and the Italkim at 14.7. The Ashkenazim of France sit at 19.1, the Tunisians at 21.5 and the Libyans at 25.9. Nothing that is not a Jewish community appears until the Maltese at 29.1, and nothing that is not a Jewish community and not an eastern Mediterranean island population appears until the Sicilians at 36.3.

Then look at the bottom of the chart. The French of the HGDP average sit at 125.8 units. The Occitans of Marseille, where a very large share of the repatriates of 1962 settled, sit at 106.0. Catalans sit at 111.8, Andalusians at 97.3. The Muslim populations of the three countries the Maghrebi majority came from sit at 137.6 for Tunisia, 144.7 for Algeria and 170.9 for Morocco, and the Kabyle Berbers at 131.9.

Those numbers deserve to be read next to each other, because together they make the central structural point of this article. The distance from an Algerian Jew to an Alsatian Ashkenazi, the two ends of the internal spread of French Jewry, is 28.1 units. The distance from either of them to the French population they live among is between 112 and 126. The distance from a Moroccan Jew to a Moroccan Arab is 160.4, and to a Moroccan Berber 181.9, while the distance from that same Moroccan Jew to an Algerian Jew is 14.2 and to a French Ashkenazi 31.1.

Whatever Sephardic and Ashkenazi mean inside a French synagogue, they do not describe two populations in the sense that a Frenchman and a Moroccan are two populations. They describe two ends of a range roughly a quarter as wide as the gap between either end and the country they are standing in. This is the orthogonality that Campbell and colleagues reported in 2012 from a different direction: North African Jewish and North African non-Jewish populations form distinct clusters that are close to perpendicular to one another, and the Jewish ones cluster with Jewish communities from everywhere else.

The Jews of Sefarad, measured directly

Now the anchor. Ten people were buried in a hurry in a Catalan ditch in 1348, and their coordinates can be compared with everything that has been proposed as their descendant.

The nearest population on the chart is the medieval Jewish community of Norwich, at 18.4 units, whose ancestors had come from Normandy two and a half centuries earlier. The nearest living population is the Ashkenazi community of France, at 19.9. Then the pooled French Jewish estimate at 21.4, the Maltese at 21.8, the Italkim, the western Erfurt Ashkenazim and the Turkish Jews all at 24.2, the Sephardim of Turkey at 25.5, Moroccan Jews at 25.9, Algerian Jews at 27.2 and the pooled Maghrebi component of French Jewry at 28.5. Tunisian Jews sit at 38.0 and Libyan Jews at 40.2.

Before anything is made of that ordering, two things have to be said about it. The first is that the Tarrega average is ten low-coverage ancient genomes and its fit in the four-source panel is 19.9 units, so gaps of a few units on this chart are not real. The five populations sitting between 24.2 and 25.9, which include the Sephardim of Turkey and the Moroccan Jews and the medieval Ashkenazim of Erfurt, are a tie and should be read as one. The second is that a great deal of the spread is drift rather than descent, and the clearest evidence for that is on the chart itself: the pooled Ashkenazi average of 1,209 individuals sits at 31.1, further than the Moroccan Jews, while the medieval western Ashkenazim of Erfurt sit at 24.2 and the modern French Ashkenazim at 19.9. The Ashkenazi founder bottleneck of the late Middle Ages moved that population away from its own ancestors. It did not move the French sample as far, which is why the French sample is used as the Ashkenazi component of the construct.

With those two caveats in place, the ordering still says something that most readers will not expect. Of every living population measured here, the one closest to the Jews of Sefarad is an Ashkenazi one. The communities that carry the name Sephardic sit further away, and the two that carry it with the strongest liturgical claim, the Tunisians and the Libyans, sit furthest of all.

The explanation is not that anybody is a fraud. It is that the medieval Jewish communities of the Latin West were, genetically, one population. Tarrega to Norwich is 18.4 units. Norwich to Erfurt is 18.3. Tarrega to Erfurt is 24.2. Those three communities were separated by the Pyrenees, the Channel and six hundred kilometres of Germany, they used different rites, spoke different vernaculars and would later be assigned to opposite sides of the great taxonomic divide of Jewish history, and they sit within twenty-five units of each other. The Sephardi and Ashkenazi categories are real as cultural traditions and as legal jurisdictions. They are not a partition of medieval Jewry into two ancestral stocks, because in the fourteenth century there was no such partition to make.

One more row on that chart matters. The Christians of medieval Iberia sit 83.8 units from the Jews of Tarrega, the Andalusians of today at 81.8 and the Catalans at 95.2. Whatever the Jews of Catalonia were in 1348, they were not converted Catalans, and the Canaanite pole at 90.7 shows they were not unmixed Levantines either. The published qpAdm figure of roughly two thirds Canaanite and one third local is exactly what the geometry here reproduces.

The Maghrebi layer

Breaking the same populations into proximal components gives the shape of each gene pool and isolates the one axis on which the Maghrebi Jewish communities genuinely differ from every other Jewish community in the western diaspora.

The pooled French Jewish estimate comes out at 53.3 per cent Canaanite, 40.3 per cent European, 5.3 per cent Maghrebi and 1.1 per cent sub-Saharan African. Its Maghrebi component alone returns 54.1, 35.3, 9.5 and 1.1. The Ashkenazim of France return 50.4, 48.7, 0.0 and 0.8. The Sephardim of Turkey return 55.4, 42.1, 2.0 and 0.5. The Jews of Tarrega return 43.1, 49.9, 6.5 and 0.5.

Pull the Maghrebi column out on its own and put the neighbours next to it.

Moroccan Jews return 13.5 per cent, Libyan Jews 12.3, Tunisian Jews 12.2, the pooled Maghrebi component of French Jewry 9.5 and Algerian Jews 6.8. Their Muslim neighbours, on the identical panel, return 65.4 per cent for Moroccan Arabs, 63.3 for Kabyle Berbers, 59.3 for Algerian Arabs, 55.2 for Tunisian Arabs and 47.4 for Libyan Arabs.

That is a factor of between five and ten, and it disposes of a hypothesis that has had a long and politically noisy life. The idea that the Jews of North Africa are essentially Judaised Berbers goes back to a passage in Ibn Khaldun, writing in the fourteenth century, about Berber tribes that had professed Judaism before Islam, and it is usually decorated with the story of the Kahina, the Jewish Berber queen who is supposed to have led the resistance to the Arab conquest of the Aures. It was revived by Nahum Slouschz in the early twentieth century and it has been recycled ever since, most recently in the service of the argument that modern Jewish populations have no ancestral connection to the Levant at all. On this panel a Moroccan Jew carries about a fifth of the Maghrebi ancestry of a Moroccan Arab and about a seventh of that of a Kabyle. If mass Berber conversion had built these communities, the number would not look like this.

It is also not zero, and that matters in the other direction. The Ashkenazim of France return 0.0 per cent on the same panel, the Romaniotes 0.0, the Syrian Jews 0.0 and the Sephardim of Turkey 2.0. The Maghrebi Jewish communities are the only Jewish populations in this article that take a real Maghrebi weight. The signal survives every substitution of the Berber pole. Swapping Shilha for Riffian, Kabyle, Shawiya, Mozabite, the eastern Zenati of Chenini, the Berbers of Djerba, or the Neolithic population of Kelif el Boroud moves the Moroccan Jewish figure between 12.6 and 19.7 per cent and the French Jewish figure between 4.2 and 8.9, and never once changes the order of the rows or lifts the Ashkenazi figure off zero. What the choice of source controls is the scale, not the ranking.

One honest complication. The Jews of Tarrega take 6.5 per cent on the same panel, and the Andalusians take 2.3. Iberia under and after Islamic rule absorbed a genuine North African component, and Jewish families moved back and forth across the strait for eight centuries in both directions. Part of what this column reads as Maghrebi in the Jewish communities of the Maghreb may therefore be a Maghrebi signal that was already present in Sefarad before anybody was expelled. The comparison that is safe is the one against the Ashkenazim and the Ottoman Sephardim, at 0.0 and 2.0, and against the local Muslim populations at 47 to 65. The comparison against Tarrega is inside the noise.

The collinearity trap: which Europe?

The obvious next question is what the forty per cent European component is. If a large part of it were specifically Iberian, that would be the measurement the title of this article is asking for. It is not obtainable, and the way it fails is worth showing in full, because this is the single most common route by which home modelling produces confident nonsense about Sephardic ancestry.

Hold the Canaanite, Berber and African sources constant and swap only the European one. With a medieval Iberian Christian source the French Jewish estimate takes 40.3 per cent European. With Andalusians it takes 40.8, with Portuguese 41.1, with Catalans 37.2, with the French of the HGDP 34.0. With Italians of Lazio it takes 55.5, with western Sicilians 66.2, with Cretan Greeks 72.4, and with a single Roman-period individual from Isola Sacra whose profile is eastern Mediterranean and Jewish-like it takes 90.3.

The fit distances across all nine of those models run from 6.4 to 23.5 units. The composition swings by fifty-six percentage points and the fit tells you almost nothing about which version to believe, because the sources are not far enough apart for the model to be choosing between them on evidence. Medieval Iberia and western Sicily are 73 units apart. Medieval Iberia and Lazio are further, but Lazio and the Canaanite pole are only 119 apart, so as the European source moves east it starts substituting for the Levantine one and the two columns trade weight rather than competing. Our working threshold on this site is about 100 units of separation, and the whole southern European end of this panel is below it.

Global25 can say that the non-Levantine component of a French Jew is southern European. It cannot say which southern Europe, and no amount of source selection will change that, because the information is not in the coordinates.

The question the calculator cannot answer

Which brings us to the actual question. Inside the Maghrebi Jewish communities there is a real historical distinction, documented in rabbinic sources, communal ordinances, surnames and dialect. The toshavim were the established Jewish population of North Africa, present in Cyrenaica and Carthage in Roman times and in the Atlas long before the Arab conquest. The megorashim were the expelled, the families of 1492 and 1497, who arrived in numbers large enough to take over the communal institutions of Fez, Tetouan, Tangier, Oran, Algiers and Tunis, who issued their own takkanot at Fez in 1494, who kept the Castilian rite as a distinct minhag for generations, and whose descendants in northern Morocco were still speaking Haketia, a Judeo-Spanish dialect, in the twentieth century.

How much of a French Jew of Algerian or Moroccan descent comes from the megorashim and how much from the toshavim? That is what the title of this article promises, and the honest answer is that this method cannot deliver it. It is worth demonstrating why, because the demonstration produces a number that looks perfectly respectable and is worthless.

Model each population with two Jewish sources rather than one: the Jews of Tarrega, standing for Sefarad in 1348, and the western Ashkenazim of Erfurt around 1300, standing for a medieval western Jewish population that certainly was not Iberian, alongside the Berber and Canaanite poles. The condition number is 29.4 and the separation between the two Jewish sources is 24 units.

Moroccan Jews come out at 32.1 per cent Sefarad, Tunisian Jews at 31.9, Algerian Jews at 27.1, the pooled French Jewish estimate at 31.8 and the Ashkenazim of France at 28.6. Libyan Jews come out at 83.3 per cent. The Sephardim of Turkey come out at 12.1 per cent, the lowest figure on the chart.

The Sephardim of Turkey are the positive control, and they fail it completely. That community descends from the exiles of 1492 by a documented route: it was settled by Ottoman decree, its congregations were named after the Iberian towns their founders came from, Aragon, Catalan, Cordova, Toledo, and it spoke Castilian until the twentieth century. If the model were measuring Iberian Jewish ancestry, the Sephardim of Turkey would be at the top of this chart. They are at the bottom, below the Jews of Tripoli, whose community is one of the oldest in the Mediterranean and which received relatively few Iberian exiles, and who score seven times higher.

This is the same failure this site has now documented in three articles. When a candidate source overlaps substantially with the target's existing baseline, non-negative least squares assigns it weight because doing so shaves a little off the residual, and the resulting number is real arithmetic describing an ancestry that is not being measured. Readers of our piece on the Yemenite Jews will recognise it exactly: there, a Canaanite source assigned three times more Judaean ancestry to Muslim Yemenis than to Yemenite Jews.

The instability is as bad as the ordering. Perturb each target by twelve units in a random direction, roughly the size of a typical fit residual in this article, and the estimate for the Moroccan Jews runs from 8 to 55 per cent across the middle eighty per cent of draws, the Tunisians from 3 to 62, the Libyans from 48 to 100 and the Sephardim of Turkey from 0 to 28. There is no version of this measurement that a reader should act on.

The reason is not a defect in the software. It is that the two populations we are being asked to separate were not separate. The Jews of Sefarad and the Jews of the Rhineland and the Jews of Carthage were, in the fourteenth century, variations within a single western diaspora that had left the eastern Mediterranean a millennium earlier and had absorbed comparable amounts of comparable local ancestry along the way. Asking a coordinate file which of them a Moroccan Jew descends from is asking it to resolve a difference of twenty-four units using an instrument whose own residuals are twelve.

None of that makes the historical question unanswerable. It makes it a historical question. The evidence that the megorashim were a large and in places dominant element in the Jewish communities of northern Morocco and coastal Algeria and Tunisia is documentary, onomastic and liturgical: the takkanot of Fez, the survival of Haketia, the Castilian surnames still carried by families from Tetouan and Tangier, the separate synagogues that kept the Castilian rite for two hundred years, and in Tunis the formal split between the Twansa and the Grana, the latter of Livornese and ultimately Iberian origin, which was maintained as a legal division of the community until the twentieth century. That evidence is strong and it is specific. Genetics adds nothing to it, and should not be asked to.

How much would have been visible

A null result is worth nothing unless the method could have produced a positive one, so the floor under the one measurement that does work, the Maghrebi component, was measured rather than assumed.

Synthetic genomes were built by adding a known amount of Berber ancestry to the Sephardic Jewish average of Turkey and pushing each one back through the same four-source model. The baseline is 2.0 per cent. Inject half a per cent and the model reports 2.49; inject one and it reports 2.98; inject three and it reports 4.94; inject five and it reports 6.90; inject ten and it reports 11.80; inject twenty and it reports 21.60. Recovery is linear and close to one to one on top of the baseline throughout.

So the Maghrebi figures of 6.8 to 13.5 per cent in the Maghrebi Jewish communities are real, comfortably above the floor, and would have been recovered accurately if they had been three times larger or three times smaller. A contribution of one per cent would sit on the floor and could not be resolved, which is the appropriate level of caution to apply to the 2.0 per cent in the Sephardim of Turkey and the 0.0 in the Ashkenazim of France: those two figures mean not detectable, not zero.

What can honestly be said

Six statements survive all of the above, and the article is worth reducing to them.

French Jewry is one population by any standard that would be applied to any other population. The distance from one end of it to the other is 28 units; the distance from either end to the country it lives in is between 112 and 126. The internal division that the community itself treats as fundamental is roughly a quarter of the size of the division it does not usually think about at all.

Its Maghrebi majority is not descended from Berber converts. Seven to fourteen per cent Maghrebi ancestry against forty-seven to sixty-five in the Muslim populations of the same countries is not what mass local conversion looks like, and the uniparental evidence says the same thing from two other directions.

Its Maghrebi majority did nonetheless absorb a real Maghrebi layer, measurable, robust to every substitution of the source, and absent from the Ashkenazim of France and from the Sephardim of Turkey. Sixty years in Paris and Marseille have not removed it and will not, because it is the one thing in the French Jewish gene pool that distinguishes the two halves of the community.

The Jews of Sefarad, sampled directly in a mass grave in 1348, were neither converted Iberians nor unmixed Levantines. They sit 84 units from the Christians of medieval Iberia and 91 from the Canaanites of Megiddo, and their nearest relatives on earth are other medieval Jewish communities in England and Germany.

The living population that sits closest to those Jews of Sefarad is the Ashkenazi community of France. That fact says something about the Ashkenazi founder bottleneck and a great deal about the artificiality of the Sephardi and Ashkenazi partition when it is projected backwards past about 1500. It says nothing whatever about who is entitled to the name.

And the proportion of Iberian exile ancestry in a given Maghrebi Jewish family cannot be measured with Global25, or, in this author's view, with any autosomal method currently available, because the populations to be told apart were twenty-four units apart at the moment of separation and the instruments have residuals half that size.

The recurring thesis of this series applies here in a form the reader may find unusually clean. Language is not genes. Neither is liturgy. A French Jew whose grandfather prayed the Sephardic rite in a synagogue in Constantine, whose great-grandmother spoke Judeo-Arabic, and whose family name may or may not have come off a ship from Valencia in 1493, is a Sephardi. That is a statement about a tradition and a family, and it is not made truer or falser by a file of twenty-five numbers.

Coordinates

The pooled construct used as the target in this article, and the four sources of the proximal panel, are given below so that readers can reproduce or contest any figure in it.

The Global25 coordinate system is the work of Davidski at Eurogenes, without which none of this would be possible. The modelling was carried out with the same non-negative least squares approach implemented in Vahaduo by Piotr Kapuscinski. Population averages come from Davidski's published files and from the Moriopoulos 2026 collection. The Tarrega genomes are from Santos and colleagues, Genes, March 2026; the Norwich genomes from Brace and colleagues, Current Biology, 2022.