How Migration Plants Surnames: The Data Trail of a Name Arriving in a New Country

In France, the surname Traoré was carried by roughly 0.10 people for every 100,000 born in the 1940s. Among people born in the 1990s, that rate is 35.76 — a 358-fold increase across five birth decades. Diallo rose about 96-fold over the same span; Nguyen, roughly fifteenfold. None of these names was created in France. Each arrived with people, and then propagated the way surnames always do — through the children born afterward. What makes this story tellable is that the propagation left a numeric trail, decade by decade, inside an ordinary civil-registry file.

This is not a claim about immigration policy, and it is not a story about people. It is a story about how a surname behaves in a registry once it has entered one, and about the specific, countable footprints that behaviour leaves. Four national statistics offices — in France, Sweden, the Netherlands and Spain — each record a different part of that footprint, and read together they show the same mechanism from four angles.

A surname you can watch arrive

France's national statistics institute, INSEE, publishes a Fichier des noms — a file of surnames tabulated by the decade of birth of the people who carry them, covering births from 1891 to 2000. Most registry files give you a single snapshot: how many people alive today carry each name. INSEE gives you a time axis. That axis is what lets a surname's arrival be watched rather than inferred.

Normalised to a rate per 100,000 people born in each decade, the pattern for several West African and Southeast Asian surnames is unmistakable:

Other Malian and Senegalese names — Cissé, Koné — climb across the same cohorts. Two things are worth noticing. First, Traoré does not merely grow — it grows from near-zero. A name that essentially did not appear on French birth records in the 1940s is, two generations later, more common among newborns than most historically French surnames are among the elderly. Second, the long-established French names are not rising to meet it; Martin, the archetypal French surname, actually declines as a rate, because the denominator — all births in the decade — is being shared among a wider repertoire of names. Nothing is displacing Martin person-for-person. The base of names is simply broadening.

There is an important caveat built into this data, and it runs in a known direction. INSEE's file counts births registered in France and excludes people born abroad — around a tenth of France's population. A first-generation arrival who was born in Mali or Vietnam is not in the file at all; only their French-born children are. So the surnames above are counted through the narrow window of the second generation onward. The real footprint of these names in the living population is therefore larger than the file shows, not smaller. The direction of the undercount is certain; its size is not.

Whole layers that were not there a generation ago

France's file is unusual in having a time axis. Most countries publish only a snapshot — but even a snapshot carries the signature of migration, because a name either appears in the top ranks or it does not, and some names now rank high that a mid-century census had no row for at all.

Sweden's statistics office, SCB, publishes surname frequency counts. Near the top of the list sit the expected patronymics — Johansson, Andersson, Karlsson, each carried by well over a hundred thousand people. Further down, but still high, sit names with a very different history. Ahmed ranks around 56th nationally, carried by roughly 12,280 people; Hassan, Ibrahim and, lower still, Nguyen also fall inside the top few hundred. Independent commentary on the SCB rankings notes the same shape from the other side — the first name of non-Nordic origin now appears well inside the top fifty. These are Somali, Arabic and Vietnamese layers, and a Swedish surname census from the 1950s would not have contained them in any measurable quantity.

The Netherlands shows the cleanest version of the same fact, because a direct before-and-after comparison exists. Dutch genealogical analysis of the surname registry observes that the most common surnames in 2007 that did not appear at all in the equivalent 1947 list are exactly: Yilmaz, Nguyen, Ali, Mohamed and Ahmed. Yilmaz — the single most common surname in Turkey — together with other Turkish-origin names such as Kaya and Öztürk now sits inside the Dutch rankings. Sixty years earlier there was no line for any of them. This is not a name climbing a ladder; it is a rung that did not previously exist.

A registry that measures how long a name has been present

Spain records something the others do not, and it does so by accident of naming law rather than by design. A Spanish person carries two surnames — the first inherited from the father (apellido 1º), the second from the mother (apellido 2º) — and the national institute INE publishes both columns separately.

A surname can only reach the second-surname column once a woman who carries it has had a child registered in Spain, and that child is now themselves counted. So the ratio

ap2 / ap1

measures a structural fact: how far a surname has propagated through the maternal line inside the Spanish registry. It is a count, not an estimate.

The twelve largest surnames in Spain, spanning a sixfold range in size, all land between 0.996 and 1.019 — a tight baseline that makes the low end legible. Singh at 0.070 is not slightly below settled; it is roughly fourteen times below it. Mohamed, by contrast, sits almost at the baseline, because the Spanish cities of Ceuta and Melilla have carried that surname for generations — the ratio reads residency history, not the language of the name.

This measure comes with a mandatory caveat, and omitting it produces a false reading. The ratio only means "generations present" for surnames that entered Spain without a second surname already attached. Countries that themselves use a two-surname system — Brazil, Bolivia, Peru — bring an ap2 with them:

By ratio alone, Mamani (1.210) looks more rooted than García — yet it reflects recent migration. The mechanism that makes the ratio meaningful for Singh simply does not hold for these names. So the honest statement is narrower than "generations in Spain": it is generations in the registry, for surnames that entered without a second surname. Anything more than that overclaims. The ratio knows nothing about any individual, records no ethnicity — Spain does not collect it — and the countries that share its naming law are invisible to it by construction.

How states handed surnames out in the first place

Migration is one way a repertoire of surnames changes. The other, historically, is administrative: a state deciding that a population needs surnames and issuing them. These decrees are the deep background to why some "national" surname stocks look the way they do.

• The Philippines, 1849. Governor-General Narciso Clavería issued a decree, on 21 November 1849, accompanied by the Catálogo alfabético de apellidos — a catalogue of some 61,000 surnames, both Spanish and indigenous, distributed province by province so that families without a fixed surname could be assigned one. Distribution was so systematic that whole towns ended up with surnames beginning with the same letter. It is the reason many Filipino surnames are Spanish, and it explains the concentration at the top of Philippine surname statistics.
• Vietnam. A set of surnames — Thạch, Sơn, Kim, Danh, Lâm among them — was assigned to the Khmer Krom population of the Mekong Delta under the Nguyễn dynasty, as an administrative identification measure. Sources differ on framing, from "bestowed" to "imposed," but agree on the mechanism: a state attaching surnames to a population for the registry's sake.
• Turkey, 1934. The Soyadı Kanunu (Surname Law) required every citizen to adopt a hereditary surname, most of which were newly coined and descriptive. It is why Turkey's entire surname repertoire is under a century old — and why Yilmaz, "undaunted," is the country's single most common name.

These decrees and the migration flows above are two faces of one fact: a country's stock of surnames is not fixed. It is added to — by administrative fiat in one era, by the arrival and settlement of people in another — and each addition is visible in the statistics if the file has the right column.

The mechanism, and its honest limits

Put the four registries together and the same shape appears in each. A surname enters a country with a person. In the first generation it shows a lopsided signature — high on the arrival side of any ledger, near-absent on the second-generation side (high ap1, low ap2 in Spain; present in birth records but not yet in older cohorts in France). As children are born locally, the name propagates, and the lopsidedness closes: the layer "grows into" the statistics. France's time axis shows the growth as it happens; Spain's paired columns show how far it has got; Sweden's and the Netherlands' snapshots show the endpoint — names that now rank where, a census or two ago, there was nothing.

The limits matter as much as the mechanism. Every registry here undercounts the layer it describes, in a known direction: INSEE excludes the foreign-born, so the French rates are floors, not ceilings; Spain's ratio cannot separate recent arrival from a community with few registered births, and is defeated entirely by two-surname origin countries. None of these numbers measures ethnicity, identity, language or citizenship — Spain and France do not record ethnicity by law, and nothing above is derived from any source that does. What the files count is narrow and mechanical: surname positions in a civil registry, and how they change over time — a smaller claim than the headline figures suggest, and the correct one.

Data as of 2026-07-18

All figures were read from the cited national-registry files and recomputed; where a figure comes from published commentary rather than a primary file, it is described as such.

• France — INSEE, Fichier des noms de famille (surnames by decade of birth, 1891–2000). Rates expressed per 100,000 people born in each decade. Note: the file excludes people born abroad, so migration-origin surnames are undercounted in a known direction. <insee.fr/fr/statistiques/353663…;
• Sweden — SCB (Statistics Sweden), name statistics. Surname frequency counts; Ahmed ≈ rank 56 (~12,280 bearers). SCB discontinued name-statistics production in 2024; current lookups are via the Swedish Tax Agency. <scb.se/en/finding-statistics/…;
• Netherlands — surname registry, 1947 vs 2007 comparison as reported in Dutch genealogical analysis: the most common 2007 surnames absent from the 1947 list are Yilmaz, Nguyen, Ali, Mohamed, Ahmed. <dutchgenealogy.nl/popular-dutch-surnames…;
• Spain — INE (Instituto Nacional de Estadística), Frecuencias de apellidos (Censo, as of 01/01/2025). Columns used: Apellido 1º / Total, Apellido 2º / Total. The ap2/ap1 ratio is a structural reading of the registry, not an ethnicity measure — Spain does not collect ethnicity data. <ine.es/daco/daco42/nombyapel/…;
• Philippines — Clavería decree, 21 November 1849, Catálogo alfabético de apellidos (~61,000 surnames). <en.wikipedia.org/wiki/Cat%C3%A1logo_alf…;
• Vietnam — Khmer Krom surnames (Thạch, Sơn, Kim, Danh, Lâm) assigned under the Nguyễn dynasty. <mdpi.com/2313-5778/8/1/16&gt;
• Turkey — Soyadı Kanunu (Surname Law), 1934.

On sensitivity. This article describes registry mechanics and counts only. It makes no claim about individuals, communities, or the merits of migration, and every figure is a count of surname records, not of people's origins.