Where Do Last Names Come From? The Origins of Surnames

Most European surnames come from one of four things: an ancestor's job, an ancestor's father, the place he lived, or something about how he looked. Smith worked metal, Johnson was the son of John, Hill lived by a hill, and Brown had brown hair or a brown complexion. That short list accounts for the large majority of hereditary surnames in English.

The part that surprises people is how recent the arrangement is. For most of English history nobody had a fixed surname. A man was Thomas at the mill or Thomas the smith, and his son might be neither. Names became hereditary — passed down unchanged whatever the son did for a living — mainly between the eleventh and fifteenth centuries in England, and in parts of Europe and Asia not until the nineteenth.

The four kinds of surname

Occupational. Smith is the commonest surname in England, the United States and Australia, for the plain reason that every settlement needed someone working metal. The rest of the trade names read like a directory of a medieval town. A cooper made and repaired barrels, from a Low German and Middle Dutch word for a cask. A fletcher made arrows and fitted their flights, from Old French flechier, from flèche, an arrow — the man who made the bow was a Bowyer. A chandler made and sold candles, from Old French chandelier, and only later came to mean a dealer in supplies, as in ship's chandler. Wright is Old English for a maker, which is why it survives mostly in compounds: Cartwright, Wainwright (waggons), Arkwright (chests). Fuller, Walker and Tucker are three regional words for one job, cleaning and thickening woven cloth.

Patronymic. The second group names the father. English did it with a suffix — Johnson, Wilson, Harrison, Dickson — and other languages did it with a prefix. Gaelic mac means son of, so MacDonald is the son of Donald. Irish ó means grandson or descendant of, and O'Brien's family traced itself to Brian Boru, who died in 1014. Welsh ap or ab fused into the following name: ap Hywel became Powell, ap Rhys became Price, ap Richard became Pritchard. Norman French fitz, from fils, gives Fitzgerald. Scandinavian -sen and -son give Petersen and Andersson, Russian possessive -ov and -ev give Ivanov, meaning Ivan's. Armenian -ian, Polish -wicz, Hebrew ben and Arabic ibn do the same work.

Toponymic. The third group names a place, in two flavours. One is landscape: Hill, Brooks, Ford, Wood, Green (the village green), Banks, Underwood. The other is a settlement the family had moved away from, which is the only reason anyone bothered to mention it — Lincoln, Preston, Hamilton, Washington. On the continent the relationship is spelled out with a particle meaning "of" or "from": German von, Dutch van, French de, Italian di and della. One common assumption is worth disposing of: Dutch van is extremely common and carries no rank at all, and German von often signalled nobility but not always. In England the Norman French de was mostly dropped as the language shifted.

Descriptive. The last group is a nickname that stuck. Brown, White, Black and Grey describe hair, complexion or habitual dress. Armstrong means what it says, strong in the arm, and belonged to a formidable Anglo-Scottish border family. Grant comes from French grand, large, and Little and Long are as literal as they look. Russell comes from Old French rousel, a diminutive of rous, red — a red-haired man — which is also what Reid and Read mean in English and what Gough means in Welsh, from coch. Not all of them are kind. Campbell is from Gaelic cam béal, crooked mouth, and Kennedy descends from Irish Ó Cinnéide, a name built on words for head and ugliness.

One caveat applies across all four. Common surnames are polygenetic: they arose independently, many times, in many places. Every village with a forge produced an unrelated Smith. A shared surname is not evidence of shared ancestry.

When surnames became hereditary

Domesday Book, compiled in 1086, is full of bynames and almost empty of surnames: men are identified by trade, father or holding, but the label belongs to the man, not the family.

What changed that was, first, the Norman aristocracy, who had brought the practice of taking a name from a landholding in Normandy and keeping it as the family's identity across generations. Landed inheritance rewards a fixed name, because it makes a claim traceable.

Then the habit spread downwards, and what drove it was paperwork. Manorial rolls, court records, rent books and above all taxation — the poll taxes of 1377, 1379 and 1381 named adults across the country — required clerks to tell one John from another and write the distinction down. Once a byname has been recorded against the same man several times it has stopped being a description and become an identifier, and his son inherits the clerk's version whether or not it still fits. The process ran from roughly the twelfth century to the fifteenth, starting in London and the south-east, finishing last in the north and west, and largely complete among the English by about 1450.

It took far longer elsewhere in Britain. Welsh families kept the living patronymic system for centuries after the English abandoned it, and fixed surnames only became general under English administrative pressure after the Acts of Union of 1536 and 1543, gentry first. That late freeze is why Welsh surnames are so concentrated today: a small pool of given names, locked in place at once, leaving the country full of Joneses, Williamses, Davieses and Evanses. Ireland, by contrast, had hereditary surnames unusually early, from around the tenth and eleventh centuries.

Surnames outside Europe

China has the oldest and most concentrated surname stock in the world. Family names appear in Chinese records going back more than 3,000 years, and the Song-dynasty schoolbook known as the Hundred Family Surnames listed a few hundred, not far off the working total. Wang, Li and Zhang are the most common, each borne by somewhere in the region of 90 to 100 million people, and government estimates put the share of the population covered by the hundred commonest at around 85 per cent.

Japan is the mirror image. Under the Tokugawa shogunate commoners generally could not use a surname in public life. The Meiji government permitted surnames for everyone in 1870 and made them compulsory in 1875, and a population named itself within a few years, frequently taking names from the landscape: Yamamoto is the base of the mountain, Tanaka the middle of the rice field, Kobayashi the small wood. The result is one of the most diverse surname stocks anywhere, with well over 100,000 in use.

Iceland never adopted family names at all. Jón Stefánsson's son is Ólafur Jónsson and his daughter Anna Jónsdóttir; a wife keeps her own name; the telephone directory is alphabetised by first name. A law of 1925 stopped the adoption of new family names, and names formed from the mother are now accepted.

Spain gives each person two surnames, traditionally the father's first followed by the mother's, and a woman does not change her name on marriage; since 1999 parents have been able to choose the order. Portuguese practice reverses it and often runs to more than two. Turkey had no hereditary surnames at all until the Surname Law of 1934 obliged every citizen to take one.

And plenty of people have no surname. A single name remains common in Indonesia, particularly among Javanese — Sukarno and Suharto each had one. In much of south India heritable family names are not traditional, and a name combines a village or house name with a father's name and a given name, often abbreviated to initials. Tibetan naming likewise has no family name.

Names that were forced or chosen under pressure

Ellis Island did not change anyone's name. This is one of the most durable family stories in the United States, and the mechanism makes it impossible. Passenger manifests were written by the shipping company at the port of departure, before the ship sailed. Inspectors at Ellis Island worked from those manifests, checking arrivals against a list that already existed; they were not writing names down and had no authority to alter them. The inspection force also included a large body of interpreters, many of them immigrants themselves. Historians who have worked through the records — Vincent Cannato's American Passage is the standard account — find no documented case of the inspector-renaming story. Name changes were real and common, but happened afterwards and mostly by choice: a spelling simplified for an employer, a name anglicised at school, a change formalised years later in naturalisation papers.

Surnames imposed by decree. Ashkenazi Jewish communities across central and eastern Europe largely used patronymics rather than fixed family names until governments required otherwise, for reasons of taxation, conscription and census. Joseph II's edict of 1787 in the Habsburg lands, the Napoleonic decree of 1808, the Prussian edict of 1812 and Russian statutes from 1804 and 1835 all compelled the adoption of permanent surnames. Many of the resulting names were chosen by the families themselves, including the ornamental German compounds such as Rosenberg and Goldstein, alongside occupational and place-based names and the retained priestly designations Cohen and Levi. Some were assigned by registrars. The frequently repeated story that officials handed out deliberately humiliating names to families who could not pay a bribe circulates everywhere, but researchers who have worked systematically through the registers, notably Alexander Beider, find the documentary evidence for it thin and the practice far rarer than the anecdotes imply.

Emancipation in the United States. Under slavery, enslaved people were recorded in most documents by a single given name, when recorded by name at all, though many used surnames among themselves long before emancipation. After 1865 those names could be used and registered openly. The choices varied: a name already in family use, the name of an earlier owner from whom relatives had been separated and might be traced, a refusal of the name of the person who had held them, or a name carrying the new status, such as Freeman or Washington. This matters for research. The census of 1870 is the first US federal census to record most formerly enslaved people by name, which is why African American genealogy so often meets a wall at that year and has to continue through Freedmen's Bureau records, probate inventories and plantation papers.

How to research your own surname

Work from records and treat meanings as the last step, not the first.

Civil registration gives you births, marriages and deaths from 1837 in England and Wales and from 1855 in Scotland, and each certificate names the previous generation. Census returns put a household together on one night with ages, birthplaces and occupations: the UK has them every ten years from 1841, with 1921 the most recent released, and the US from 1790, with 1950 opened in 2022. Parish registers carry baptisms, marriages and burials back to 1538 in England. Immigration records are useful because naturalisation papers often record both the original name and the adopted one.

Ignore spelling. Before general literacy the clerk wrote what he heard, and one family can appear as Shepherd, Sheppard, Shephard and Shepard within three generations. Search by sound.

Surname-meaning websites are unreliable for three reasons. They hand you one origin for a name that usually has several unrelated ones. They are very often attached to a business selling certificates or heraldic prints. And they blur the difference between the etymology of a word and the history of a family. For Britain and Ireland the serious reference is the Oxford Dictionary of Family Names in Britain and Ireland (2016), built on record evidence for roughly 45,000 names rather than on folklore.

The point to hold on to: a surname's etymology describes one ancestor at one moment, probably before 1400, and nothing about anyone since. Every Smith is descended from a smith. They are not descended from the same smith. Surname DNA projects test whether two lines sharing a name share a paternal ancestor, and frequently show that they do not.

Frequently asked questions

Where do last names come from?

Most European surnames come from one of four sources: an ancestor's occupation (Smith, Baker, Cooper), an ancestor's father (Johnson, MacDonald, Petersen), a place (Hill, Lincoln, van Dijk) or a personal characteristic (Brown, Little, Armstrong). They began as informal bynames distinguishing two people with the same given name, and hardened into inherited names as written administration spread — mainly between the twelfth and fifteenth centuries in England, and far more recently elsewhere.

What is the oldest surname?

There is no clean answer, because "oldest" can mean the earliest recorded name-word, the earliest hereditary use, or the earliest line of descent anyone can still document, and those give different winners. Chinese surnames are the usual claim, attested in records over 3,000 years old, although the system as used today consolidated later and unbroken documented descent that far back is very rare. Katz and Cohen are also commonly cited, but Cohen worked as an inherited priestly designation long before it operated as a surname in the modern sense. In Europe, the Irish hereditary surnames of the tenth and eleventh centuries, such as O'Brien, are among the earliest traceable as family names.

Why do some surnames end in -son?

It is a Germanic patronymic construction meaning "son of" whoever is named in the first half: Johnson, Wilson, Harrison, Jackson. Scandinavian languages used the same endings, -son and -sen, but kept them as live patronymics changing with every generation until nineteenth-century laws froze them into inherited family names, which is why Scandinavian surname stocks are so narrow. Iceland never made that change and still forms names afresh each generation.

Can I find out what my last name means?

Usually yes, if you use a scholarly dictionary rather than a commercial surname site — but the meaning of a word and the history of a family are two different questions. Because common surnames arose independently in many places, knowing that Cooper meant a barrel-maker tells you nothing about which barrel-maker, or whether the man who first carried the name into your line practised the trade at all. For your own family the answer is in civil registration, census and parish records, not in the etymology.

Where to go next

Surnames show how ordinary history behaves under inspection. Something that feels ancestral turns out to be an administrative convenience most families acquired within the last thirty generations, often from a clerk. The etymology is the easy half; the interesting half is why anybody needed fixed names, and who benefited from writing them down.

If you want more of that kind of detail, our 10 fascinating history stories collection is a good place to begin, and the best history books for beginners covers where to go if you would rather read at length.

Chunks publishes one free five-minute story every day — written by people, narrated properly, on iOS and Android.