How Old Norse Patronymics Actually Work

Eriksson is not a surname. It never was one, not for the man who first carried it and not for the thousand generations of "-sson" and "-dóttir" names that followed. It's a sentence folded into a single word: son of Erik. Anyone building Old Norse patronymics for a novel, a campaign, or just out of curiosity runs into the same wall fast — most guides stop at "add -son to the father's name" and leave the actual grammar untouched.

That shortcut works some of the time. It also silently breaks on a large share of real Old Norse names, because the suffix doesn't attach to the name you'd say out loud — it attaches to a grammatical case, and Old Norse had noun classes that changed shape between cases in ways modern English simply doesn't do. This guide walks through how the genitive actually forms the patronymic, why the Viking Age never had hereditary surnames, where matronymics fit in, and where fiction and games quietly get the naming wrong. Along the way we'll be straight about where our own generators simplify the mechanic, and why that's a reasonable trade for a game but not for a lore document.

What a patronymic actually is (and isn't)

A patronymic is a name element built from a parent's given name, marking someone as "child of" that person. In Old Norse it takes two shapes: -son (son) or -dóttir (daughter), attached to the father's name — or, less often, the mother's. Leif Eiríksson is Leif, son of Eiríkr. Freydís Eiríksdóttir is Freydís, daughter of the same man. Nothing here is a family name in the sense English speakers mean it.

A surname is inherited downward, unchanged, across generations — a brand that survives the person who first carried it. A patronymic is recalculated every single generation from whoever your father (or mother) happened to be. Leif's children were never Eiríksson; they took a patronymic built from Leif's own name. That distinction sounds pedantic until you realize it's the single most common naming mistake in Viking-inspired fiction, and we'll come back to it in the fiction-versus-fact table below.

There's a third layer too — the byname, or epithet, like "the Boneless" or "Bloodaxe." It sits alongside the given name and patronymic but works on completely different rules: it's personal, earned, and not passed to children at all. More on that later; for now, the point is that a full Old Norse name could carry three separate, non-overlapping pieces of information, and only one of them behaves anything like a surname.

The real mechanic: genitive case, not just "add -son"

Here's the part almost every article on this topic skips. Old Norse was a case language — nouns changed form depending on their grammatical role in a sentence, the way "he" becomes "him" in English but far more extensively. It had four cases (nominative, accusative, dative, genitive), and the genitive is the one that matters here, because a patronymic suffix doesn't attach to a name's dictionary form. It attaches to that name's genitive stem — the possessive form, roughly equivalent to "Erik's" in English.

For a lot of common masculine names, the genitive stem happens to look almost identical to the nominative plus an "-s," which is exactly why "just add -son" survives as folk wisdom. Eiríkr becomes Eiríks in the genitive, and Eiríks plus son gives you Eiríksson — the rule of thumb works. But Old Norse nouns fell into several declension classes, strong and weak among them, and weak-stem names don't play by that pattern at all. A name ending in weak -i or -a swaps that ending out for -a or -u in the genitive, which means the patronymic built from it looks nothing like "name plus -s plus son." Get the stem wrong and you get a name no Old Norse speaker would recognize.

This is the gap every competing guide leaves open. Runenym and genname.io both reduce the whole system to father's-name-plus-suffix; Nordic Names gets closer, naming the genitive explicitly, but stops short of showing which noun classes actually change shape. None of them give you a table you could use to check a name. So here's one.

How the genitive actually forms — table by declension class

The table below lines up a nominative name, its declension class, the genitive stem that class produces, and the resulting patronymic or matronymic. One row is flagged as genuinely disputed among specialists — treat that one as an example of how the written record gets messier the further back you push, not as settled fact.

Notice the shape of the pattern: strong names mostly just tack on "-s." Weak names swap their final vowel before anything gets added. If you only remember one thing from this table, make it that a weak name's genitive stem is not "the name plus a letter" — it's a different word shape entirely, and the patronymic is built from that different shape.

Worked examples from the sagas

Abstract rules are easier to hold onto with real names attached to them, so here are four that show the mechanic doing actual work.

Snorri Sturluson. The 13th-century historian and author of the Prose Edda is patronymically the son of Sturla Þórðarson. Sturla is a weak masculine name ending in -a; its genitive stem is Sturlu, not "Sturlas" or "Sturlason." That's why the correct, historically attested form is Sturluson — a name you'll see rendered wrong more often than right in casual fantasy writing, precisely because the weak declension is invisible if you only ever apply the "-s" rule.

Leif Eiríksson. Eiríkr is strong, and its genitive is the straightforward Eiríks — so Leif, the son of Eiríkr (Erik the Red, in the more familiar English spelling), is Eiríksson. This is the pattern the shortcut gets right, and it's also the most common declension class among Old Norse masculine names, which is exactly why the shortcut has survived as long as it has.

Freydís Eiríksdóttir. Same father, same genitive stem, different suffix. Freydís is Eiríkr's daughter, so the patronymic is Eiríksdóttir — proof, in one pair of names, that the genitive stem doesn't change with the child's gender. Only the suffix does.

Njáls synir. Njáll, the tragic protagonist of Njáls saga, has a strong genitive stem, Njáls. His sons collectively are referred to as Njáls synir — "Njáll's sons," using the plural noun synir rather than a compound suffix. It's a reminder that Old Norse could express "son of" either as a bound suffix on the individual's name or as a free-standing plural noun for the group, depending on context.

Why there were no family surnames

A patronymic resets every generation, and that's not an accident of the grammar — it's the whole point of the system. A son's name says whose son he is, not which dynasty he belongs to. What carried weight across a Viking Age lifetime wasn't a family brand; it was personal reputation, and the reputation of the specific father or mother a name referenced, refreshed with every new person born.

Hereditary surnames — names that stay fixed across generations regardless of who your father was — arrived in Scandinavia much later, and unevenly. Noble and merchant families in Denmark, Norway, and Sweden began adopting fixed family names in the later medieval period and especially from the early modern era onward, often modeled on continental European practice. Iceland took a different road almost entirely, and kept the patronymic system running long after its Scandinavian neighbors had largely moved to surnames — a difference that still shapes how Icelandic names work today, which the later section on modern Iceland covers in more detail.

Matronymics: when the name came from the mother

A matronymic follows the same grammar as a patronymic — genitive stem plus -son or -dóttir — except it's built from the mother's name instead of the father's. It existed, and there are attested examples in the sagas and other Old Norse sources, but it wasn't the default. Matronymics tended to appear for specific reasons: the father was unknown, dead, or in some kind of disgrace, or the mother was of visibly higher status or better known than the father.

Here's where honesty matters more than a tidy answer. How common matronymics actually were in daily Viking Age life is genuinely debated among scholars, and the debate has a structural cause: most of what survives are sagas, and sagas are overwhelmingly organized around male genealogical lines. A written record skewed that way will always undercount matronymic naming relative to how often it might have occurred in speech, in less formal contexts, or in households the sagas simply weren't interested in narrating. So treat any confident claim about matronymic frequency — including ones on other naming sites — with a raised eyebrow. The honest version is: matronymics were real, attested, and functioned exactly like patronymics grammatically, but exactly how common they were is not something the sources let us pin down precisely.

Our own Norse name generator's everyday category draws on attested names like Kolbeinn, Þórir, Bárðr, Vigdís, Þuríðr, and Ásdís — ordinary homestead names rather than saga-hero names, which is exactly the layer of the naming record where matronymic use is hardest to trace and most likely underrepresented.

Bynames and epithets: the third layer

Alongside a given name and a patronymic, many Old Norse people also carried a byname — an epithet describing something distinctive about them. Bynames cluster into recognizable categories: appearance (the Fair, the Red), character or notable deeds (the Bold, the Peaceful), occupation, place of origin, an animal association, and — often enough that modern readers find it startling — blunt or outright crude humor.

Ivar the Boneless and Eirik Bloodaxe are the two most famous surviving examples, but the written record has plenty of rougher company. A 2025 Medievalists.net roundup of attested Viking nicknames catalogues names built around physical description, reputation, and jokes that clearly landed differently to a Viking Age audience than they read to us now — bynames were not always flattering, and nobody was checking whether they'd offend the recipient.

The critical point for anyone borrowing this for fiction: a byname is not a surname. It was personal, usually earned or assigned by other people rather than chosen for yourself, and it died with the person who carried it. Nobody's children inherited "Bloodaxe." Role and status could shape which flavor of byname someone was likely to pick up — a raider earning something fierce, a jarl earning something dignified, a seer earning something closer to poetic — but the byname itself was never handed down as a family marker.

Iceland today: the system that never died

It's easy to talk about patronymics as a piece of dead grammar from a saga-era past. Iceland makes that framing wrong. Icelanders still use patronymics — and, increasingly, matronymics — as their primary family-name-equivalent, right now, in 2026. A 1925 law effectively froze the creation of new hereditary surnames in Iceland, and a naming committee, the Mannanafnanefnd, has governed permissible personal names ever since, balancing tradition against individual choice.

The system has kept evolving inside that framework rather than staying frozen. In 2019, Iceland introduced the gender-neutral suffix -bur ("child of") as an alternative to -son and -dóttir, for people who don't identify within the binary the older suffixes assume. Icelandic phone directories are still organized by first name rather than family name, because for most Icelanders the "family name" is a patronymic that's different from their father's own patronymic and different again from their children's. This isn't trivia — it's the strongest evidence available that what this guide has been describing isn't an extinct grammatical curiosity. It's a naming system with an unbroken line to the present day.

Fiction vs fact: where fantasy gets Norse naming wrong

Put the mechanic and the social history together and you get a short, useful checklist for anything Viking-flavored in fiction or games. The table below lines up the mistake against what Viking Age practice actually looked like.

Building Norse names for your own characters

Not every name in a novel or a campaign needs a philologist's sign-off. A background character who appears for one scene can reasonably get a simplified patronymic — nobody's going to stop reading over a missed weak-stem vowel shift. Where it's worth slowing down is a named, recurring character, or a scene where the naming itself is doing lore work — a reveal of parentage, a moment where a matronymic signals something about the family, a genealogy a reader might actually check.

It's worth being candid here about what our own tools do, because the honest answer helps more than pretending otherwise. Our Viking name generator's makePatronymic() function takes a father-root and appends "-sson" or "-sdóttir" directly, with no stem change — a deliberate simplification. For strong-declension roots, that lines up with the real historical pattern most of the time. For a weak-stem name, it would miss the vowel shift that a genuinely attested form like Sturluson requires. That's a fine trade-off for a fast, playable name generator; it's not a substitute for the table further up this page if you need a name a specialist wouldn't wince at.

The Norse name generator takes a different approach entirely — it doesn't build patronymics at all. Instead it draws from curated pools of mythic, heroic, and everyday names with real attested meanings attached, which sidesteps the genitive question completely by not generating compound patronymic forms in the first place. And the Skyrim Nord name generator is doing something different again: it's styling names to evoke a specific game's aesthetic, which deliberately drifts from historical Old Norse form in favor of what reads as recognizably "Nord" to players of that game. None of these three approaches is wrong for its purpose. They're just answering different questions — playability, curated authenticity, and game-world flavor — and only one of them is trying to model the actual grammar.

• Quick NPC, background character, one scene: the simplified father-root-plus-suffix form is fine.
• Named protagonist, family lineage that matters to the plot, or anything a reader might fact-check: work from the genitive stem, using the declension table above.
• Deliberately stylized settings (a game, a specific franchise aesthetic): historical accuracy can take a back seat to consistency with that setting's established naming conventions.

FAQ

Is "Eriksson" a surname?

Not in the Viking Age sense. It's a patronymic — "son of Erik" — recalculated for every person based on their own father's name. It only became a fixed, inherited surname later, in places and periods where Scandinavian naming shifted toward continental European convention.

What's the difference between -son and -sson?

Nothing systematic — it mostly comes down to whether the father's genitive stem already ends in a consonant that doubles the "s" sound, and how a given source chooses to spell it. Eiríksson and Ragnarsson both show a doubled "s" because the genitive stem already ends in "s"; a name whose genitive stem ends differently takes a single "-son."

Did Vikings actually have last names?

Not in the modern sense of a fixed name inherited unchanged across generations. What they had was a patronymic (or occasionally matronymic) that changed with every generation, plus an optional byname that was personal and non-hereditary. Neither functions like a surname.

Why does Iceland still use patronymics?

Iceland never fully adopted the hereditary-surname convention that spread through Denmark, Norway, and Sweden. A 1925 law froze the creation of new family surnames, and the patronymic (now also matronymic, and since 2019 gender-neutral via -bur) system has continued as the primary naming convention ever since, regulated by the Mannanafnanefnd.

Can a name have both a patronymic and a matronymic?

Historically, most recorded names carry one or the other, not both — a person is either "child of father" or "child of mother" in the name itself. Some modern Icelanders do choose to carry both a patronymic and a matronymic today, but that's a contemporary practice rather than a documented Viking Age norm.

Is Bjǫrn's genitive really disputed?

The u-stem class that Bjǫrn historically belonged to is one of the messier corners of Old Norse noun morphology, and how it was handled — and later reanalysed — isn't something scholars describe with total agreement. It's a good example of why a single confident rule can't cover every attested name.

Sources and further reading

• Nordic Names — Patronym
• Wikipedia — Icelandic name (1925 law, Mannanafnanefnd, the -bur suffix)
• Wikibooks — Old Norse/Noun Declension
• Wikipedia — Old Norse morphology
• Medievalists.net — "From Butter Penis to Fairhair: 517 Real Viking Nicknames" (2025)
• Medievalists.net — "Viking Nicknames" (2014)
• Landnámabók and Íslendingabók — medieval Icelandic genealogical sources containing many of the attested patronymics and matronymics referenced above, including the Sturluson lineage.