Patronymic and Matronymic Surnames Explained

Johnson, Fitzgerald, Marriott, Tillotson — how fathers (and, more rarely, mothers) gave the world its commonest surnames, what the suffixes actually mean, and how to research a line whose name was once a moving target

Occupational surnames, the subject of our previous guide, freeze a job into a family name. Patronymic surnames freeze a person — the father’s given name, carried by his children. The logic is so natural that most of the English-speaking world’s commonest surnames are exactly this: the genitive -s (“Robert’s” → Roberts), the Norse -son (“Anderson”), the Norman Fitz- (“Fitzgerald”), the Gaelic Mac and Irish O’, the Welsh ap (our Welsh guide covers that system in depth) — and, far more rarely, the matronymic: a name from a mother or another woman, as in Marriott (from Mary), Mabbott (from Mabel) or Tillotson (from Till, a pet form of Matilda).

The consequences are enormous, and they run counter to instinct. Because every man who was ever a father is potentially a surname-founder, patronymic names were coined independently in thousands of places — which is why ten of the thirteen commonest surnames in England & Wales are patronymic-type names, and why “sharing a patronymic surname” is close to meaningless as evidence of kinship. Likewise, a matronymic in your line is a poignant, information-rich clue — but it may record something other than biology, as our case studies show.

As throughout this series, three layers stay separate: meaning (Johnson = John’s son), origin (a specific event, family by family, when the label froze), and family history (the documented descent of your line, provable only in records). See also our earlier pillars: Welsh Surnames: Patronymics, Spelling Changes, and History; Surname Spelling Variations: How to Search Historical Records; Occupational Surnames: How Jobs Became Family Names.

Key takeaways

• A patronymic names a man from his father; a matronymic (also styled metronymic) names a man from his mother or another woman. Iceland still uses the living form — Jónsson/Jónsdóttir — where most other European systems froze theirs.
• “Son of” is encoded differently across Europe: -s and -son (English), Mac/Mc, O’ and Fitz (Celtic and Norman), -ez/-es (Spain/Portugal), -ovich/-wicz (Slavic), ibn/bin/ben (Arabic/Hebrew), and so on.
• English patronymics come in three families: simple genitive -s (Williams, Adams), the Norse -son (northern England), and the pet-form layer — -kin, -cock, -ot, -on diminutives that turned Dawson (Daw/David), Gibson (Gib/Gilbert), Hickinson? (Dick/Richard), Hopkins (Hob/Robert) and Tillotson.
• Matronymics are scarce because medieval record-keeping was patriarchal — but they arise where a woman was the recorded head of a household: widowhood, illegitimacy, an absent father, or a propertied mother.
• The genitive is ambiguous. John Mabbes (1309) may mean “Mabel’s son” — or “Mabel’s servant”. Medieval -s could also be a later addition unrelated to any genitive (McKinley).
• Dictionary etymologies of metronymics have been disproven: the FaNUK project showed Yorkshire Maud/Mawhood is not “Matilda’s son” at all, but a toponym from de Mohaut (Mold, Flintshire).

1. The words, and the two states of a patronymic system

A patronymic (patri- = father, -onym = name) identifies a person by their father’s name; matronymic/metronymic does the same by a mother’s. It helps to distinguish two states of such systems:

• A living patronymic system changes every generation: the Icelandic Jónsson’s son is not Jónsson but, say, Ólafsson, and Jón’s daughter is Jónsdóttir. Iceland keeps exactly this today, as surname scholars from Robert Fowkes onward have noted (“surnames… may not have arrived even now”). The Nordic -son/-sen, -dóttir/-dotter/-datter, and the Gaelic Mac began as live systems too.
• A frozen patronymic system is what most European countries converged on across the medieval and early modern period: at some point, in some place, “Edward’s son” stopped referring to Edward’s current sons and just became the family name. In England, per the Oxford Family Names project, most surnames became hereditary between the thirteenth and sixteenth centuries — with regional lag (south-Lancashire patronymics were still being coined in the early 1600s, as Richard McKinley showed). In Wales, patronymics persisted into the eighteenth and nineteenth centuries in rural areas (see our Welsh guide).

Everything in this article is really about the freeze: which label stuck, why, and what it lets you — and does not let you — conclude.

How “son of” is encoded: a cross-cultural table

A special case worth naming: FitzRoy, “son of the king” — the style borne by acknowledged royal illegitimate sons, such as Henry VIII’s son Henry FitzRoy. Here the “patronymic” is a statement of parentage in the absence of a paternal surname — a bridge to the matronymic logic of §4.

2. The English systems: -s, -son, and the pet-forms

England is unusually rich in patronymic surnames, largely because its surname-freeze coincided with a small stock of very dominant baptismal names (John, William, Robert, Richard, Thomas…). Three formation families account for nearly all of them.

(a) The genitive -s

The simplest: the medieval genitive, anglo-Norman-inflected, attaches to the father’s name — William’s → Williams, Adam’s → Adams, Edward’s → Edwards, Rich’s → Riches, Phillip’s → Phillips, Harry’s → Harris. This is the pattern English and Welsh record-keepers favoured (in Wales, clerks added -s to patronymics wholesale in the Tudor freeze — see our Welsh guide). One warning before you read any -s as “son of”: McKinley’s research, summarised by the FaNUK team, shows that adding an inorganic final -s to already-hereditary surnames became fashionable in the sixteenth and seventeenth centuries, and the Oxford researchers can watch the same surname “vacillating between the zero and the -s form in the same parish”. A Roberts may be Robert + s as a genitive — or Robert with a sixteenth-century flourish.

(b) The Norse -son

The -son suffix descends from the North Germanic genitive that the Scandinavian settlers of the Danelaw used habitually. Surname geographers have long noted its concentration in northern England, the lowland Scottish border zone and East Anglia’s old Danish areas: Anderson, Peterson, Thompson, Robinson, Richardson, Watson (Wat’s son), Allison? (skip). -s and -son are thus two dialects of the same grammatical instinct, often overlapping even within counties — Roberts in the south, Robson, Robinson in the north.

(c) The pet-form layer — the real engine

Medieval English loved diminutives, and they doubled up on given names so freely that the same forename sired five or six different surnames. Peter McClure’s studies (“The Kinship of Jack”, in Nomina) map the suffix system — -kin, -ke, -man, -cot, -cok, -cus — that built diminutive bynames out of pet forms. The mapping below, from surname-dictionary evidence (Reaney & Wilson; FaNUK), is the single most useful tool for researching patronymic lines:

(*Welsh-frozen forms — see our Welsh guide for the apostrophe of ap-contractions there, e.g. Bevan from ab Evan.)

Two things to internalise from this table. First, a family can switch pet forms: a Richardson line may appear as Dickinson, Dixon, Riches or Dickens in earlier documents — because the underlying father-name was always Richard. Second, pet forms create surname look-alikes in both directions — which is exactly where metronymic research goes wrong (§4, case study 1).

3. Matronymics: the rarest common type

Matronymics are scarce for an obvious structural reason: the medieval record-keeper was overwhelmingly interested in men — as landholders, taxpayers, jurymen. A child was “of” his father for most legal purposes. Matronymics arise in the niches where that assumption collapses:

1. Widow-headed households. When a widow was the household’s principal — paying her own poll tax, holding her own tenancy — her sons could be recorded as her dependants, and a genitive like Margaretae could freeze as a surname. (The poll taxes of 1377–81, edited by Carolyn Fenwick, are the standard source where widows appear as principals.)
2. Illegitimacy. Where the father was unnamed, the mother stood alone in the record — the classic origin cited across surname literature. (Compare FitzRoy: parentage declared via the father’s title.)
3. Absent fathers — sailors, soldiers, itinerant workers — leaving the mother as the known parent.
4. Propertied or notable mothers. A woman of property could anchor a family’s identity: “son of Maud” as a status statement.
5. Clerical habit and ambiguity. Some apparent matronymics are simply how a clerk handled a dependent relationship — to which we return below.

The genitive problem: Mabbes, 1309

The FaNUK team’s discussion is worth quoting almost verbatim: in a 1309 Bedfordshire subsidy roll appears John Mabbes — a name that means “Mabel’s son or other relative… or perhaps Mabel’s servant”. The medieval genitive -es marks belonging, not necessarily parentage. A man recorded by his mother’s or mistress’s name might be her son, her widowed mother’s kin, or a dependent in her service. For the researcher, that means: a matronymic is a clue to a relationship, not a birth certificate.

The English metronymic catalogue

The pet-form suffixes of §2(c) work on women’s names too, giving the classic metronymics:

• Dyson — from Dy(e), pet form of Dionisia (Denise; French Denis‘s feminine)
• Tillotson — from Till, pet of Matilda (+ -ot, sometimes -son)
• Ibbotson / Ibbetson — from Ibb(ot), pet of Isabel
• Mabbott, Mabbs — from Mabb, pet of Mabel/Mahalt
• Marriott — from Mary/Marie (+ -ot)
• Emmott / Emmett / Emms — from Emma
• Catlin, Catling — from Cateline, French pet of Catherine
• Tibbs, Tibbett — from Tib, another pet of Isabel
• Nanson, Annison — from Anne/Nan, northern -son
• Annis / Anness — from Agnes

Each deserves the standard cautions: check medieval bearers, check the map, expect competitors (a “Mariot” might be Marie — or something else entirely), and treat the -son and -ot suffixes as evidence of dependency-name rather than proven descent.

4. Polygenesis: what patronymics do to kinship logic

Count the damage. In the ONS National Health Service Central Register survey of England & Wales (1991–2000, analysed by the Guild of One-Name Studies), the top thirteen surnames run: Smith, Jones, Williams, Taylor, Brown, Davies, Evans, Wilson, Thomas, Johnson, Roberts, Robinson, Thompson. The only non-patronymics in that list are Smith, Taylor and Brown — one job, one job, one colour, against ten father-names frozen into family names. Among the next fifty, patronymic-type names continue thickly: Jackson (25th), Harrison (38th), Watson (45th), Griffiths (50th), Richardson, Wilkinson, Simpson, Gibson, Dawson, Sanderson, Dickinson…

The inference is unavoidable and is the standing rule of this series: a patronymic surname was frozen independently in a very large number of unrelated families. Two Richardsons are not kin; two Thompsons in one village are probably not kin. Where you suspect kinship among bearers of a top-100 patronymic, use the tools of our earlier guides — distribution mapping (the Waterer lesson from the occupational pillar), variant searching, DNA and one-name studies — rather than the name itself.

The same logic explains the modern Nordic and Icelandic parallels: because -sen/-son/-dóttir remained live into the nineteenth century in Denmark and Sweden (and because Iceland never froze), the “surname pool” argument is really an argument about freeze dates — freeze early (Wales? no, Wales froze late; norm in England 1250–1450) and names multiply slowly; freeze late (Iceland: still) and the “surname” pool barely exists.

5. Three case studies

1. Maud was not a metronymic — the dictionary revised. Reaney’s dictionary grouped Maud, Mahood, Mawhood, Malt, Moult, Mowat together as metronymics from Matilda (via Medieval English Maud). The Oxford FaNUK team’s distribution work reversed much of it: Yorkshire Maude/Mawhood is an Anglo-French toponymic, from de Mohaut — the hill fort at Mold, Flintshire; Scottish Mowat is a toponym with no connection to Matilda; and the Irish-named Mahood bearers clustered around Liverpool and Glasgow in 1881 are a Gaelic patronymic, MacHood. One innocent-looking cluster of “matronymics” turns out to be a Welsh place, a Scottish place, and an Irish father-name. Lesson: pet-form look-alikes are the easiest surname category to misclassify; etymology must survive distribution.

2. Mabbes (1309): the genitive is a relationship, not a gene. As above — “Mabel’s son or other relative… or perhaps Mabel’s servant.” Lesson: for matronymics especially, prove the relationship in documents before you name a child after a mother who may have been only the household’s head.

3. Fitzpatrick: a Mac wearing a Fitz. Per MacLysaght’s Surnames of Ireland, Fitzpatrick — the only great Fitz- surname of Gaelic origin — is an anglicising of Mac Giolla Phádraig (of Ossory), the Norman-French prefix adopted by Tudor-era record-keepers for a Gaelic Mac name. Lesson: prefixes are not watertight proof of anything; a Fitz- may be a Mac, and (as with the Welsh ap forms) the label’s grammar matters less than the line’s documents.

6. How to research a patronymic or matronymic surname

1. Decode the suffix before you decode anything else. Decide the family: -s (genitive, possibly later addition), -son (Norse north), Mac/Mc/O’/Fitz, ap/ab (→ Welsh guide), -kin/-cock/-ot (pet-forforms), -ez/-ovich/-ian (immigrant origins). Use FaNUK (2016) first, Reaney & Wilson for medieval bearers, Black (1946) for Scotland, MacLysaght for Ireland, Morgan & Morgan and the Rowlands for Wales.
2. Map it. Patronymics are the most polygenetic surname type, so regional distribution is your best lead (Surname Atlas / gbnames profiler, as in our occupational guide). The epicentre of your spelling’s variant set is more informative than the dictionary entry for the bare name.
3. Build the variant list from the forename, not the surname. With Table (§2(c)) in hand, enumerate every pet form of the root: a Richardson line may surface as Dickins(on), Dickens, Rich(es), Hitchcock; a Hobson line as Hopkin(s), Hobby? (careful), Dobbs. Apply the Guild’s variant/deviant discipline from our spelling guide, and search Soundex-style groupings for the common names.
4. Test the -s. If your line’s -s could be a sixteenth/seventeenth-century addition (McKinley’s finding), check earlier spellings in the same parish cluster for the bare form (Robert vs Roberts).
5. For matronymics, prove the mother in the documents. Look for her directly: as a poll-tax principal (Fenwick’s edition), a widow in manor court rolls, a tenant in charters and rentals, a “widow” so styled in parish registers. The surname records an administrative moment — your research recovers the person.
6. When the name tells you nothing, drop it. Pre-freeze patronymic zones (rural Wales, the far north, any illegitimate line) require the cluster method of our spelling pillar: given name + age + place + associates, working sideways through siblings.
7. Use DNA as the sorter, not the name. For the big patronymics especially (Johnson, Wilson, Thompson), surname DNA projects and one-name studies are the honest way to separate the dozens of founding lines — and, where an Y-DNA mismatch turns up, to catch a matronymic concealed inside a patronymic-shaped surname (an Ibbotson line whose Y-chromosome belongs to an unrelated “Smith” is exactly the sort of history the records can then explain).

FAQ

Is Johnson really “John’s son”? Linguistically, yes — but the -s could equally be a later genitive addition, and in any case the name was frozen in thousands of unrelated families. “John’s son” is the name’s meaning, not your genealogy.

Why are there fewer metronymics than patronymics? Because record-keeping tracked men. Matronymics cluster in the niches — widow-headed households, unnamed fathers, absent fathers, propertied women — and in regions where clerks recorded women as principals. Each surviving matronymic is therefore unusually likely to have a specific story behind it.

My surname is Wilson/MacDonald/O’Brien — was the founding father Wilson/Macdonald/O’Brien? The forename under the suffix is usually real (Will, Donald, Bréanán), but the family it founded is one of many; and anglicisation layers (as Fitzpatrick shows) can disguise the original language entirely.

Do living patronymic systems still exist? Iceland’s is the classic case (Jónsson/Jónsdóttir), plus the Gaelic Nic/Ní female forms in Ireland — and, in frozen-form countries, occasional modern revivals (see the Bryn Terfel example in our Welsh guide).

Can a maiden-name surname hide a matronymic? Yes in effect: a child recorded under a widowed or unmarried mother’s surname appears in the records as a matronymic even when the name itself looks like anything else. Treat “unusual parent-name in a patronymic family” as a research question, not an error.

Sources and further reading

• Patrick Hanks, Richard Coates & Peter McClure (eds), Oxford Dictionary of Family Names in Britain and Ireland (OUP, 2016); and the FaNUK methodology paper (Hanks, Coates & McClure) — the Maud/de Mohaut correction, the John Mabbes 1309 discussion, McKinley on inorganic -s, and south-Lancashire patronymic coinage (citing McKinley 1990).
• P.H. Reaney & R.M. Wilson, A Dictionary of English Surnames (3rd edn, 1991) — medieval pet-form evidence; use with FaNUK’s caveats.
• Peter McClure, “The Kinship of Jack” I & II (Nomina 26, 28) — the diminutive suffixes -kin, -ke, -man, -cot, -cok, -cus.
• Edward MacLysaght, The Surnames of Ireland — Fitzpatrick/Mac Giolla Phádraig; George F. Black, The Surnames of Scotland (1946).
• T.J. & Prys Morgan, Welsh Surnames (1985); John & Sheila Rowlands, The Surnames of Wales (1996/2007) — see our Welsh pillar.
• Carolyn C. Fenwick (ed.), The Poll Taxes of 1377, 1379 and 1381 (OUP/British Academy) — widows as principals.
• Guild of One-Name Studies — “Top 500 names in England and Wales” (ONS NHS Central Register survey 1991–2000), and “Variants and deviants”.
• Richard A. McKinley, A History of British Surnames (1990) and county surname series; David Hey, Family Names and Family History (2000); George Redmonds, Turi King & David Hey, Surnames, DNA and Family History (2011).
• Robert A. Fowkes, “Welsh Naming Practices…” (American Name Society, 1978) — on Iceland’s living patronymics and Celtic parallels.
• Cross-links: our pillars Welsh Surnames: Patronymics, Spelling Changes, and History; Surname Spelling Variations; Occupational Surnames.

Individual patronymic and metronymic surnames — Johnson, Roberts, Watson, Gibson, Marriott, Tillotson, Ibbotson and the rest — have their own profiles on Select Surname List, each with the underlying forename’s pet forms, regional distribution and research notes.