Where notable people with each surname were born in Great Britain: 39,206 born in the 1800s against 48,944 born since 1950, from Wikidata. Search your surname and watch it spread.
A surname is a small piece of geography. Many British surnames were fixed in the Middle Ages and stayed close to where they began for centuries: Joneses in Wales, Tre- names in Cornwall, a family named after its village living a few miles from it. Railways, cities and two world wars shuffled the deck. How much? Without open census data, I cannot count everyone, but I can follow a well-documented subset: the people Wikidata records.
I took every person in Wikidata with a family name, a birth year from 1700 and a birthplace in Great Britain: 133,761 people with 22,041 different surnames, born in 10,260 places. Then I compared where each surname's bearers were born in the 1800s with where they were born from 1950 on. The map lets you search any surname with at least two notable births.
Wikidata is built by volunteers from encyclopaedias, sports databases, parliamentary records, peerages and museum catalogues. Who gets an entry follows who was recorded. Of the 133,761 people here, 7,939 were born in the 1700s, 39,206 in the 1800s, 37,672 between 1900 and 1949 and 48,944 from 1950 on. Most are men: women are 14% of those born in the 1800s and 22% of those born since 1950.
London is the giant: 24% of notable people born in the 1800s were born there, and 21% of those born since 1950. Scotland's share falls from 17% to 12%. These are shares of the notable, so a surname that "moves to London" on the map may only be following where notable people tended to be born. Every comparison below allows for that: a surname counts as regional only when its births cluster more than notable births in general.
To measure how regional a surname is, I asked what share of its notable births would have to move region for its spread to match all notable births in the same era (a dissimilarity index). Small samples look lumpy by chance, so I compared each surname with random draws of the same size from all births, used the same sample size in both eras, and kept only the 349 surnames with at least 15 notable births in both the 1800s and since 1950.
The typical common surname was 48% less regional for people born since 1950 than for those born in the 1800s.
For the median surname, the share of births beyond what chance would give fell from 13.7 percentage points to 7.1. 79% of the 349 surnames became less regional. The region where a surname was most over-represented stayed the same for only 42% of them, and the median surname's centre of gravity moved 46 km, against 11 km for all notable births together.
The most regional names in the 1800s were Welsh and Scottish: Rees, Davies, Owen and Jones in Wales; McKenzie, Reid, McLean and Cameron in Scotland. Most stayed regional but much less so. A few held on or grew: Paterson, Thomson and Robertson in Scotland, Robson in the North East. Names of England's industrial north, such as Parkinson, also stayed put more than most.
Most Welsh surnames come from patronymics: Evan ap Thomas became Evan Thomas, and the "son of" names with an added s (Jones, Davies, Roberts, Edwards) or a fused ap (Powell from ap Hywel, Price from ap Rhys, Pugh, Pritchard, Bowen) became some of the commonest names in Britain. I took the 14 patronymic surnames named in Wikipedia's article on Welsh surnames.
In the 1800s, 65% of notable people with these names were born in Wales, against 8% of all notable people. For those born since 1950 the figure is 35%, against 6%. Still five or six times the national rate, but most notable Joneses, Davieses and Evanses are now born in England, led by London (329 since 1950) and the North West (314). Jones alone has 1,768 notable births in the data, more than any other surname.
"By Tre, Pol and Pen shall ye know all Cornishmen", runs the rhyme recorded by Richard Carew in 1602: tre a homestead, pol a pool, pen a headland. Do notable bearers of Tre-, Pol- and Pen- surnames come from Cornwall? I took every surname in the data that starts with those letters (76 Tre-, 31 Pol- and 48 Pen- names). That catches false friends (Trevor is Welsh, Pennington is a Lancashire place name, Penn and Penny are English), so this is a generous test.
Cornwall is small in this data: only 826 notable births in all, under 1% in every era. Against that, 7.4% of 1800s notables with Tre-, Pol- or Pen- surnames were born in Cornwall, about 9 times the rate for everyone, and 2.0% of those born since 1950. The rhyme points the right way, but most notable Trevelyans, Trefusises and Penroses were born outside the county, often into landed or naval families with homes elsewhere.
Many English surnames began as a place: someone who came from Haigh, Heywood or Warburton. I matched every surname to the cities, towns and villages in OS Open Names with exactly the same spelling, keeping names with at most three namesakes so that the place is identifiable. That gives 1,410 surnames. A spelling match is not proof of origin (Wilson, Green and Moore are also village names), so this is a test of tendency, not of any single name.
In the 1800s, 4.5% of notable people with a place-name surname were born within 25 km of a place of that name, 2.9 times what chance would give. For those born since 1950 it is 3.9%, 2.0 times. The pull of the namesake has weakened but not gone. The clearest homing names are northern: of 7 notable Haighs born in the 1800s, 4 were born within 25 km of the village of Haigh near Barnsley; of 16 Leeses, 6 near Lees by Oldham or Lees in Derbyshire. Since 1950 none of the Haighs and none of the Leeses in the data were.
My surname is a good warning about this data. Hotham is a village in the East Riding of Yorkshire, population 233 at the 2011 census. Its Wikipedia article records that it was once the land of Sir John Hotham, 1st Baronet, Governor of Hull, and that the Hotham family estate is now at nearby South Dalton. A surname that matches a village a few miles from the family seat is the classic shape of a place-name surname, though I have not traced its origin here. Wikipedia's page for the name lists 14 people called Hotham. But only 2 notable Hothams in Wikidata have both a family name and a birthplace in Great Britain from 1700 on. Others were born abroad or before 1700, or their entries lack a family name or a birthplace.
Two points cannot make a pattern. 67% of all surnames in the data have two notable births or fewer, so most people searching the map will meet the same thinness for their own name. That is the honest limit of an open, notable-only source: it is rich for Jones, Smith and Campbell and thin for most names. Show Hotham on the map.
One way to see how far notable people mirror everyone is to compare them with real counts. National Records of Scotland publishes the surnames that appear most often in Scotland's birth, marriage and death registers each year, under the Open Government Licence. For the top 100 surnames of 2025, I compared the number of register entries with the number of notable people with that surname born in Scotland since 1950.
The ranks agree well: a Spearman correlation of 0.85 (0.77 for notables born in the 1800s). Smith leads both lists. So for common names, notable people track the population better than you might fear. The differences are telling: some names are over-represented among the notable (McKenzie, Kerr, Gibson) and some under-represented (Johnstone has none, Dickson, Green and Hill few), and spelling matters, because Wikidata and the registers do not always agree on it.
People. Wikidata, queried on 29 October 2026 through the QLever SPARQL endpoint: every human with a family name (property P734), a place of birth (P19) with coordinates, and a date of birth in 1700 or later, where the birthplace's country is the United Kingdom or one of its predecessor states. That gave 141,273 people. A birthplace counts as in Great Britain if an OS Open Names settlement lies within 10 km of it; this drops Northern Ireland, the Isle of Man and the Channel Islands, which OS Open Names does not cover, and leaves 133,761 people. Each birthplace takes its region and county from the nearest OS Open Names settlement. Surnames are Wikidata's English labels of the family-name items, so spellings are kept apart (Thomson and Thompson, Mcdonald and McDonald).
Measures. Regional clustering is the dissimilarity index between a surname's births across the 11 regions and nations and all notable births in the same era, averaged over 200 random subsamples of the same size in both eras (at most 60), minus the average index for random samples of all notable births of that size. The centre of gravity is the mean position of births. Signature surnames use the location quotient: a surname's share of a region's births divided by its share of all births. Namesake distances are to the nearest OS Open Names city, town or village with exactly the surname's spelling, compared with a random sample of 8,000 notable births per era.
Limits. This is not a census. Notability is skewed by sex, class, occupation and place, and by what volunteers have entered: Wikidata coverage of sportspeople and politicians born since 1950 is far better than of farmers born in 1820. A birthplace is often a town or city rather than a parish, and "London" is one point for millions of births. People with two family names count under each. Spelling variants are separate surnames. The 1881 census, which would answer the original question for everyone, is held by the UK Data Service with surnames only under a special licence, and is not used here.
The counts of notable births by surname, era and region, the regional measures for each surname, the place-name and Scottish comparisons and all code are in the GitHub repository and archived with the paper at doi.org/10.5281/zenodo.23761427. Only aggregate counts are published, never lists of people.
Related: What is your town's name made of? on the old words inside place names, and High Street or Main Street? on street names.