GBNames allows users to explore the British geography of their family name. Many family groups first settled in different parts of Great Britain and typically spread out over time. Uneven geographies mean that family groups may be concentrated in different types of neighbourhoods. This site uses surnames to introduce users to the SmartCensus, developed by Smart Data Research UK's Geographic Data Service (GeoDS).

GBNames is intended to show interesting geographical patterns in names, rather than mapping individuals or families. We do not map or show statistics for names with fewer than 100 occurrences per year. In total we have statistics for over 40,000 unique surnames; these names account for approximately 78 per cent of Great Britain's adult population - around 52 million people aged 18 and over, estimated from ONS and National Records of Scotland mid-year population estimates (mid-2023).

This page explains, in plain terms, where the data comes from and how the map is built, with links to the underlying research for anyone who wants more detail. Each statistic on a name's results page comes with its own explanation alongside it.

For the Historic Census period (1851-1921), we use digitised census records for England, Wales and Scotland from the Integrated Census Microdata (I-CeM) project. Addresses are linked to the historic parish they fall in; see Higgs and Schürer (2014) for how this was built. Census data for a given year only becomes available for research once it is 100 years old, which is why 1921 is currently the most recent census year we can use. Scotland's census records for 1911 and 1921 are missing from I-CeM.

The SmartCensus is an ongoing project of the Geographic Data Service (GeoDS) to link conventional censuses and surveys with new and exciting 'smart' data sources - providing neighbourhood estimates of circumstances including health, resilience to cost of living pressures and gambling behaviour - in greater geographic detail and with greater update frequency than more conventional statistical sources. Smart data are typically created when we use smartphones and other digital devices to carry out everyday tasks. These data are inherently less reliable than surveys and statistics, and these updates are typically developed for the same neighbourhood areas that are used for the UK Census. We document the quality of our smart and other data on our website (data.geods.ac.uk) and in research publications, including Lansley et al. (2019) and Van Dijk et al. (2021) which describe the core data infrastructure development behind the SmartCensus. We have been updating many of our SmartCensus datasets every year since 1997, allowing researchers to monitor the intricacies of local changes throughout the 21st century.

Visit the website of the Geographic Data Service to find out more about our data products, related maps and how to gain access to our Open, Safeguarded and Secure datasets.

Rather than plotting every single address (which would risk identifying individuals), we use a technique called kernel density estimation (KDE): it smooths out the locations of everyone who shares a name into a continuous surface, showing where the name is more or less concentrated, the same way a heatmap works. If we simply mapped where people with a surname live, almost every map would look the same: bright spots over London, Birmingham and Manchester, because that is where most people live. So we adjust for population: each area is judged partly by how many bearers it has and partly by how unusually common the name is there compared with everyone else. The coloured contours enclose the areas where the name is most concentrated on this adjusted surface. See de Smith, Goodchild and Longley (2018) and Cheshire and Longley (2011) for the technique itself, and Van Dijk and Longley (2020) for how we apply it here.

Because some surnames have far more bearers than others, the shading is always relative to that name's own spread, not a fixed scale used for every name. The three bands each enclose the smallest area holding a set share of that name's own (population-adjusted) presence nationwide, so a name with 150 bearers still gets a full set of three bands, just like a name with 150,000 - what differs is how widely those bands are spread out, not whether they appear at all. Each time period's map is also worked out fresh from that year's own data, so an area moving in or out of the shading between years means the name became comparatively more, or less, concentrated there that year - not that nobody with the name lives there any more.

We only build a map, or show any statistic, for a name with at least 100 people recorded in a given year - never fewer. This is to make sure nobody's individual circumstances can be inferred from the page, and it's also why a search for a relatively rare name may come back with no results at all. For the 1911 and 1921 censuses, Scotland has no data of its own; rather than show a name's map with a blank hole where Scotland should be, a name that is strongly concentrated in Scotland instead shows its 1901 map for those two years, with a note explaining the substitution. Other names keep their own 1911 and 1921 maps, with Scotland shaded grey to show that it has no data for those years.

GBNames uses Google Analytics to understand how the site is used. To find out more about how GeoDS and its partner institutions process data more broadly, see the GeoDS Privacy Notice.

Funded through the Geographic Data Service (GeoDS), a Smart Data Research UK Investment: ES/Z504464/1.