Britain is not supposed to have earthquakes. It has thousands. I mapped every one in the British Geological Survey's two catalogues, from the Dover Straits shock of 1382 to 2024, to ask where they happen, how big they get, how often a big one comes, and what the biggest has been near you.
At about six o'clock in the evening of 6 April 1580, in Easter week, the ground shook in London, across south-east England and over the sea in Flanders and northern France. Two people were killed in England. The British Geological Survey puts the shock in the Strait of Dover, at magnitude 5.8, and it is still one of the biggest earthquakes ever to hit England. A Briton who felt it would not have called it rare: the same straits had shaken in 1382, interrupting a church council in London that became known as the Earthquake Synod.
Britain sits in the middle of a tectonic plate, far from the plate edges where the world's great earthquakes happen. Its earthquakes are small. But they are not rare. The BGS catalogues list 471 known earthquakes from 1382 to 1969, worked out from letters, newspapers and diaries describing what people felt, and 6,561 recorded by seismometers from 1970 to 2 September 2024. Most are far too small to feel. A few have killed people, toppled chimneys or twisted church spires.
This study puts both catalogues on one map. It asks four questions: where Britain's earthquakes cluster, how often a big one (magnitude 4 or 5) comes round, how much the record is shaped by our ability to detect small shocks, and how many of the recorded shocks were set off by coal mining. And it lets you look up the biggest earthquake ever recorded near your town.
Each circle is one earthquake at the epicentre the BGS gives, sized by magnitude. Sepia circles are historical earthquakes (before 1970), located from felt reports; red circles are instrumentally recorded ones (1970 on). Type a town or village to find the biggest earthquake recorded within the distance you choose, or drag the slider to watch the record build up.
Magnitude is a logarithmic scale: each whole step means about ten times more ground motion and about thirty times more energy. In Britain a magnitude 3 is felt by people near the epicentre; a magnitude 4 rattles windows and cups over a wide area; a magnitude 5 is felt across much of the country and brings down chimneys near the epicentre. The catalogues hold 19 earthquakes of magnitude 5 or more on land in Britain or within 20 km of its coast, and 105 of magnitude 4 or more.
The biggest of all is offshore: the Dogger Bank earthquake of 7 June 1931, magnitude 6.1, about 104 km from the nearest British town, Withernsea on the Yorkshire coast. Close to the coast, the biggest is the Dover Straits shock of 1580 (5.8). On land, three historical shocks share the record at 5.3: Caernarfon in 1852, Hereford in 1896 and Derby in 1957. The biggest instrumentally recorded are Market Rasen in 2008 and the Llŷn Peninsula in 1984, both 5.2 in the BGS catalogue, which places the Llŷn epicentre just off the coast. Tap a card to see it on the map.
Earlier chronicles describe bigger events still, such as the earthquake of 1185 that is said to have damaged Lincoln Cathedral and the one of 11 September 1275 that destroyed the church of St Michael on Glastonbury Tor. These are not in the BGS catalogue, which starts in 1382, and medieval writers sometimes blamed earthquakes for buildings that simply fell down.
Britain's earthquakes are not spread evenly. The south and east of England are quiet: east of 1° west and south of 53° north lies 17% of Britain's land but only 3% of its recorded earthquakes. The west is busier: north and south Wales, the Welsh borders, Cumbria, the Scottish Highlands and the west coast of Scotland. The ten busiest 20 km squares on the map hold 27% of all the instrumentally recorded earthquakes in Britain.
Two kinds of place top the list. One is the natural hotspots. The Llŷn Peninsula is the busiest in Wales: the magnitude 5.2 earthquake of 1984 was followed by 56 more recorded earthquakes within 30 km in the next year, most of them aftershocks, and the historical catalogue lists 28 earthquakes within 20 km of Pwllheli from 1534 on. Comrie in Perthshire, close to the Highland Boundary Fault, is the most famous: 30 historical earthquakes within 10 km of the village, 19 of them in the swarm of 1839 to 1846, the largest of magnitude 4.8 on 23 October 1839. The village is sometimes called the "Shaky Toun", and its Earthquake House, built in 1874 to hold an early seismometer, still stands.
The other kind is more surprising. The four busiest squares in Britain are not on any famous fault. They are Midlothian, around Bonnyrigg and Dalkeith; the Potteries, around Kidsgrove and Stoke-on-Trent; Clackmannanshire; and north Nottinghamshire, around Ollerton and Market Warsop. They are all coalfields.
When coal is removed from deep underground, the rock above and around the workings settles and the stresses shift. Small earthquakes can follow. They are shallow, which makes them easy to feel at the surface even when they are small. The catalogue does not label which earthquakes were induced, but the pattern is plain. Of the 3,616 onshore earthquakes recorded from 1970 to 2014, 1,460 (40%) had epicentres within 5 km of a coal mine mapped in OpenStreetMap, even though only 10% of Britain's land lies that close to one.
Three more clues point the same way. The near-mine earthquakes were shallow: a median depth of 1.7 km, against 7.4 km for the rest. They kept to the working week: only 19% of the near-mine earthquakes up to 2014 fell on a Saturday or Sunday, against 27% of other onshore earthquakes and the 29% you would expect by chance. And they stopped when the mining stopped. After the last deep mines closed in 2015, the share of onshore earthquakes near a coal mine fell to 17% in 2015 to 2024, and to 15% since 2020.
The busiest earthquake zones in modern Britain were made by people: the coalfields of Midlothian, the Potteries and Nottinghamshire.
Fracking near Blackpool also set off small earthquakes: a magnitude 2.3 at Preese Hall in 2011 and a 2.9 near Little Plumpton in 2019, after which the government announced an indefinite suspension of fracking. Neither appears in the catalogue layer I used, which has no events within 10 km of those wells in 2011 or in 2018 and 2019, so induced events may be handled separately by the BGS. I report them from published accounts, not from these data.
Seismologists describe how often earthquakes of different sizes happen with the Gutenberg-Richter law: for every step up in magnitude, earthquakes become about ten times rarer. The slope of that relationship is called the b-value. For the 802 earthquakes of magnitude 2.0 or more recorded in Britain from 1980 to 2 September 2024, I estimate b = 0.92 (95% interval 0.86 to 0.99), close to the global norm of about 1.
The fit says Britain should expect about 2.2 earthquakes of magnitude 3 or more a year (the catalogue recorded 2.0 a year), a magnitude 4 or more about every 3.9 years (95% interval 2.9 to 5.2) and a magnitude 5 or more about every 32 years (22 to 50). The long record agrees well for magnitude 4: the catalogues hold 91 magnitude 4 or bigger earthquakes in Britain since 1750, one every 3.0 years. But it holds 15 of magnitude 5 or more, one every 18 years, more than the fit predicts. Either recent decades have been quiet for big earthquakes, or the magnitudes estimated from felt reports for historical earthquakes run a little high, or both. With so few big events, nobody can say which.
The record grows over time, and almost all of that growth is in what we can detect, not in what the Earth is doing. Before 1700 the historical catalogue includes only earthquakes of magnitude 4 or more, and only 19 of them for Britain survive in the record. In the nineteenth century, newspapers and amateur observers caught many more small shocks: 213 in the 1800s. From 1970 seismometers took over, and the network kept growing.
In the 1970s, 69% of the instrumentally recorded earthquakes in Britain were smaller than magnitude 2; by the 1990s it was 87%. The number below magnitude 2 rose 3.2 times, from 298 in the 1970s to 952 in the 1980s, as stations were added (and as the coalfields were at their busiest). The number of magnitude 3 or bigger, which the network could catch throughout, did not rise: 45 in the 1970s, 19 in the 2010s. So a map of every recorded earthquake partly maps where we were listening, and when.
For each of the 1,426 cities and towns in Ordnance Survey's gazetteer I found the biggest earthquake in either catalogue within 25 km. 41% of them (582) have had one of magnitude 4 or more that close, and 125 have had a magnitude 5. Only 46 have no recorded earthquake within 25 km at all, among them Aberdeen and St Albans. London's biggest within 25 km was a magnitude 3.1 in 1750, one of two London shocks a month apart, in a year with six earthquakes in the catalogue.
Historical earthquakes are reconstructions. For events before 1970 the BGS works out a location and magnitude from where and how strongly the shaking was felt. They can be tens of kilometres out, and the magnitudes are estimates. Where the catalogue gives an instrumental magnitude (ML) I use it; otherwise its macroseismic magnitude. Five historical events have no magnitude. The catalogue's own magnitudes can differ from those quoted elsewhere: it gives the 1984 Llŷn earthquake 5.2, where many accounts say 5.4.
Detection changes over time. The historical catalogue is selective (before 1700, magnitude 4 or more only), and the instrumental one gets better at catching small events decade by decade. The layer I used starts at magnitude 1.0. Counts of small earthquakes say as much about the seismometer network as about the Earth. The Gutenberg-Richter fit uses only magnitude 2.0 and above from 1980, where the b-value is stable.
Induced earthquakes are not labelled. I infer mining-induced seismicity from closeness to mapped coal mines, depth and timing. OpenStreetMap's coal mines are incomplete and include old shallow pits, so this is an indicator, not a classification. The fracking-induced earthquakes near Blackpool are not in the layer at all.
Cleaning. The modern layer contained 7,834 rows; 1,273 were duplicates (the same event entered more than once, often once by an automatic system and once by an analyst), leaving 6,561 events. "Britain" here means epicentres on land in Great Britain or within 20 km of its coast; the catalogues also include 1,476 events further offshore (942 of them in the northern North Sea, north of 58° N), and 32 in Ireland.
Earthquakes. I downloaded the BGS historical catalogue (all known UK earthquakes before 1970, 471 events) and the modern catalogue (instrumentally recorded since 1970) from the BGS map services, merged them, removed duplicate rows and placed each epicentre: on land in Great Britain (Natural Earth outline), within 20 km of the coast, in Ireland, on the continent or further offshore. Each event was given its nearest OS Open Names town and settlement.
Statistics. b-value by Aki's maximum likelihood estimator with Utsu's correction for magnitudes rounded to 0.1; the completeness magnitude (2.0) is where the estimate stops changing as the threshold rises; intervals by bootstrap. Return periods are the inverse of the fitted annual rate. Weekend test: binomial test against 2/7.
Coal mines. OpenStreetMap features whose name mentions a colliery, coal mine or pit, or tagged with coal as the resource; distance from each epicentre to the nearest. Land share from a 2 km grid of points on British land.