The Moon's shadow is a narrow band, so any one spot can wait centuries for a total eclipse. I computed every path of totality and annularity that crossed Britain and Ireland from AD 1000 to 2200 and tested 17,103 cities, towns and villages one by one. When did yours last go dark in daytime, and when will it next?
On 3 May 1715, at about nine in the morning, London went dark. Edmond Halley had predicted it and printed a map of where the Moon's shadow would fall, and for three and a half minutes the city saw the Sun's corona. Nobody in London has seen that since, and on my calculation nobody will before 2200. In the whole window from AD 1000 to 2200, 3 May 1715 is London's only total eclipse.
That is not bad luck. A total eclipse happens somewhere on Earth every year or two (920 in these 1,200 years), but the Moon's dark shadow is only a narrow strip, so any one place has to wait a long time for it to pass overhead. Averaged over the land of Britain and Ireland, I find a total eclipse every 359 years, close to the figure of about 375 years often given for the world as a whole.
To find out what that means town by town, I used the open-source Astronomy Engine library to compute the positions of the Sun and Moon for every solar eclipse from AD 1000 to 2200 (2,850 of them worldwide), found the 60 whose total or annular shadow touched land in Britain, Ireland, the Isle of Man or the Channel Islands, and tested every city, town and village in the Ordnance Survey's gazetteer, plus the towns of Ireland, Northern Ireland, Man and the Channel Islands, second by second. The calculation reproduces the eclipses that history remembers, from King Henry's eclipse of 1133 to Cornwall's cloudy morning in 1999, with some honest caveats about the Middle Ages and the far future.
Type the name of a city, town or village in Britain, or a town in Ireland, Northern Ireland, the Isle of Man or the Channel Islands. The card gives the last total eclipse before today, the next one, and every eclipse between AD 1000 and 2200 whose central shadow crossed the place. Tap an eclipse to draw its path on the map.
Between AD 1000 and 2200 I count 30 total eclipses whose shadow touched land in Britain and Ireland, 21 of them before today and 9 still to come, plus 30 annular eclipses, when the Moon is too far away to cover the Sun and leaves a burning ring. On average a total eclipse touches land somewhere in the islands every 41 years, but they come in clusters. Halley's eclipse of 1715 was followed only nine years later by another, across Ireland, Wales and the south. Then nothing: no path of totality touched land for 203 years, until the brief eclipse of 29 June 1927. The whole nineteenth century had none. The twenty-second century, by contrast, has 7.
Totality needs the Moon's dark cone, the umbra, to reach the ground, and when it does it covers only a narrow strip that sweeps across the globe in a few hours. Over 1,200 years those strips criss-cross the islands. On my 4 km grid of the land, 98% of Britain and Ireland sees at least one total eclipse between AD 1000 and 2200, and the average spot sees 3.3. That works out at one every 359 years.
Luck still varies a lot. The 1,819 cities and towns in my list average 3.2 totals each. The luckiest, with 8, are Kintore, Moffat, Biggar, Hawick, Langholm and Haltwhistle. At the other end, 15 towns have none at all in the whole window: the corner of Kent around Tenterden, Hythe, Lydd, New Romney and Ashford, and a belt across south Ulster and the border counties of Ireland, among them Omagh, Dromore, Castlederg and Virginia. The longest wait between two totals is 1,011 years, from 1140 to 2151, shared by 68 towns around Manchester and east Lancashire, among them Accrington, Altrincham, Bolton and Bury. The shortest is 2 years: towns in the north-east of Scotland such as Banff, Elgin, Fraserburgh and Huntly saw totality in 1652 and again in 1654.
Most places last saw totality a long time ago. For 981 of the towns (54%) the last total eclipse was Halley's of 1715 or its sequel of 1724. Only 161 saw one in the twentieth century, in 1927 or 1999, and 88 have had none since AD 1000. The median town has waited 311 years. Looking ahead, 660 towns (36%) will see no total eclipse before 2200; for 165 the next comes in 2090.
Each card gives my computed figures for the eclipse, with the history from the Wikipedia articles cited at the foot of the page. Press "Show" to draw the path on the map above.
The Peterborough Chronicle and William of Malmesbury both recorded this eclipse, and it was remembered as an omen: it fell as Henry I left England for Normandy for the last time, shortly before the civil war of the Anarchy. In my calculation totality crossed Scotland and the far north of England: 259 cities and towns, with 4 min 22 s in Edinburgh and 3 min 45 s in Newcastle.
William of Malmesbury saw in it a sign of King Stephen's capture at Lincoln the next year. The Peterborough Chronicle says the day darkened in Lent "when men were eating; and they lighted candles to eat by". My path runs from the south of Ireland across Wales and the Midlands: 685 cities and towns, Derby with 3 min 43 s.
Known in Scotland as Mirk Monday and also called Black Monday (it was 29 March in the calendar then used in Britain). Nicholas Culpeper and William Lilly predicted catastrophe; John Wybard watched totality from Carrickfergus. In my calculation the shadow crossed Ireland, the Isle of Man, the Lake District and Scotland: Dublin had 2 min 38 s, Edinburgh 2 min 50 s, Carrickfergus 1 min 37 s.
Edmond Halley predicted it, drew a map of the shadow's path and watched it from London; his prediction was about four minutes and 30 km out. London had 3 minutes 33 seconds of totality (22 April in the British calendar). It is the only total eclipse in London in my whole window: I get 3 min 21 s.
Nine years after Halley's eclipse, a second path crossed Ireland and southern Britain near sunset, from Galway to Wales, Devon, Hampshire and Sussex, passing south of London. Thomas Simpson of Lincolnshire, then 13, took up astrology after seeing it and later became a noted mathematician. In my calculation 528 cities and towns saw totality, and it was the last total eclipse anywhere in Ireland until 2090.
The first total eclipse on the British mainland for 203 years lasted seconds, early on a cloudy morning. The Astronomer Royal's camp at Giggleswick School saw it; about a quarter of a million people stood on the shore at Southport. I get 26 s at Giggleswick and 25 s at Southport.
The BBC broadcast live from west Cornwall with Patrick Moore, but cloud hid totality almost everywhere on the mainland; the Lizard, around Goonhilly, was one of the few places where it cleared. Ferries in the Channel and the crowd on Alderney did better. I get 1 min 48 s at Plymouth (published: 1 min 42 s).
The next total eclipse on the British mainland comes late in the afternoon of 23 September 2090, close to sunset, across the far south-west of Ireland, Cornwall, Devon, Dorset, Hampshire and the Isle of Wight. It will be Ireland's first since 1724. I get 2 min 18 s at Plymouth and 57 s at Southampton.
A calculation like this is only worth trusting if it reproduces the eclipses people actually saw. Where a duration is published for a named place, mine agree to within about 12 seconds: London in 1715 comes out at 3 min 21 s against 3 minutes 33 seconds, Southport in 1927 at 25 s against an observer's 23, and Plymouth and Alderney in 1999 at 1 min 48 s and 1 min 43 s against 1 min 42 s and 1 min 41 s. London and Dublin come out partial in 1999 and 2090, as published. The sunrise eclipse of 1230, which Wikipedia's list says peaked at 2 min 27 s near Bridlington, peaks in my calculation at 2 min 30 s at Filey, just up the coast.
Wikipedia's list of eclipses visible from the British Isles, based on Sheridan Williams's book of 1996, names 31 total and annular eclipses over the United Kingdom after AD 1000. I find 30 of them. The one I do not, the annular eclipse of 2048, is described there as passing just north of Shetland, which is what I get too. Wikipedia also says that no total eclipse was seen from Ireland between 1724 and 2090; neither does my calculation. I find some the list does not mention: many annular eclipses (the list is not complete for them), totals seen only in Ireland, such as 1679, a substantial total in 1185 across the Hebrides and the north of Scotland (48 towns, among them Inverness and Kirkwall), and a marginal one in 1630 that clipped the far south-west of Ireland and the tip of Cornwall for under 20 seconds.
Some differences remain. For 1424 the list gives almost two and a half minutes; I get 3 min 59 s on Orkney. For 2142 the list puts Dover and Hastings in totality, but my path stays about 22 km offshore. Fred Espenak's current path table for 2142 on EclipseWise also leaves both towns just outside, and the gap between his path and mine is almost all Delta T: he uses 178 seconds for that year, the formula built into my software 308.
One correction was needed to get here. Astronomy Engine's series for the Moon has a slightly smaller long-term slowing of the Moon's mean motion than the lunar theory that the Delta T formulas assume. Left alone, its Moon runs ahead by about 5 minutes of time in 1023 and 24 seconds in 1715, which slid my medieval paths more than a hundred kilometres east. I hold the Moon back by that amount (details in the paper). With the correction, and the same Delta T, my centre lines for 1140 and 2142 lie within about 7 km of Espenak's published ones.
The Earth's rotation is slowing, unevenly, mostly because of the tides. Eclipse calculations run on uniform time, so to know where on the turning Earth a shadow fell you need the difference between uniform time and the clock the Earth keeps, called Delta T. In 1999 it was about a minute (64 seconds in my model). Wikipedia's article gives historical values of about 1,570 seconds in AD 1000 and 200 seconds in 1500, worked out largely from records of ancient and medieval eclipses. My calculation uses the Espenak and Meeus formulas built into Astronomy Engine, which give 932 seconds for 1140.
An error of one minute in Delta T turns the Earth a quarter of a degree under the shadow and slides the path about 16 km east or west at British latitudes. For the eclipse of 1140, shifting Delta T by a minute either way changes the verdict for 32 to 39 of its 685 towns, mostly at the edges of the path. The sunrise eclipse of 1230 is more fragile (49 to 65 of 324), because near sunrise a small shift moves the shadow a long way over the ground. The far future is uncertain too: Delta T for 2090 and 2142 is a forecast, not a measurement, though for 2090 a minute either way changes only 4 to 8 of 175 towns. So read the dates for towns near the edge of a medieval or twenty-second-century path as "probably", and those near the centre line as safe.
Eclipses. I listed all 2,850 solar eclipses from AD 1000 to 2200 with Astronomy Engine (version 2.1.19), screened each on a one-degree test grid over the islands, and scanned the survivors on a grid of 0.04 by 0.06 degrees (about 4 km; 17,921 land cells) and at each of the 17,103 places, first every 30 seconds and then every 2 seconds around the moment of closest approach. At each moment I take the Moon and the Sun (with light-time and aberration) from Astronomy Engine, turn the place's position on the Earth's ellipsoid into the same frame using sidereal time and precession and nutation, and measure its distance from the axis of the Moon's shadow against the radius of the umbra (total) or antumbra (annular) at that point, as Astronomy Engine's own local eclipse code does. A place counts only if the Sun is above its horizon (no refraction). Durations are the time inside the central shadow with the Sun up. I hold the Moon back along its orbit to match the long-term lunar acceleration that the Delta T formulas assume (section 05).
Places. Great Britain: every OS Open Names populated place of type City, Town or Village. Ireland, Northern Ireland, the Isle of Man and the Channel Islands: GeoNames places of 1,000 people or more (522 places; those of 1,500 or more count as towns). "Cities and towns" in the statistics means OS City and Town plus those GeoNames towns: 1,819 places.
Caveats. Delta T is uncertain before about 1600 and after today, so path edges are uncertain by kilometres to tens of kilometres (section 06). The Moon's mean radius is used: mountains and valleys on its limb change durations by a second or two and matter only for the thinnest paths. Weather is ignored: most of these eclipses were probably clouded out somewhere. Dates are in the Julian calendar before 15 October 1582 and the Gregorian after it; Britain and Ireland used the Julian calendar until 1752, so the page also gives the local date. "Today" is 28 October 2026. The eclipse of 12 August 2026 was total in Iceland and Spain but not here.
Every eclipse in the window whose total or annular shadow touched land in Britain, Ireland, the Isle of Man or the Channel Islands, with my computed figures. Dates are in the Julian calendar before 15 October 1582 and the Gregorian after it, as astronomers usually give them; the date in brackets is the date in Britain and Ireland, which kept the Julian calendar until 1752. Tap a row to draw it.