The holy and wishing wells of Britain and Ireland. I mapped 3,440 holy wells, read the saint and the language in their names, and counted the cures, rags and offerings recorded at Ireland's wells.
Throw a coin into a well and make a wish. Most of us have done it, at a garden centre or a shopping arcade, without thinking about where the habit came from. For a thousand years and more, people in Britain and Ireland came to springs named for saints, walked rounds, prayed, drank, bathed sore eyes, and left something behind: a rag tied to a thorn tree, a pin, a medal, a coin. This page follows that thread with data, from the saint's well to the wishing well.
All numbers are computed in the code that comes with this page; see Methods for how each was made and what it cannot tell you.
Each dot is a well. In the Republic of Ireland the dots are the National Monuments Service's 3,025 records of holy wells, the state's own inventory. In Northern Ireland, Britain and the Isle of Man they are wells that volunteers have mapped in OpenStreetMap or entered in Wikidata, filtered from 66,517 loosely matching names down to 415 holy wells and 1,107 other named wells. Colour the dots by saint, by the language of the name, or by what the Irish descriptions record. Tap a dot to read its record.
The Republic of Ireland's inventory holds 3,025 holy wells, 43.6 for every 1,000 km². Cork has the most (358); for its size, Dublin (120.7 per 1,000 km²), Kilkenny, Wexford and Clare are the densest counties, and Monaghan the sparsest (6.3).
Across the border the picture collapses, and that is the first lesson of this project. Northern Ireland shows 2.7 wells per 1,000 km², about one sixteenth of the Republic's (15.9 times lower). The wells did not stop at the border; the record did. North of it I only have what volunteers have mapped. Even within OpenStreetMap alone, the Republic has 5.5 times Northern Ireland's density (13.8 against 2.5), which suggests that Irish mappers have used the official inventory; 36% of the inventory's wells have an OpenStreetMap well within 150 m.
So Britain's 373 wells (1.6 per 1,000 km²) are a floor, not a count, and I do not compare them with Ireland as if they were. Within Britain, the open data still find the places you would expect: Cornwall has 57 mapped holy wells, 16.0 per 1,000 km², far ahead of any other county, and the north Welsh counties of Clwyd and Gwynedd follow.
A well's name often says whose it is: St Patrick's Well, Toberbride, Ffynnon Fair, Lady Well. I read a saint only where the name says so plainly, in English ("St X's Well"), in Irish or Gaelic (Tobar Phádraig, Tobar Mhuire) or in Welsh (Ffynnon Fair is Mary's well, Ffynnon Ddewi David's). For a few anglicised Irish forms, such as Tobermurry and Toberaraght, I used them only because the inventory's own descriptions give the saint (Muire, St Attracta). Anything else I leave unread rather than guess.
Of the Irish inventory's 2,142 named wells, 1,007 (47%) name a saint, 245 are simply holy (Holy Well, Trinity Well, Sunday's Well) and 890 (42%) I cannot read: Toberinneenboy, Tobergal and hundreds of other Tober names whose meaning I have not checked. Among the wells with a saint, Mary (162, 16%) and Patrick (161, 16%) are level, and Brigid is third (113, 11%). Then come John, Colmcille, Brendan and Ciarán, and a long tail of local founders: the 189 different names include Mac Duagh, Attracta, Gobnait and Moling.
In Britain, Mary leads in England (33 of 167), Wales (12 of 49) and Scotland (14 of 63). Below her the lists turn local: Helen and Anne in England, Cybi, Teilo, Winifred and Beuno in Wales, Ninian and Kentigern in Scotland. Patrick leads in Northern Ireland, with 7 of the 28 wells there that name a saint.
In my earlier map of saints and holy wells across Western Europe, Brigid held 23.1% of named Irish holy wells. That figure came from the 1,021 Irish holy wells in Wikidata, of which 515 had a dedication read from their labels. Here I use the full official inventory of 3,025, its Ordnance Survey names, and a reader that also takes Irish forms such as Tobermurry and Toberpatrick. Brigid's count is about the same (113 here against 119 there), but Mary's and Patrick's are far higher (162 against 88, 161 against 29), so Brigid's share falls. The two lists label the wells differently and cover different wells. The two figures measure different lists, and I trust this one more for Ireland because it is the complete inventory.
The word for well carries its own map. In the Republic, 45% of named holy wells have Irish-language names, almost all in the anglicised spellings of the nineteenth-century Ordnance Survey (Toberbride, Tubbrid, Tobermore), and 53% have English names (St Brigid's Well, Lady Well). The saints mostly travel in English: of the Irish wells whose name gives a saint, 23% have an Irish form. In Wales 63% of mapped named wells are a Ffynnon; in Scotland 14% carry a Gaelic Tobar or Fuaran. I found no Manx Chibbyr and only one Cornish Venton among the mapped wells, a gap in the open data rather than in the languages.
A saint's well was used. People came on the saint's day for the pattern, walked rounds, said prayers at stations, took water for a cure, and left tokens. The Irish inventory's descriptions record some of this. I searched them for words that mark each practice, ignoring mentions that are negated ("no votive offerings", "no devotions now"), and counted the wells that mention each one.
Two cautions. First, 397 records carry only a placeholder, so I count 2,628 described wells. Second, the open dataset carries only the first 254 characters of each description, and 86% of descriptions reach that limit. Many practices are mentioned further on in the full text. Every share below is therefore a floor.
Even in those short openings, 19% of described wells mention a cure, rags, offerings or a pattern. Patterns, pilgrimages, rounds and stations are the most common (12%). Cures come next (5.3%), and the eyes are the commonest ailment: 52 descriptions mention eyes, often in the name itself, Tobar na Súl, the well of the eyes. Rags appear at 32 wells (1.2%) and offerings or votive objects at 74 (2.8%), among them 14 with coins. One in five descriptions (21%) records a well that is dry, lost or known only as a site.
Coins dropped into the water, a tree with coins pressed into its bark: the wishing well's habit is already there at the saint's well.
Wells that name a saint are about as likely to have recorded practices as those that do not (19% against 19%), so the rituals belonged to the holy well as a type, whoever its saint.
And the wishing well itself? In the open map data the name is rare and mostly modern. Of 81 OpenStreetMap features called "Wishing Well" in Britain, Ireland and the Isle of Man, 38 passed my filter as real wells or springs; 13 are pubs, cafes or restaurants, 6 nurseries or care services, and 6 hotels, cottages or holiday lets. Names with rags, pins, eyes or warts are rarer still: 4 rag wells, 5 eye wells, 3 wart wells and 1 pin well. The folk practice survives better in the descriptions of Irish holy wells than in the names on the map.
Holy wells sit close to churches. I measured the distance from each well to the nearest church building in Wikidata (41,965 churches in Britain and Ireland) and compared it with 20,000 random points spread across the same country.
In the Republic, 15% of holy wells are within 500 m of a church in Wikidata, against 2.6% of random points: 5.7 times more often. In Britain, 61% of the mapped holy wells are, against 8.8% of random points. And the saints match. Where a well names a saint and there is a church with a saint's name within 2 km, they share the saint 36% of the time in the Republic and 55% in Britain, against 8% and 10% when I shuffle the wells' saints.
The British figure is high partly because volunteers map the wells they can find, and wells beside churches are easy to find. Even so, other named wells in Britain, the farm wells and village springs, are less often near a church (28%). The holy well was part of the parish landscape, not apart from it.
Republic of Ireland. The National Monuments Service's Sites and Monuments Record, every record classed "Ritual site, holy well" (3,025 records), with county, townland, coordinates, the names shown on the Ordnance Survey maps, and the opening of the archaeological description. This is an official inventory and the best source here.
OpenStreetMap. 66,517 features in the United Kingdom, Ireland and the Isle of Man whose name loosely matched a well word. Most are not wells (Wellington Road, Bridewell, Powell, spas, pubs and care homes), so I filtered them in steps, shown in the table. In the Republic, OpenStreetMap wells within 150 m of an inventory well were treated as the same well (1,137 features); the 407 that were not are kept in the data file but not counted, so that the Republic's figures rest on the inventory alone.
Wikidata. Items that are holy wells, 897 of them in the study area; 114 were not already within 150 m of a well from the other sources and were added.
A well counts as holy outside the Republic if its name gives a saint or says holy (Holy Well, Trinity, Lady Well), or if OpenStreetMap tags it as religious or a holy well, or Wikidata lists it as one. The language is read from the generic word only: well; ffynnon; tobar, tober, tubber or tubbrid; fuaran; chibbyr; venton. Saints are read as described above, from a fixed table of name forms; a name after "St" that is not in the table counts as its own saint, as written.
Each practice is a set of words searched in the description (for example rag, rags, rag tree, cloth, ribbons; or coins, pins, medals, rosaries, votive, offerings). A mention that follows "no", "not" or "without" within a few words is ignored. I checked the matches by reading samples, some of which are quoted above. The full word lists are in the code.
Recorded is not surviving: the inventory includes wells that are dry or lost, and the map shows neither all wells that once existed nor only those that do now. The comparison between Ireland and Britain mixes an official survey with volunteer mapping, so the gap between them says more about records than about the Reformation or belief, and I do not read it as a measure of either. The descriptions are cut at 254 characters, so practice shares are lower bounds. Wikidata's churches are uneven by country. Name readings miss saints hidden in unfamiliar Irish forms, and a few may be wrong.
Companion: my earlier map of saints and holy wells across Western Europe, which looks at church dedications from Wikidata.