Britain is full of dragons: worms coiled round hills, knuckers in bottomless pools, serpents in the marsh. Does every dragon sit next to a church, a barrow, a hillfort or a river? I placed 26 named legends and 135 dragon-word place names on the map and measured each against fifty ordinary places of the same kind nearby.
Every county seems to have a dragon. The Lambton Worm wrapped itself round a hill above the River Wear; the Knucker of Lyminster lived in a pool so deep that the church bell-ropes, tied together, could not find the bottom; the Dragon of Mordiford was painted on the wall of the village church until 1811. On the map itself, the old words for dragon still turn up: Old English draca lies behind some Drake names and wyrm ("dragon", "serpent", "worm") behind some Worm names.
Folklorists have long noticed that dragons keep particular company. In Beowulf the dragon guards a treasure in a burial mound, and two places called Drakelow are said to mean "dragon's mound". Dragon-slayers lie in churches; St George and St Michael, the Church's two great dragon-killers, have churches of their own. Hillforts and rivers turn up in the stories too. So I asked a simple question of the map: does every dragon sit next to a church, a barrow, a hillfort or a river?
There are two kinds of dragon here. The first is the legend: a named story, such as the Sockburn Worm or the Dragon of Wantley, that a written source ties to a place. I collected 29 from Wikipedia and from D. L. Ashliman's collection of British dragon texts, and placed 26 of them at the village, hill or wood the source names. The second is the name: every place on the Ordnance Survey's open gazetteer whose name contains Dragon, Drake, Worm or Serpent. There are 135, once the dragonflies, Serpentine lakes and wormwood are set aside.
The short answer: the legends cling to the church, and that is the one strong pattern. They sit no closer to barrows, hillforts or rivers than ordinary places of the same kind. The place names are different creatures. As a whole they sit no closer to anything, except that in the north they are often near a recorded burial mound. And the most famous dragon mound of all, Drakelow, has no recorded barrow anywhere near it.
Large scarlet marks are the legends, placed at the place their source names. Small marks are the place names, coloured by their word. Tap any mark for its source, the reason it sits where it does, and the distances to the nearest church, barrow, hillfort and river. Churches of St George and St Michael, barrows and hillforts can be switched on as layers.
The legends are an English crowd: 24 of the 26 I could place are in England, with one each in Scotland and Wales. That says more about the sources than about the dragons; Ashliman's collection and English Wikipedia lean heavily on English county folklore. The names are spread more widely, but they too are thin in Wales (two) and in much of Scotland, where Welsh and Gaelic names would not contain these English words at all.
Each legend was compared with fifty ordinary OS places of the same kind (villages with villages, hills with hills) from the same 50 km square, and with fifty random points of land. Of the 26 legend places, 77% lie within 1 km of a church recorded in Wikidata. Of the matched places, 55% do, and of random land, 34%. In 2,000 replicate draws of matched places, a gap this large turned up with chance 0.015. The median legend sits 303 m from a church, against 824 m for matched places.
Part of this is how the legends reached us. In eleven of the 26 stories the source itself points to a church: the slayer's tomb at Nunnington, the Slayer's Slab at Lyminster, the carved stone at Linton Kirk, the dragon painted at Mordiford, the dragon asleep on the tower at Llandeilo Graban. A tomb or a carving in a church is exactly the kind of thing that keeps a story alive and gets it written down. But the legends without a church in the story sit close to churches too (73% within 1 km, against 51% for their matched places), so the pattern is not only in the telling.
The dragons in the stories live where the people lived: in the parish, a short walk from the church.
Most legends (19) are placed at a village or hamlet, because that is the place their source names. Those 19 are very close to a church (89% within 1 km, against 66% of matched villages and hamlets). The seven placed at a named hill or wood (Worm Hill, Bignor Hill, Loschy Hill, Linton Hill, Dragon Hill, Shervage Wood and St Leonard's Forest) are too few to test on their own.
The other three companions do not show up for the legends. 23% of legend places lie within 2 km of a recorded barrow or burial cairn, against 20% of matched places. 23% lie within 2 km of a hillfort, against 18%. 19% lie within 500 m of a named river, against 16%. None of these gaps is beyond what luck gives with 26 places.
So does every dragon sit next to something? Nearly. 92% of the legend places are next to at least one of the four, against 75% of matched places and 63% of random land (chance 0.041 against matched places). But almost all of that comes from the church. Britain is so crowded with churches, mounds, forts and rivers that most ordinary places are next to one of them too.
The place names are a different kind of evidence. A name containing Worm or Drake may come from a dragon, but it may just as well come from a snake, a male duck, a person called Drake or a personal name that happens to begin with Worm. I do not try to decide which. I sort the names by form only, and treat every one as a possible dragon name unless a source supports a dragon reading.
The biggest classes are names that begin with Worm- (51, such as Wormley, Wormington and Wormit) and other Drake names (28, many of them possessive, like Drake's Wood). The names most likely to carry a dragon are fewer: 30 with Worm as a word of its own (Worm Hill, Worms Cleuch, Wormhill), 11 that join Drake to a mound or hill word (Drakelow, Drake Law, Drake Howe), 11 with the word Dragon and four with Serpent. Only five names, at four places, have a dragon reading in a source I could check: Drakelow in Derbyshire and in Worcestershire ("dragon's mound"), Wormhill in Derbyshire, and Dragon Hill at Uffington, where St George is said to have killed the dragon.
Taken together, the names sit no nearer churches, hillforts or rivers than ordinary names of the same kind in the same area. Only barrows show a difference: 26% of the names lie within 2 km of a recorded barrow or cairn, against 20% of matched names (chance 0.048).
The barrow result comes from the north. In the north of England and in Scotland, 33% of the 60 names lie within 2 km of a recorded barrow or cairn, against 16% of matched names (chance below 0.001). In the south and the Midlands the share is 20%, against 22%. Northern dragon names are often hills (Worm Hill, Worm Law, Drake Law, Worms Hill) on high ground where cairns and barrows survive and are recorded, so some of this may be the ground and the records as much as the dragons; the matched names are hills from the same squares, which helps, but cannot remove that entirely.
The best single example is Drake Howe on Cringle Moor in the North York Moors, the second name of an OS feature. A howe is a mound, and the nearest recorded barrow, called "Round barrow at Drake Howe", is 10 m away. But the two Drakelows whose name is explained as "dragon's mound" have no recorded barrow nearby at all: the Derbyshire one is 7.3 km from the nearest, and the Worcestershire one 13.8 km. Their mounds, if they were barrows, may have been ploughed away long ago, or may never have been barrows. Dragon Hill at Uffington, by contrast, sits 330 m from a recorded barrow and 970 m from a church, right under the White Horse.
The north-east is famous for its worms: Lambton, Sockburn, Spindleston Heugh and, just over the border, Linton. In my set of legends, four of the five stories whose creature is called a worm are in the north, against one in the south (Wormhill, in Derbyshire). The southern legends are dragons and serpents. That fits the old idea that "worm" lasted longer as the word for a dragon in the north, though with so few stories it is a pattern, not a proof.
The place names do not follow. Per 100,000 named OS features, the north has 12.6 names with Worm as a word against 9.8 in the south, a small difference; for all Worm- names the south has more (34.6 against 26.0). The densest region for dragon-word names is the West Midlands (10.7 per 10,000 features), thanks largely to a run of Worm- names in Herefordshire, followed by Yorkshire and the Humber (8.4). Scotland (2.9) and Wales (1.1) are low, as language alone would lead you to expect.
If dragon stories and dragon-slaying saints went together, churches of St George and St Michael should be commoner near the legends. They are, a little. Of the 317 churches within 5 km of a legend place, 7.3% are dedicated to St George or St Michael, against 5.4% of churches near the matched places and 3.9% of all 36,435 churches in Britain in Wikidata. 27% of legend places have a St George church within 5 km, against 14% of matched places (chance 0.090), which is suggestive but no more.
The clearest single case is one where the source says it outright: at Brinsop in Herefordshire, the Dragon's Well lies on the south side of the church, and the church is dedicated to St George. At Uffington, St George himself is the slayer, and the bare patch on Dragon Hill is where the dragon's blood fell. Elsewhere the slayers are local knights and heroes (Lambton, Conyers, Somerville, Shonks), and the dragon names on the map show no St George effect at all (13% with a St George church within 5 km, against 12%).
I took every British legend with a place from the English Wikipedia "List of dragons in mythology and folklore" (wikitext of 27 October 2026), the articles on the Lambton, Sockburn, Laidly and Linton worms, the Knucker, the Mordiford dragon and Dragon Hill, and D. L. Ashliman's British Dragons collection of old texts. For each legend I recorded the phrase that names the place and placed it at the most specific of the named places that OS Open Names also lists: a named hill or wood if there is one, otherwise the village or parish. 29 legends are in the table and 26 are placed; the Stoor Worm (a sea monster), the red and white dragons of Wales and the dragon of the Nine Maidens' Well (whose places are not in OS Open Names) are listed but not mapped. Every placement and its reason is in data/legends.csv. A Wikidata query for dragon items timed out, so Wikidata adds no legends here.
I searched all 267,744 OS Open Names features for words beginning Dragon, Drake, Worm and Serpent, and set aside 34 Dragonfly, Serpentine, Serpentwalk and Wormwood names. The classes are about form only. Where I mark a name as supported, the source is linked in the data; every other name is only possible.
For every legend and name I drew 50 ordinary OS names of the same local type from the same 50 km square (100 km or the whole country when fewer than 20 were available), leaving out every name with a dragon word, and 50 random points of land in the same square. Each test compares the real places with 2,000 replicate sets, each taking one control at random per place; chance values are two-sided. Distances are straight lines to: the nearest of 36,435 church buildings in Wikidata (any date, with dedications as parsed in Whose Church Is It?); 12,387 barrows, burial cairns and mounds in OpenStreetMap and Wikidata (from Is the Barrow a European Thing?); 3,610 hillforts in the Atlas of Hillforts of Britain and Ireland (as used in Within Sight); and named rivers in OS Open Rivers (watercourses named River or Afon, and tidal rivers).
The legends are few and come from sources that favour England and favour stories with a church monument. A legend placed at a village shares that village's church by construction, which is why the controls are other villages; even so, villages with a Wikidata church item may differ from those without. Wikidata's churches are not only medieval churches: they include buildings of every date. The barrow records are much fuller in some areas than others, and many barrows have been ploughed out. River crossings, which matter in several stories (the Lambton Worm in the Wear, Conyers at the Tees), cannot be tested properly with these data, so I used distance to a named river instead. I ran 13 measures on many groups, so single chance values near 0.05 should be read with caution. The church result for the legends and the barrow result for the northern names are the two I would put weight on.