Between the Norman Conquest and 1516, English kings licensed markets in their thousands, each one a bet that a village would grow into a town. I followed every one of the 2,173 market places in the Gazetteer of Markets and Fairs to today. Which became towns, and which are villages or fields now?
In 1251, according to the Gazetteer, a market was granted at Ravenserodd, a port on the sandbanks at the mouth of the Humber. It was a borough, with two markets and a fair on record. Nothing is there now: the sea took it in the fourteenth century, and there is no named settlement within 1.5 km of its grid reference. Ravenserodd is the extreme case. But it is a useful reminder that a royal market charter was a licence to try, not a guarantee of anything.
In the century and a half after 1200, the Crown issued market grants at a rate never seen before or since. Landlords paid for them because a market brought tolls, rents and, they hoped, townsfolk. Some of those places became Ashford, Stratford upon Avon or Salisbury. Many more stayed exactly what they were: a church, a manor and some fields, with a charter in the rolls to say that a market might once have been held there on a Tuesday.
This is the third of my "is it destiny?" studies, after Domesday Destiny and Minted. The question is the same: does early importance last? This time the test is the market charter, the medieval state's own bet on which places would be towns. The short answer: about 31% of medieval market places are towns today, and you could have predicted many of the winners in 1300.
Each dot is a place with a market recorded in the Gazetteer, at the grid reference the Gazetteer gives, coloured by what Ordnance Survey's gazetteer of place names calls the place today. Drag the slider to watch the markets appear: each place shows from the first year it has a market on record. Tap a dot for its story in numbers, or search for a place.
The Gazetteer records 1,879 royal grants of a market in England and Wales between 1066 and 1516, 1,549 of them by charter, the rest by letter close, letter patent or other grant. Another 710 markets were held by custom ("prescriptive" markets: old boroughs, mints and places whose market was recorded without a known grant). Grants of fairs ran alongside: 2,486 of them.
88% of all grants fall between 1200 and 1348. The busiest decade was the 1250s, with 254 market grants, and the 1300s were not far behind with 159. Then the curve falls off a cliff. From the 1350s to 1516 the Crown granted an average of 9.7 markets a decade, against 111 a decade from 1200 to 1349. The Black Death killed perhaps a third or more of the population; there were simply fewer people to buy and sell, and the markets England already had were more than enough.
For each market place I looked up its own name in OS Open Names, near the Gazetteer's grid reference. I could find the same or a very similar name for 2,080 of the 2,173 places (96%). The result:
Only 665 (31%) are towns or cities today, and a further 144 (7%) have been swallowed by a bigger town, like Southwark, Croydon or Kingston upon Thames in London. More than half, 1,168 (54%), are villages. 129 are hamlets and 67 have gone from the map altogether. Where a 2022 built-up area of the same name exists, the medieval market towns that are still towns have a median of 11,936 residents today.
Timing mattered enormously. Places with a market by custom, which usually means an Anglo-Saxon or Norman borough, did best: 54% are towns today. Of places whose first market was a grant before 1200, 50% are towns. After that the odds fall with every half century: 29% for first grants in 1200 to 1249, 17% in 1250 to 1299 and just 11% in 1300 to 1348.
The late arrivals mostly failed: they were squeezed into the gaps between markets that already existed.
The thirteenth-century lawbook known as Bracton set out when a new market was a "nuisance" to an older one. A reasonable day's journey, it says, is twenty miles, divided into three parts: the morning to go to market, the middle to buy and sell, and the evening to get home. A rival market within a third of that, six and two-thirds miles or about 10.7 km, could be challenged. (Thorne's translation renders the unit as leagues; the twenty-mile day is what fixes the distance.)
In practice the rule did not stop the Crown packing markets in. Of 1,484 places whose first market was a grant, 1,216 (82%) lay within 10.7 km of a market already on record, and the median distance to the nearest older market was 5.8 km. But closeness was fatal for growth. Only 15% of those squeezed-in markets are towns today, against 41% of the 268 that had room around them.
Taking each feature one at a time, the best predictors of a market place being a town today were being a borough (60% are towns, against 16% of the rest), having had a mint (82% against 28%), sitting on a major river (45% against 23%) or a Roman road (42% against 28%), and not being squeezed in beside an older market.
Many of these features go together: boroughs were often old, on rivers and on Roman roads. So I put them in one logistic regression for all 2,173 placed market places. Being a borough multiplies the odds of being a town today by 3.6, a mint by 2.9, a market by custom by 2.5, a fair by 3.1, a major river within 1 km by 2.0 and a Roman road by 1.7. An older market within 10.7 km roughly halves the odds (odds ratio 0.54). Once all that is known, the date of the grant adds little.
What about size in 1086? For the 1,288 market places I could match to a Domesday place with a household count, the biggest Domesday villages did best: 53% of places with 100 or more households are towns today, against 26% of those with fewer than 20. But below 100 households size made little difference, and once borough status and the rest are in the model, Domesday households add almost nothing (odds ratio 1.07 per doubling). Domesday size mattered mainly because the big Domesday places were already boroughs.
Alan Everitt's list of market towns around 1500 to 1640, which the Gazetteer flags, gives a halfway check. Of the 2,173 medieval market places, 732 (34%) were still market towns around 1600. The sorting had mostly happened by then.
Being on Everitt's list is the single strongest predictor I found: 74% of the c.1600 market towns are towns now. Most medieval markets that had not made it by 1600 never did. Still, 162 of the c.1600 market towns are villages, hamlets or gone today, among them Lavenham in Suffolk, Castle Combe in Wiltshire and Dunwich. They lost their place at some point in the four centuries since.
The saddest category is the borough, a place whose lord gave it a charter of urban privileges, that is now a hamlet or nothing at all. I count 30:
Some were destroyed by the sea: Ravenserodd on the Humber and Old Winchelsea on the Sussex coast. Old Sarum lost its people to New Salisbury in the thirteenth century. Several Welsh ones, like Castell y Bere, Dinefwr and Dryslwyn, were small boroughs beside castles. A further 221 boroughs are villages today, including Mitchell and Bossiney in Cornwall and Torksey on the Trent, where Domesday Book counts 264 households and the parish has 777 residents today.
In Cambridgeshire only 9% of medieval market places are towns today, and in Norfolk, with 154 market places, just 16%. East Anglia had a dense web of small chartered markets (the median Norfolk market place had another within 3.1 km), most of which stayed villages. In the north and west, where markets were fewer and often further apart (a median of 7.0 km in Cheshire, 7.3 km in Cumberland and 5.9 km in Wales), a much larger share became towns: 50% in Cheshire, 49% in Lancashire and 56% in Wales. Tap a column heading to sort.
A charter is not a market. The Gazetteer records grants and other evidence; it cannot tell us whether every granted market was ever held, or for how long. I use the first year with evidence, which for prescriptive markets is the first time a market is mentioned, not when it began.
"Town" is Ordnance Survey's word. OS Open Names classes places as City, Town, Village, Hamlet, Other Settlement or Suburban Area. These labels are a modern cartographic judgement, not a legal status, and they are imperfect: Upton upon Severn and Lavenham are "villages", and some sizeable places are listed as "Other Settlement". I used built-up area populations from Census 2021 to catch districts of big towns and sizeable hamlet-type entries.
Matching names is fallible. 1,947 places matched an OS name exactly, 133 by a similar spelling or word order, and 79 only by the nearest settlement; 14 have no settlement within 1.5 km. 7 grid references were corrected where the printed reference was clearly mistyped. The median distance between a grid reference and the OS point of the same name is 226 m.
Associations, not causes. Boroughs, rivers and Roman roads all go together with age and lordship. The regression separates them only as far as these crude measures allow. Domesday households could be matched for only 1,288 places, and Domesday did not survey the far north.
Markets. I parsed the county pages of Samantha Letters's Online Gazetteer of Markets and Fairs in England and Wales to 1516: 2,411 place entries, 2,590 markets and 2,936 fairs. For each place I kept only derived variables: its grid reference converted to latitude and longitude, the number and kind of markets and fairs, the first year with evidence of each kind, and flags for borough, mint, a market town c.1600 in Everitt's list and a fair in Harrison's list of 1587. The Gazetteer's grid references are two four-digit numbers in hectometres on the National Grid; I converted them with pyproj and checked the result on sixteen well-known places, which all land within 962 m of the OS point for the town of that name.
Today. OS Open Names populated places, matched by name within 4 km of the grid reference (letters only, ignoring case and punctuation, with similar spellings and word order accepted within 3 km). Census 2021 populations of 2022 built-up areas of the same name come from ONS small-area data. Domesday households come from the Hull Domesday Project data prepared for my Domesday Destiny study, matched by name within 4 km.
Geography. Major rivers: OS Open Rivers watercourses whose named course is at least 40 km long. Roman roads: the Itiner-e reconstruction (conjectured roads only). Coast: Natural Earth. Odds ratios come from a logistic regression; one-at-a-time comparisons use Fisher's exact test.