Britain's oldest surviving road map, drawn about 1360 with east at the top. I found 157 of its 179 way stations by eye, fitted its shape to the real island, and checked 169 of the distances written beside its red roads against the real roads. The shape wanders. The numbers, mostly, hold.
In the Bodleian Library in Oxford is a map of Britain on two skins of parchment, about 115 by 56 centimetres. Nobody knows who drew it. It is named after Richard Gough, the antiquary who bought it in 1774, and it is usually dated to about 1360. East is at the top, so Scotland stretches away to the left, Wales bulges out at the bottom and Kent sits up near the top right corner. Across England and Wales run thin red lines, joining towns drawn as little houses and churches, and beside most of them a Roman numeral: how far it is to the next town.
Those red lines make it the oldest surviving road map of Britain. In 2024 the historians Eljas Oksanen and Stuart Brookes published a database of them: 179 way stations, the 188 red lines between them with the figure written on each, and the real road each line most likely stands for, traced on the ground and dated from Roman roads to the road books of the seventeenth century. Their database tells you where each town really is. It does not tell you where the map draws it. So I found 157 of the 179 towns on the map by eye, and set the two side by side. Click a town on either map to see it on the other.
Left: the Gough Map (Bodleian Library, MS. Gough Gen. Top. 16), photograph on Wikimedia Commons, public domain. Rings: the 157 way stations I located, strong for a label read or the right drawing on the right red line, faint where I was unsure; blue rings mark the control points in Scotland and on Hadrian's Wall. In Map layers you can lay the National Grid and the modern coastline over the parchment, bent to fit. Right: the way stations at their true places, the red lines with their written miles in Roman numerals, and the real roads under them by period (Oksanen and Brookes, CC BY 4.0). Drag the slider to bend the island, its towns and red lines into the shape the Gough Map gives them. Land: Natural Earth.
Fit the located towns to their real places with a single scale and a single turn, the best a flat copy of the real island could do, and the map comes out turned 95° from north-up: east at the top, as everyone says, but leaning a further 4.6°. Allowing for that, the typical town sits 17.6 km from where it should, and 76% of them are within 25 km. A general stretch in one direction barely helps: the best straight-line fit is only 1.11 times longer one way than the other. For a map drawn without any survey, this is remarkably good.
How far each region is out of place, and at what scale it is drawn
For each region: the median distance between a town's true place and where the map puts it (bars), and its own best scale against the map's average (labels). Hover a bar.
The errors are not spread evenly. England's south-east and the north are drawn closest to the truth, with median misplacements of 13.9 and 12.6 km. Wales is the outlier: its towns are misplaced by a median 33.2 km, and the whole country is drawn at 1.22 times the scale of the rest, as if seen through a magnifying glass. The coast of Cardigan Bay is pulled out the furthest: Aberystwyth is 70 km out of place, and Aberdovey, Barmouth and Harlech more than 60 km. The north of England, on the other hand, is drawn small, at 0.90 of the average scale.
Scotland is where the map gives up. There are no red lines north of the border, and few names I could read with confidence, so I added four Scottish places I could, and the two ends of Hadrian's Wall. The Wall itself is drawn well, its ends within about 31 km of their places. Beyond it the land is squashed: Orkney lies 450 km from the Wall in reality but only 356 km on the map, and it is drawn 246 km from its true place. Perth is drawn at 0.67 of its true distance from the Wall, while the islands at the mouth of the Forth are pushed out. With so few points the slider above bends only England and Wales: north of the Wall it would be guesswork.
Of the 188 red lines, 169 still carry a legible figure; on 19 it is faded away. The figures run from 4 to 50, with a median of 11. Following each line along the real road that Oksanen and Brookes traced for it, one written mile covers a median 2.08 km of road, or 1.89 km as the crow flies. That is about 1.29 of today's statute miles, so the map's mile was a longer, older, customary mile, not the one fixed by statute in 1593.
Each red line: the figure on the map against the length of the real road
169 red lines with a legible figure, coloured by the period of most of the road under them. The dashed line is 2.08 km per written mile; the band is 25% either way. Hover or tap a dot.
The figures track the roads well: 72% of them are within 25% of the road length at that rate, and 89% within 50% (rank correlation 0.74). The longest figure on the map, L (50) for Bristol to Marlborough, is also one of the most generous, at 1.24 km a mile; that line runs 62 km on the ground. The worst misfits look like a slip of the pen or a misread: Rye to Winchelsea is written as 8 miles for 4.7 km of road.
Was the mile the same everywhere? Roughly. By region of the line's first town it ranges from 1.86 km in the south-west to 2.34 km in the Midlands. Short lines run a little longer per mile (2.21 km for figures up to 8) than long ones (1.93 km for figures over 12), as you would expect if a short stage were rounded up to a whole mile, or measured along a winding lane.
Here is the strangest thing about the Gough Map. For 146 red lines with both towns located, I can compare two ways of reading the distance: the length of the line as drawn, at the map's average scale, and the figure written beside it. The written figure is wrong by a median 13%; the drawing by 31%. The figure is closer to the real road on 66% of lines, and within 25% on 71% of them, against 47% for the drawing. Whoever made the map, or the itineraries behind it, knew the roads by distance, not by shape.
The drawing against the numbers
For each red line, the error of the drawn length and of the written figure against the real road. Lines to the left of the diagonal: the number is better. Hover a dot.
Could you have planned a journey with it? Adding up the figures along the red lines from London, the road to York comes to 202 miles, which at 2.08 km a mile is 420 km, against 424 km along the real roads beneath. To Bristol the map gives 202 km for 203; to Carlisle 485 for 493; to St David's 445 for 432. Over a long journey the slips cancel out. The worst is the road to Exeter, where the map's 165 miles make 343 km, 1.18 times the road.
Journeys from London, by the map's own figures
Shortest route over the red lines, adding up the written miles (a few faded figures filled in from the road), converted at 2.08 km a mile, against the real road along the same lines. Hover a bar for the route.
Which roads do the red lines follow? Of the 4,505 km of real road that Oksanen and Brookes traced under them, 35.4% is Roman road. Another 28.9% has medieval evidence: 13.1% is known from before Domesday Book in 1086, from Anglo-Saxon charters and place names, and the rest from bridges, charters and travellers' accounts of the centuries after. 32.7% is known only from the road books of the seventeenth and eighteenth centuries, Ogilby and Cary, though the map shows it was already a route by 1360. 71 of the 188 red lines (38%) run mostly along a Roman road.
The real roads under the red lines, by period
Kilometres of traced road under the red lines, by the period code in the database (my reading of the codes). Hover a block.
The age of the road makes no difference to the figure. Lines mostly on Roman roads come out at 2.10 km a mile, those known only from the later road books at 2.10, and the share within 25% is much the same. The map's distances were the distances people travelled, on whatever road they used.
The data. Oksanen and Brookes' GIS database (Archaeology Data Service, 2024, CC BY 4.0): 179 way stations with modern names and historic counties, 188 red lines with the written figure, and 455 traced road segments with period codes. The codes are numbers without a key in the files I could reach; I read them from the categories and comments as Roman road, evidence before Domesday, dated medieval evidence after Domesday, other medieval evidence, early modern road books only, river and undated. Road distances follow the traced segments between each line's two towns (shortest path, towns snapped to the nearest traced road; road lengths as given in the database).
Towns on the parchment. No table gives where each town is drawn, so I found them by eye on the photograph at full resolution (12,224 by 6,021 pixels): 33 by a label I could read, 56 by the right drawing at the end of the right red line with the right numeral, and 68 with real doubt, mostly among faded drawings. 22 I could not place, 6 of them in Cornwall, where the parchment is worn. A check: leaving each town out, a smooth fit through the others predicts its place to within a median 8.1 km. Fitting only the surer ones turns the map 96° instead of 95°, and changes no conclusion here.
Shape. Similarity and affine fits by least squares from British National Grid to image pixels; misplacement is the distance between a town's true place and its map place taken back through the overall similarity. Region fits use at least 6 towns. The warped grid, the bent coastline and the morph use a thin-plate spline through all located places, and say nothing about land far from them. The Scottish points are few and partly guessed.
Miles. Kilometres per written mile, medians over lines with a legible figure. The statute mile (1.61 km) is only a yardstick: the map's own mile was not standardised. Journeys use the shortest route by written figure; faded figures are filled from the road length.