by Graham Robb
I now knew that this apparently defective land map of ancient Britain and Ireland had been based on information of an extraordinarily high quality. The coherence of the information had made the process of ‘hacking into’ the map unexpectedly straightforward: first, the map had to be resized to conform to the original graticule; second, the land data had to be separated from the coastal data. The third and final key presented itself quite naturally, though it took some time to realize that I was looking at the only contemporary cartographic evidence of the Roman conquest of Britain: the map of the towns of Britain is not a single map but an atlas, comprising five distinct maps, each with its own graticule (figs 8 and 9). Here again, any attempt to reconstruct the original source as though it had been a coherent whole was doomed to failure.
Ptolemy’s re-plotted coordinates confirmed the system described in the book on the ancient Celts, but they also showed that the basic principle had been applied more widely than I had thought. In analysing the organization of roads and settlements in pre-Roman Britain, I had found that the ratio of 4 and 3 operated only in the southern half of England, as far north as Mediolanum (Whitchurch in Shropshire). I had tentatively identified this as a sphere of Druidic influence beyond which no such pattern was detectable. This was confirmed by Ptolemy’s atlas, in which the 4:3 graticule operates only up to Mediolanum.
North of Mediolanum, it seemed at first as rough and ready as one would expect an ancient map to be. Even in those days, ‘the North’ was considered less sophisticated, and I assumed that the lack of evidence was a sign of backwardness. Writing the book in a land where sheep outnumber human beings, I had come to think of Oxford and London as palaces of learning shining on a far horizon, and I was not predisposed to discover in the region of Lancashire, Yorkshire and Cumbria the intellectual treasure of a maligned civilization.
*
The unsuspected glories of map-making at the dawn of British history are a wonderful and disconcerting sight. The map of southern England seems to be unique in the history of cartography, though it may have been typical of ancient Celtic mapping. It functions like a direction finder on a viewing platform or like a giant signpost: each finger on the signpost points in the right direction, while its length gives a rough idea of distance. The distances are approximate but the directions are accurate to within two degrees from certain nodal points – in particular, London.
This is, to all appearances, a land map designed to be used in Londinium (fig. 10). Though its practical uses would have been limited, it would have suited the purposes of an ancient Briton who, like a Muslim praying to Mecca, wanted to know the exact direction in which a certain place lay but for whom distance and journey time were unimportant. The item acquired by the Library of Alexandria might have resembled one of the portable Roman sundials which occasionally come to light. The original map was certainly more accurate than Ptolemy’s sample since the bearings could not have been so precisely calculated without knowledge of the correct distances.
The map of northern England, by contrast, is so much like a modern map that, without the medieval copies of Ptolemy’s Geography, it could be mistaken for a modern forgery. It uses a graticule of 3 by 2. Distances as well as directions are shown with a precision unmatched until the Renaissance. Its only obviously exotic feature is its orientation, which is based, conveniently, on the sun rather than the North Pole (a sensible rather than superstitious convention since, at that time, there was no star at the pole).
These are by far the most accurate maps to have survived from the ancient world. For the purposes of the original surveyors, they were effectively faultless: expressed anachronistically in degrees, the bearings appear impossibly precise, but each angle is the product of a simple whole-number ratio (3 and 5 for Colchester, 5 and 9 for Winchester, etc.). This is the proof which Hugh Davies lacked in 1998 when he argued, controversially, that the Roman road system must have been based on maps produced by triangulation. It also answers a long-standing question: how did Marinus of Tyre know that the exact distance between London and Chichester was fifty-nine Roman miles? It was comparatively easy to measure distances by road, but this, remarkably, is the distance in a straight line.
In Ptolemy’s mind, the lack of British coordinates based on astronomical readings was a severe shortcoming. As we now know, he was right to think that only measurements of the sun’s shadow and the length of day could precisely define a location on the earthly sphere. The problem was that, although the theory was correct, the technology (especially chronometers), like most latest technologies, was full of bugs and glitches.
The result can be seen in Ptolemy’s map of Gaul. This long-established Roman province, with its road network and universities, should have been one of the best-mapped parts of the empire. In theory, the map of Gaul was more sophisticated, but in reality, it would have been unusable: Reims appears as a suburb of Paris, Paris is shown to the south rather than north of Nantes, and several parts of the map are almost upside-down. Its main cartographic value is as a record of erroneous readings. In Britain, older but more reliable methods of determining geographical position had remained in use. These methods, like those used for medieval portolan charts, were perfectly adequate for relatively small parts of the globe such as northern England.
The division of England into separate maps, each with its own conventions, is a graphic guide to the early Roman conquest of Britain. The towns on the map of southern England had all been conquered by about AD 60. Those on the northern England map were conquered only later, in the AD 70s. Perhaps, at the end of each campaign, the information obtained by Roman officers from natives or merchants was sent back to Rome, from where it eventually reached Alexandria. The new frontier was then consolidated in preparation for the next advance.
These borders which temporarily marked the northern limit of the Roman Empire have long been suspected but never traced in detail. They may already have existed as tribal boundaries. Use of the 3:2 rather than 4:3 system probably reflects a cultural divide. Most of the map of northern England covers the territory of the tribe or tribal federation of the Brigantes. ‘Brigantia’, previously thought to have stretched to the borders of what is now Scotland, must have ended much farther south at Epiacum. This name is currently applied to the Roman fort of Whitley Castle, advertised as ‘the best preserved fort in the Roman Empire’, but the restored map of Ptolemy shows beyond doubt that it belongs to the lonely fort at Low Borrowbridge.
Places recorded on the separate maps of Ptolemy’s British atlas, with the approximate foundation dates (AD) of the Roman forts and towns. For more detail, see figs 8–12.
North of Low Borrowbridge, the road and the railway climb to the pass of Shap which many first-time travellers from the south, impressed by the obvious geographical frontier, mistake for the Anglo-Scottish border. In the AD 70s, the Roman legions pushed on beyond Shap, up the western side of the Pennines, to Whitley Castle and Carlisle, leaving the eastern side for a later campaign. These two towns of the Selgovae tribe accordingly appear on the map of northern England. The forts on the eastern side were established only later, in the AD 80s, which is why they appear on the later, Caledonian map.
The Pennines are still a barrier, especially when the Hartside Pass is snowbound and landslips have closed the railway. To find this geographical reality in the digital hieroglyphics of a second-century atlas was a strangely thrilling experience. There on the edge of the map of northern England, Carlisle looked like an outpost, as it had done in 2010. Beyond, lay other places with enigmatic names. Only one of them – Bremenium – had ever been identified. But if the map of Caledonia followed a similar logic, there was every hope that, after Carlisle, the skies of the borderlands would remain clear.
25
The Kingdom of Selgovia
It took several weeks to reach this point in the atlas and I still had no idea what was waiting in the borderlands. The maps were certainly accurate enough to allow previously unidentifiable places to be ident
ified.46 Since Ptolemy noted the tribes to which each town belonged, this had already led to a significant modification of the tribal map of Britain. I knew, too, that many of the towns shown on the maps had certain features in common. All of them stood at junctions of the Roman road network, almost one-third were inland ports and many, not surprisingly, lay in regions known to have been mined for precious minerals. It seemed likely that the Solway Firth and the inland port of Netherby/Castra Exploratorum would have been of particular interest to merchants and military commanders. But first there was the problem of the Caledonian map to be solved.
The peculiar rotation of Scotland to the east is easier to correct than to explain. The original map may have shown west at the top and Ptolemy reproduced the map as he found it, assuming north to be at the top. The other possibility is that this is an early example of southern meteorological prejudice. The Roman geographer Strabo had asserted that human life was impossible anywhere north of Hibernia (Ireland), and so Ptolemy ‘corrected’ the map which implausibly showed several tribes and towns in the supposedly frozen north by turning it ninety degrees to the right. In either case, the map simply has to be re-rotated ninety degrees anti-clockwise. Though the Caledonian map is less accurate than the northern England map – probably because it covers a much larger area – it can be read without much difficulty on a graticule of 4 by 1. The smaller number of Roman forts in Scotland makes identification relatively unproblematic since there are fewer candidates for each site (fig. 12).
The fort of Bremenium in the bottom-right corner of the restored map is known from the second-century Antonine Itinerary to be the fort of High Rochester which lies on the Roman road, Dere Street, just below Carter Bar on the Anglo-Scottish border. The place labelled ‘Alauna’ would have stood on or near Hadrian’s Wall. The most likely candidate is Corbridge – either the Roman fort or the neighbouring civilian settlement. Both these Northumbrian places are attributed by Ptolemy to the Votadini tribe, while the two Cumbrian forts to the west are attributed to the Selgovae.
In this sparse configuration of points, two other towns lie close together. Their names are Colanica and Curia. These are Celtic words, thought to be generic terms meaning ‘tribal centre’ or ‘meeting place’. Neither town has been identified until now.
Despite their proximity, each town is attributed to a different tribe – Colanica to the Damnonii of southern Scotland, Curia to the Votadini of northern England. This must account for the inclusion of two places so close together. Nowhere else on the British atlas are two towns shown in such proximity. The original map-maker evidently thought it important to indicate this tribal frontier. A merchant travelling from one to the other would have wanted to know with which tribe he would be dealing.
The accuracy of the restored map makes it possible to identify these neighbouring places with complete certainty. Thanks to the data gathered in Britain and digitized by Ptolemy, the cryptic coordinates of Colanica and Curia are as legible as the road signs on the A7 which crosses the border between Longtown and Langholm. Curia of the Votadini was the ‘ancient Citie’ whose ‘strange and great ruins’ lie under the fields of Netherby by the Esk, while Colanica of the Damnonii was the British town and Roman fort eight miles upstream at Broomholm, on the grassy platform at the end of the Roman road which forms the northern boundary of the Debatable Land.
*
Towns of the Damnonii, Selgovae and Votadini (straddling Hadrian’s Wall) and approximate tribal territories. The shaded area is the Debatable Land. For more detail, see figs 12 and 13.
It seemed (and was) an extraordinary stroke of luck to discover on a map of the second century AD the exact markers of the northern and southern boundaries of the Debatable Land. For some reason, this obscure corner of Outer Britannia was one of the most precisely mapped parts of the Roman Empire. It was strange to think that, in the second century and in 1552, when Henry Bullock created his ‘platt’ of the Debatable Land, this small region, which, even now, is missing from most guidebooks, was depicted on the most accurate maps of the time.
But was it really so astonishing to see the Debatable Land rise from the night of ages? In the land on the other side of the Liddel, two tribes, each controlling a vast area, had faced one another across a frontier. Apart from the rivers themselves and the Cheviot watershed, this is the earliest sign of the future Anglo-Scottish border and one of the oldest clues to any political division in pre-Roman Britain. The emptiness revealed by the archaeological record was not misleading. The medieval documents had been right to call that oddly resilient realm ‘ancient’. Neither Scottish nor English in the Middle Ages, in the late Iron Age it had been neither Damnonian nor Votadinian.
Ptolemy’s maligned atlas of ancient Britain also reveals a crucial difference between the frontier which still exists and its Celtic ancestor. The inter-tribal zone was not just a buffer between two states. A short distance to the south, the map shows a town or fort called Uxellum, attributed to the Selgovae tribe. This is Uxellodunum, where Hadrian’s Wall reaches Carlisle on the bluff above the river Eden.
It is easy to think of the tribes of ancient Britain as either proto-English or proto-Scottish, but the territory of the Selgovae belongs to a different era. Ptolemy’s map proves that the Selgovae inhabited both sides of the Solway Firth, which now divides England from Scotland. Their kingdom was a Mediterranean of the north: it stretched from the highest Pennines in the east to the middle of the Galloway peninsula in the west. The broad arm of the Solway looks like the permanent marker of an ancient division, but two thousand years ago, when a Selgovian stood where King Edward I gazed at Scotland for the last time, he would have been looking at his own country, and when he walked out over the sands at low tide, he would have crossed the water without leaving the motherland.
*
The kingdom of the Selgovae would have met those of the Damnonii and the Votadini in the region of the Debatable Land. Though the Selgovae possessed the Cumbrian Plain and both shores of the Solway, the vital inland port at Netherby belonged to the Votadini, while the Damnonii controlled the northern end of the Esk corridor at Broomholm, where roads converged from north, west and east, tapping the rich mining and agricultural areas of lowland Scotland. This tripartite division centred on a march is typical of the ancient Celts. A very similar situation existed in Gaul, where the great tribal federations of the Aedui, the Arverni and the Bituriges formed a broad frontier zone, later called the Marche, which is still recognizable as a linguistic and cultural watershed. The arrangement is rare if not non-existent in the modern world, despite its obvious geopolitical value.
With shared boundaries in the flatlands on the edges of the Debatable Land, all three nations had access to the trade routes of the Irish Sea. This inter-tribal zone still looks like a no man’s land. The landscape has been transformed by roads and railways, wind farms and pylons, but the physical and even the human geography is unaltered. At Longtown, the farmers of Cumbria and Dumfriesshire meet at Britain’s biggest sheep market. In 2001, the area’s centrality was demonstrated by an infected sheep sold at Longtown Mart, from where foot-and-mouth disease spread rapidly to the rest of the country.
The proto-Debatable Land may have arisen as a buffer zone in the aftermath of a destructive war, or it may have been created by consensus or divided inheritance when the region was settled by Celtic tribes. These tripartite divisions cushioned by a broad frontier could be as sturdy as the arch of a bridge sustained by the force of compression. Each tribe had an interest in preserving the mutually beneficial arrangement. The closest examples today are antagonistic rather than cooperative. The buffer zone in Cyprus, which separates Greek and Turkish Cypriots, is administered by a third party, the United Nations. Large parts of the ‘Dead Zone’ have been abandoned to nature and are useless to the people on either side. To judge by the long absence of any settlement, the Celtic solution, which was still being applied by the Anglo-Scottish borderers five hundred years ago, was to make the buffer zone �
�batable’ rather than debatable. The livestock of opposing tribes or nations maintained the unoccupied zone as pasture and prevented it from turning into wasteland.
The oldest known trysting place in the borderlands was still in use when the Debatable Land was partitioned in 1552. The Lochmaben Stone at the foot of the Sark, on the south-western tip of the Debatable Land, where English and Scottish officials discussed international matters, bears the name of a local Celtic god, Maponus (here). It has often been suggested that the sacred stone was once a meeting place of Celtic tribes. This now looks more than likely. The monolith on the ocean’s edge, where the river-borne rubbish of two nations gathers in the mud before being carried out to the Irish Sea, stands in the area where the three Celtic nations of the future borderlands came together.
The worn stone of Maponus now has nothing to tell us, and the trail revealed by the second-century map ends like the ghostly footprints on the stairs of Netherby Hall. The history of these tribal and national divisions can be traced only in the centuries-long emptiness of the Debatable Land. Perhaps the sanctity of the buffer zone was preserved by a religious veto, later materialized in the chapels which stood on its borders. The only other clue is a name traced faintly on the map of 1552, which identifies the land by the Esk at the end of the Scots’ Dike as ‘Dymisdale’. A document of the same period calls it ‘Dimmisdaill, as the common people say’. The name disappeared from maps and local memory long ago. Three other Dymisdales or ‘Doomsdales’ in Britain are associated with justice and execution; the Dymisdale of Inverness, for example, was the way that led to Gallows Hill. Did the ancient Celts and their Dark Age successors picture that sacred enclave, which lay deserted and silent after sundown, as the other world, inhabited only by the spirits of the dead?