Long before the Gauls had long-distance roads and a telegraph system, and even before there was a Lady of Vix living in a palace above the Seine, in an age so remote and poorly documented that it can scarcely be called protohistoric, the lands of the Celts were divided into small tribal enclaves. One of the few signs of inter-regional coordination in this period is a name. ‘Mediolanum’ was one of the commonest and oldest place names in the Celtic world. It occurred over a vast area, from the Irish Sea to the Black Sea, and from Spain to northern Germany, but principally in Gaul.* Some of the places called Mediolanum eventually became Roman towns; others vanished altogether. Within a few generations of the Roman conquest, ‘Mediolanum’ was an obscure and baffling term, its meaning known, presumably, only to the Druids who lived in exile or who stayed behind secretly in Gaul. Educated Romans – even those who had a smattering of Gaulish – had no idea what it meant. In the fourth century AD, the citizens of a town in northern Italy called Mediolanum (now the city of Milan) decided that ‘medio – lanum’ must be a Latin term signifying ‘half wool’ and adopted as their coat of arms a weird and impossible creature wearing half a fleece. Not long afterwards, the Gaulish language was extinct, and almost nothing remained to indicate that the Mediolana were relics of the earliest attempts by Celtic tribes to map and organize their territory.
After two hundred years of scholarship, the Gaulish language is better known today than it was at the end of the Roman empire. The literal sense of the word ‘Mediolanum’ is now well established. The Gaulish dictionary compiled by Xavier Delamarre defines it as ‘a term of sacred geography’: ‘a holy centre . . . perhaps a central point of reference on the vertical axis of the three worlds – upper, middle and lower’. For students of the Celts, this is familiar ground. In Celtic mythology, ‘middle’ was a three-dimensional term. It referred not only to the earth that lies between the upper and the lower worlds, but also to the intersection of lines based on the cardinal points. According to Celtic legend, Ireland was divided in the first century AD into four kingdoms, each of which gave a part of itself to form a fifth, central kingdom called Mide (or Meath), signifying ‘middle’. This is the cruciform pattern that can be seen on the ceremonial Celtic bronze spoons which began to appear all over Europe in about 800 BC. It usually takes the form of two perpendicular lines with a circle at the centre. The so-called Celtic cross of the early Christian Church is probably a direct descendant of those designs.
The concept of a geographical and symbolic centre – of a temple, a town, a nation or the entire earth – is common to most Indo-European religions. The Druids, according to Caesar, met at a particular time of the year in ‘a region considered to be the centre of the whole of Gaul’. In ancient Greece, the equivalent of a ‘holy centre’ was the omphalos or ‘navel’ stone, which marked the centre of the world, the most famous being the omphalos at Delphi. The exact location was said to have been discovered when Zeus released two eagles or crows at opposite ends of the earth: flying in straight lines at the same speed, the birds met at the place that became Delphi. In Jerusalem, there are two omphaloi: the Temple Mount, where Adam was created, and the rock of Golgotha in the Church of the Holy Sepulchre, where Jesus Christ was crucified. On medieval maps of the world, the omphalos of Jerusalem is placed at the centre of the earth, with Paradise at the top and the Pillars of Hercules at the bottom.
No such map of the Celtic world has come to light – or nothing that has been recognized as a map – and a glance at the distribution of Mediolana shows why the search for a ‘sacred centre’ of the Druids has so far proved fruitless. The map of Mediolana printed here is the most detailed yet established, but still no pattern can be seen. The problem is, there are ‘sacred centres’ or ‘middle sanctuaries’ all over the Celtic world: Herakles would have passed by six of them on his way to the Matrona Pass, and Caesar’s campaigns would have taken him to within a few Roman miles of at least a dozen Mediolana. Some of the Mediolana are now towns or villages, but most of them are so insignificant that they exist only as a small farm or a field. A ‘Melaine’ in the Compiègne forest was identified as a Mediolanum only when two lead coins of the third century AD were found there inscribed with the abbreviations ‘MED L’ and ‘MEDIOL’. A Mediolanum that was a small island in the Atlantic Ocean near the Île d’Aix was swallowed by the waves some time after 1430. Another Atlantic Mediolanum – a headland near Carnac called the Pointe de Meylant – was never recorded on a map: it turned up during research for this book in a guide for coastal pilots published in 1763.
8. Mediolana
The majority of these ‘central sanctuaries’ were in Gaul, but there were also Mediolana in England, Wales, Belgium, Germany, Switzerland, Italy and Spain. The furthest flung (fig. 43) were in Lower Austria and Bulgaria. The best known – but not the earliest – is the city of Milan. In all, thirty-six places are known for certain to have been called Mediolanum. There are also one hundred and one other places for which no documentary or epigraphic evidence survives but whose names – Meaullens, Miolan, Molliens, etc. – are almost certainly derived from ‘Mediolanum’. (For a complete list, with coordinates, see www.panmacmillan.com/theancientpaths.)
Châteaumeillant is a relatively straightforward example. Twenty-three of its earlier forms are recorded: Mediolanum became Mediolens, then Melianum, Millandum, Maiglen and finally, like a pebble tumbled and smoothed by a river, Meillant. The original name is cited by Gregory of Tours (sixth century AD) and on a thirteenth-century map of the world (the Peutinger Map) based on ancient ‘itineraries’ or lists of staging posts dating from the fourth century AD. The oldest remnant of all can be seen by anyone who cares to stand with a small ladder in the middle of the route nationale to Paris in the village of Bruère-Allichamps, where a third-century milestone marks one of the supposed centres of France. Near the top of the column, a Roman inscription indicates that ‘MED[IOLANUM]’ lies ‘L XII’ (twelve Gaulish leagues or 32.5 kilometres) from the milestone.
Many of the Mediolana are places that a traveller might well describe as ‘the middle of nowhere’. It is extremely hard to tell why these unimportant places bore such a portentous name. For several centuries, Iron Age scholars and the seekers of mystical truths known to French archaeologists as ‘Celtomaniacs’ puzzled over the meaning of this tantalizing term, which combines geography with religion. Theoretical expeditions returned empty-handed or in disarray. According to some, ‘Mediolanum’ meant ‘middle of the plain’ (because Milan is in the middle of a plain): this is disproved by the fact that most Mediolana are on hills or in hilly terrain. Others claimed that the Mediolana were tribal centres: but some Mediolana are only a few kilometres apart and none is in the centre of a Celtic tribal territory.
One of the most persistent attempts to explain the mystery was made by a professor of literature at the University of Bordeaux. Between 1972 and 2000, Yves Vadé developed a theory that these ‘middle sanctuaries’ had been part of a system of coordinated points. In Vadé’s reconstruction of the hypothetical pattern, one of the key points was Châteaumeillant, which lies approximately the same distance from Meilhan-sur-Garonne as from Évreux (formerly Mediolanum). These two places in turn are more or less equidistant from Milan (724 and 728 kilometres). Observing that a few other miscellaneous sites of Celtic and pre-Celtic significance lay close to the lines connecting these points, Vadé drew various combinations of equilateral and isosceles triangles over the Gaulish part of Europe. ‘Unless the maps deceive us’, he concluded, the places called Mediolanum represented a spectacular application of the principles of sacred geography.
Unfortunately, the maps do deceive us. A map is a flat representation of part of a sphere. This translation of a curved surface into two dimensions inevitably produces a certain distortion. For the same reason that a small piece of orange skin can be flattened without tearing its edges, the distortion is insignificant over short distances, but on the scale of Vadé’s ‘system of equidistances’, sizeable discrepancies occur. As a result, some of his triangles are skewed by several degrees and some of his equidistances are out by several kilometres. With the correct equations and projections, it is perfectly possible to replicate the kind of survey that an ancient civilization was capable of producing (p. 147). But even then, no pattern emerges, which is hardly surprising: Vadé assumed that practically all the Mediolana are known, but since relatively few protohistoric place names survive, even the one hundred and thirty-seven Mediolana shown on the map above must be a small minority of the original total.*
Most other Celtic historians who pondered the Mediolanum mystery sensibly added it to all the other unanswerable questions of Iron Age history and hoped that, one day, the mystery would be solved by archaeology.
Surprisingly, archaeology has provided a kind of answer – but an answer so multifarious and unexpected that it took me several months of research even to notice its existence. The stupendous reference work known as the Carte archéologique de la Gaule is now at its one-hundred-and-nineteenth undigitized volume. It describes every archaeological find, however small or dubious, made on the territory of modern France. Again and again, it shows that the places once called Mediolanum do indeed have something other than a name in common. The entry for Meylan (Lot-et-Garonne) is fairly representative: ‘some iron slag’ was found and nothing else. At Molain (Jura), there is a pre-Celtic burial mound, and then the record is blank apart from some Roman coins. In 1873, at the Mediolanum near Pontcharra on the road to Lyon, a subterranean passage was discovered: it led to a small dry-stone chamber; at intervals along the passage, holes for lamps had been carved and there were traces of smoke; but it turned out to belong to a Gallo-Roman dwelling. Despite its Celtic name, Miollan appeared not to have existed until the Roman conquest. For many other Mediolana, there is no entry at all, or just a terse ‘nothing to report’.
Bizarrely, the one distinguishing feature of these ‘sacred centres’, apart from the obviously Celtic name, is the absence of Celtic remains. With a single exception (p. 81), even the few Mediolana that became important towns after the Roman conquest have a void in their history. The Mediolanum which is now the city of Saintes in western France prides itself on its independent Gaulish past, and yet, after several decades of intense excavation, it has not a single museum-worthy object with which to celebrate its pride, not even a Celtic coin, just an undatable potsherd and a small piece of harness that might have come from somewhere else.
Despite this deluge of insignificance, it seemed impossible that the places called Mediolanum had nothing in common but a lack of Celtic evidence and a name – an oddly ineradicable name, attached to nothing in particular, yet as durable as a stone temple – and so, in 2008, remembering the six Mediolana that lie along the Via Heraklea, I decided to include as many ‘middle sanctuaries’ as possible in each cycling trip to France. The first sign of something out of the ordinary was the fact that it was exceptionally difficult to fit them into long-distance routes. Many of them were in awkward, out-of-the-way locations, and few of them had anything of interest apart from the name on the road sign. To judge by the raised eyebrows of farmers and dog-walkers, no one since the Iron Age had ever asked for directions to Les Miolans or Le Mayollant, and no tourist had ever stopped to photograph the muddy pastures of Maulain or the monolithic grain silos on the otherwise featureless hill of Montmeillant.
In the Burgundian hamlet of Meulin (cited in a ninth-century cartulary as ‘Mediolanensis ager’), the GPS unit went blank and never recovered. This was certainly due to the five hours of torrential rain that fell between Paray-le-Monial and Mâcon, but it seemed quite appropriate: even with weeks of advance planning, many of the Mediolana were tricky to find and even harder to leave. On the map, the route looked simple, but on the ground, the ‘first left’ or ‘second right’ turned out to be a plausible lane, which ended in an open field. Tracks led off at odd angles and veered in a different direction as soon as they left the village. There was often a choice of two apparently identical onward routes. The church, if there was one, sometimes appeared to belong in a different setting, as though its original village had been lost to another dimension. Yet once the site had been identified, the church spire would remain in view for several kilometres.
By the tenth Mediolanum, this much was obvious: the typical Mediolanum was a tiny place on a hilly site, devoid of Celtic remains, with a disproportionate capacity to bewilder and confuse, set apart from any natural trade route or long-distance road, and inhabited by an agitated dog who knew that any visitor was in the wrong place. None of them looked as though it had ever been a Celtic Jerusalem or Delphi.
It was only later that I learned of a form of cartographic analysis that might have explained the confusion or made it easier to predict. The technique, developed by the archaeologist Éric Vion, is based on the fact that once a route has been created, it almost never disappears completely: thousands of years after it was first tramped out, it might survive as a track, a field boundary or the edge of a wood. It follows that a large-scale modern map is not just a snapshot of the network at a given moment, but a multi-dimensional scan that reveals all the successive strata which make up the present roadscape. Just as an oak tree tells the story of its life in growth-rings and twisted boughs, the whole road system – rather than just the segments that are known to be ancient – is a record of its own evolution.
Even the simplest application of the technique can produce some exciting results. The first step is to identify all the junctions in a given area. Next, one simply counts the number of branches at each junction, including tracks as well as roads. The junctions with the largest number of branches (say, ten or more) are nearly always important places: these are the cities and large towns which remain visible when the map is zoomed out, and which are directly connected to one another over long distances. If the junctions with fewer branches are added, then the small towns and large villages with more local connections appear, and so on, until the whole populated landscape is accounted for.
Sometimes, in this apparently logical system, anomalies appear – a crossroads in open country which turns out to mark the site of an abandoned village or a Roman marching camp; a prehistoric trail that meanders like a sleepwalker, missing out all the villages and cutting across later routes. There may also be some many-branched crossroads that belong to an earlier network. Some of the branches may have lost their original purpose and yet survive in truncated forms. Isolated from the rest of the map, these ancient intersections look like fallen trees with their roots in the air.
Most of the places called Mediolanum fit this description, which explains why it was so easy to get lost in their vicinity. As it entered and left each ‘middle sanctuary’, the bicycle, unconsciously reading the coded messages of the road system, was slipping in and out of different eras, plunging with a turn of the handlebars into the distant age when Greek sailors were founding trading posts on the Mediterranean and when the solar trajectory of the Heraklean Way was first projected over half a continent.
* The Greek and, probably, Gaulish, form was ‘Mediolanon’. The better-known Latin form, ‘Mediolanum’, is more commonly used. The plural is ‘Mediolana’. ‘Lanon’ is unrelated to the Welsh ‘llan’ and Breton ‘lann’ (an enclosed piece of land, especially the site of a monastery or church). The nearest Welsh equivalent is ‘llawn’ (‘full’ or ‘complete’).
* An additional problem is that the names on maps can be just as deceptive as the maps themselves. Toponymy is like a dream in which familiar faces turn out to belong to strangers. The city of Milan owes its name to ‘Mediolanum’, but most of the other places called Milan have an entirely different origin. A village that was called ‘Moulins’ in the eighteenth century (now Molain) is mentioned in an eleventh-century document as ‘Villa Mediolanis’, but all the other ‘Moulins’ are mills. The etymological record shows that, of the thirty-nine ‘middle sanctuaries’ included by Vadé in his calculations, sixteen were never Mediolana. One of the key points in his system is Meilhan-sur-Garonne, which, though it looks like a name derived from ‘Mediolanum’, was probably the estate of a Roman citizen called Aemilianus.