The Celts
in a Hurry

The Whole Thing in One Page

The Battersea Shield is famously labelled Celtic. It is an ornate bronze facing that once covered a wooden shield, and no inscription identifies its maker or bearer. Its workmanship proves exceptional craft and display; its recovery from the Thames may reflect a deliberate offering. It is not a membership card.

The familiar picture joins torcs, druids, knotwork, hillforts, blue-painted warriors and resistance to Rome into one civilisation spread across ancient Europe. Those pieces belong to different regions and centuries. There were ancient Celts, and Celtic languages once covered an immense range. There was no Celtic nation, empire, church or shared political command.

The word hides several questions. Who used a Celtic name? Who spoke related languages? Who made objects in related styles? Who claimed common ancestry? Who later chose the inheritance? The answers overlap without producing one border. Greek and Roman writers used Keltoi, Galatai and Galli for changing populations. Linguists recover a family stretching from Iberia to Anatolia. Archaeologists use Hallstatt and La Tène for material patterns that cross political and linguistic boundaries.

Inside those patterns were regional societies. Most people farmed, raised animals, spun cloth, worked wood and iron, traded and lived within communities whose politics were close to home. Elites displayed rank through feasts, wagons, weapons, gold and imported wine. Bibracte gathered workshops, storage, coinage and political authority. In Britain, the Snettisham Great Torc put more than a kilogram of gold alloy around a neck. In Ireland, the Broighter hoard joined a gold collar to a miniature boat.

Kinship adds another map. At sampled southern British cemeteries, a long maternal line ran through the community while newcomers were mostly male. Comparisons with other sites are consistent with matrilocal residence in more than one region. The finding concerns where partners lived, not who ruled, and cannot be projected across Celtic-speaking Europe.

Religion was local. Communities placed offerings in rivers, lakes, pits and sanctuaries and worshipped gods whose names changed by region. Druids were learned figures with religious and legal roles in parts of Gaul and Britain, not a continental church. Ritual killing best explains some cases, but no evidence turns every bog body, severed head or violent burial into one Celtic rite.

People moved at several scales. Celtic-speaking groups crossed the Alps, entered the Balkans and founded Galatian communities in Anatolia. Trade, marriage, raiding and military service moved people and practices farther. Language, genes and artefacts did not travel together in one wave.

Rome brought imperial resources into these local worlds. It used manpower, engineering, supply, professional command, terror, diplomacy and existing rivalries. Caesar turned intervention into conquest; Vercingetorix assembled a broad coalition only after years of Roman advance and lost at Alesia. Rome later conquered much of Britain and crushed Boudica's revolt. Political plurality made coordination harder, but division alone did not build the siege works or replace Roman armies.

The endings differed. Gaulish faded under Latin. Ireland remained outside Roman rule. Brittonic survived in western Britain and travelled to Brittany. Christian scribes preserved older stories in altered forms. Later scholars, artists and political movements gathered languages, literature and selected artefacts into a wider Celtic identity unknown to any Iron Age community.

The inheritance is real because its materials are real. It is reconstructed because history did not preserve them in one piece.

That is the book.

Why You Should Care

Around 470 BCE, a woman was buried at Vix in Burgundy on a four-wheeled wagon, wearing gold and surrounded by objects drawn from several worlds. Beside her stood a Greek bronze vessel 1.64 metres high, made for mixing wine and capable of holding about 1,100 litres. Mediterranean craft, northern materials and local authority met in one grave. Her precise role is unknowable, but exceptional female status is not.

Vix corrects the ancient Europe many people carry in their heads. Beyond Greece and Rome lies a grey margin of forest tribes, waiting to be conquered and described by somebody literate. The grave belongs to another world: wealthy elites, long-distance exchange, specialised craft, ceremonial display and communities capable of drawing remarkable objects hundreds of kilometres from where they were made. Rome did not bring history to an empty continent.

The Celts also rearrange the linguistic map. Related languages were once spoken from Iberia and Gaul to northern Italy, the Balkans and central Anatolia. Their living descendants now survive mainly along Europe's Atlantic edge. What appears marginal today once extended across much of the continent, and its retreat records conquest, prestige, migration and language change over more than two thousand years.

The subject then exposes a problem hidden inside almost every ancient history: how do we decide that scattered people belonged together? Language, outsider names, ancestry and material styles provide different evidence. Confusing them can turn a sword into proof of a nation or a language family into one political identity. Keeping them separate does not dissolve the Celts. It shows what the category can support.

Britain and Ireland make that problem tangible. The Snettisham Great Torc weighs more than a kilogram and carries a late Iron Age British version of La Tène art. The Broighter hoard, found near the old shore of Lough Foyle, contains an Irish gold collar and a miniature boat. Both belong to the wider subject. Neither proves that their makers used the same name, obeyed the same rulers or understood the designs identically. Connection is visible. Uniformity needs evidence.

Burials can now reveal relationships as well as objects. At Durotrigian cemeteries in southern Britain, ancient DNA found a long maternal line and predominantly male incomers. That is strong evidence for matrilocal residence in those communities, while comparisons suggest a wider British pattern. It says nothing directly about queenship, inheritance or Celtic-speaking Europe. Kinship, residence and political power remain separate questions.

The Roman story changes in the same way. Caesar did not defeat one country called Gaul. He entered a field of Aedui, Arverni, Sequani, Helvetii and many others whose rivalries and alliances could be used. Yet local division is only part of the explanation. Rome brought experienced armies, replacement capacity, engineers, supply systems and a commander whose career profited from continued war. Local choice and imperial force operated together.

The subject restores ordinary life beneath the military account. Most people produced food, raised animals, made textiles, worked wood and iron, carried salt, traded and maintained households. Some communities built dense fortified centres with streets, storage, workshops and coinage before Roman armies arrived. Religion appears through sacred water, local gods, feasts, human remains and valuable objects removed from use, rather than through one druidic church.

The name then outlived the ancient political world. Celtic languages remained strongest in Ireland and western Britain, medieval writers preserved traditions already altered by Christianity, and later movements built a wider inheritance from them. The useful task is to distinguish what was transmitted, recovered, adapted and invented without demanding perfect continuity or total fraud.

The book moves from salt mines, graves and workshops to Italy, Galatia, Atlantic Europe, Alesia and Roman Britain, then into the languages and legends that survived. Rome is no longer the whole foreground, and the Celts are no longer its scenery.

The Core Ideas

A Name Without a Nation

Caesar begins his account of Gaul with a sentence that appears to settle the matter. One part of the country, he says, is inhabited by people who call themselves Celtae and whom Romans call Galli. There were people in antiquity who used a Celtic name. The trouble begins when that sentence is stretched across half a continent and made to cover everyone later described as Celtic.

Ancient identity worked in layers. A man might be an Aeduan because that was his political community, a Gaul when contrasted with Romans, and perhaps a Celt in a broader setting. The same person could belong to a kin group, household, client network and religious community without treating any one label as the complete answer. Modern nationality encourages the opposite habit. We expect one people to possess one territory, one government and one name, then become suspicious when the past refuses the arrangement.

The ancient labels were never stable enough to supply it. Greek writers used Keltoi and Galatai for different populations at different times. Roman writers preferred Galli, then applied related names to people in northern Italy, Gaul and Anatolia. Some communities were included because of where they lived, some because of language or custom, and some because outsiders found them similar. The map moved with the writer.

Local names mattered more in daily politics. The Aedui, Arverni, Sequani and Helvetii of Gaul made treaties, fought wars and pursued rivalries as distinct peoples. In Britain the Iceni, Brigantes, Catuvellauni and dozens of others did the same. None sent representatives to a Celtic parliament. There was no common king, law code, military command or priesthood governing them all. Shared speech and custom could make negotiation easier, but they did not erase the interests of the polity collecting tribute, defending land or trying to dominate its neighbours.

Untidy borders do not make the category unreal. Ancient writers identified Celts and Gauls, Celtic languages formed a real family, and names, artistic forms and religious practices travelled across wide networks. The useful model is graded resemblance. Communities could share speech while differing in burial, or exchange the same brooch style across a linguistic boundary. Broad affinities could matter without replacing the local allegiance that decided taxes, war and law.

That distinction changes the whole story. The Celts did not fail to build a country that ought to have existed. They inhabited a world in which identity was nested and authority dispersed. What later looks like disunity was normal political structure. Rome would exploit it, but Rome did not reveal a missing nation. It confronted many functioning societies that had never agreed they were one people.

Language Is the Strongest Thread

Five is cúig in Irish and pump in Welsh. The words look distant until the same sound difference appears again and again: an ancestral consonant remained k-like in Goidelic and became p in Brittonic. One small comparison opens a relationship that no shield, grave or classical label can establish by itself.

The most defensible way to draw a Celtic map is therefore to listen rather than look. Celtic is a branch of the Indo-European language family, related at a deep level to Latin, Greek, Germanic, Slavic and several other branches. Its ancient speakers left no common literature and no single inscription announcing their unity, but the languages preserve patterns of sound, grammar and vocabulary too systematic to be accidental.

The surviving fragments once covered a remarkable distance. Lepontic was written in northern Italy. Gaulish appeared in inscriptions and names across much of Gaul. Celtiberian was recorded in the interior of the Iberian Peninsula using an adapted local script. Galatian was spoken in central Anatolia by descendants of migrants who had crossed the Balkans. Ancient British and early Irish belong to the same family, though the precise routes and dates by which Celtic speech spread across Atlantic Europe remain contested.

The evidence is often small but cumulative. A dedication scratched into pottery, a coin legend, a personal name on a Roman tombstone and a place-name preserved by a later language can each reveal part of the pattern. Gaulish inscriptions used Greek and then Latin letters. Celtiberian inscriptions adapted signs developed for Iberian languages. The alphabet changed because writing was borrowed from neighbours; the linguistic structure underneath did not.

The surviving languages are commonly grouped into branches whose descendants still exist. Goidelic includes Irish, Scottish Gaelic and Manx. Brittonic includes Welsh, Cornish and Breton. Breton needs one correction of its own. It is not the last direct survival of ancient Gaulish in France. It descends mainly from Brittonic speech carried from Britain to Armorica during late antiquity and the early Middle Ages. Even the modern map therefore records migration rather than a population standing still.

Language is stronger evidence than a decorated shield because linguistic relationship can be demonstrated through recurring correspondences. It is weaker than a passport because speech does not dictate political loyalty, ancestry or self-description. People change language through trade, intermarriage, conquest, prestige and administration. A Celtic-speaking population could absorb newcomers. A community could retain much of its material life while adopting Latin. Genes do not conjugate verbs.

This also explains why archaeology cannot locate the first Celtic speaker with confidence. Pots and swords can move without language. Languages can spread without a mass migration visible in cemeteries. The old picture imagined a Celtic homeland in central Europe whose people carried Hallstatt and La Tène culture westward in waves. Movement certainly occurred, sometimes on a large scale, but material style cannot be used as a substitute for speech. The first people to make a La Tène brooch were not required to pass a language test.

The linguistic thread is firm and modest. It shows that populations from Iberia to Anatolia once spoke related languages without proving one origin story, political identity or uniform culture. That is enough to give Celtic a defensible centre. A language family can connect societies that never formed a nation and remain discoverable after the governments that once spoke it have disappeared.

Hallstatt and La Tène Were Networks, Not Countries

Hallstatt is a lakeside town in the Austrian Alps. La Tène is a site beside Lake Neuchâtel in Switzerland. Their names now cover centuries and thousands of kilometres because archaeologists needed labels for recurring patterns in graves, settlements, weapons and art. The labels are useful. Turning them into peoples is where the trouble starts.

Hallstatt usually describes much of the earlier Iron Age in central Europe, especially from around the eighth to the fifth centuries BCE. The original site grew wealthy from salt. Mining preserved timber, leather, textiles and human waste, revealing dangerous labour as well as distant trade. Its cemetery contained burials with marked differences in wealth. Salt could be cut from the mountain, carried across Europe and exchanged for goods from far beyond the Alps.

Other centres show what those routes produced. Around 600 BCE the Heuneburg on the upper Danube acquired a whitewashed wall of sun-dried mud brick built with Mediterranean knowledge. Thousands lived on and around the fortified height. At Vix, an elite burial contained one of antiquity's largest bronze wine vessels. Wine, wagons, metalwork and jewellery moved through chains of exchange linking northern rulers to Etruscans and Greek colonies.

None of this creates a Hallstatt kingdom. The centres competed, rose and declined. Local burial customs varied. Most people lived in farmsteads and villages beyond the princely mounds. The common pattern came from routes, technologies, elite tastes and imitation rather than central government. A Greek vessel in Burgundy records movement and appropriation, not a change of nationality.

La Tène begins as a different material pattern from around the fifth century BCE. Its metalwork favoured curves, transformed plants, animals and faces that appear to shift as the object turns. Swords, scabbards, torcs, brooches, vessels and horse equipment carried versions of the style across Europe. Its makers borrowed from Greek, Etruscan, steppe and local traditions, then altered what they borrowed.

The Snettisham Great Torc shows what happened at the Atlantic edge. Made in Britain during the first century BCE, it weighs 1,084 grams and combines a heavy twisted neck ring with elaborate terminals. It belongs in a La Tène artistic history while remaining intensely regional. The Broighter collar in Ireland belongs there too, though its construction and imagery follow another local route. A style was a conversation in metal, not a uniform issued by a nation.

The type site of La Tène produced weapons and other objects placed beside or into water. The name later expanded far beyond the community that deposited them. La Tène-style pieces appear where ancient writers identified Celtic speakers and where they did not. Britain developed its own forms. Ireland received the style selectively. A spiral can show contact, prestige or imitation before it shows language.

By the second and first centuries BCE, large fortified settlements called oppida appeared across parts of temperate Europe. Bibracte contained workshops, imported goods, coinage and public spaces. It belonged to a wider urban change rather than a single Celtic plan. Some Celtic-speaking societies built large towns. Others remained dispersed.

Hallstatt and La Tène organise the archaeological record. They become misleading when painted as successive Celtic empires or exact maps of speech. The objects reveal societies watching, borrowing from and competing with one another across political boundaries. The classification is scaffolding. The past is the building behind it.

Farmers, Craftspeople and Warrior Elites

An Iron Age household owned few objects that could survive two thousand years. Timber rotted, cloth vanished, food was eaten and tools were repaired until little remained. A sword placed in a grave had the opposite career. It was protected from reuse, preserved by corrosion and found by an archaeologist. The result is a distorted population in which nearly everyone seems armed and nobody spends the morning grinding grain.

Most people in the societies called Celtic were farmers. They grew cereals, tended cattle, sheep and pigs, stored food against winter and managed land whose yield decided whether a household survived. Settlement varied by region. Britain had roundhouses and enclosed farmsteads. Central Europe contained rectangular buildings, open villages, hilltop centres and later towns. Ireland followed its own patterns. There was no standard Celtic house waiting to be copied from the Atlantic to Anatolia.

Work filled the year. Grain had to be sown, harvested, threshed and ground. Wool had to be cleaned, spun, dyed and woven. Timber became buildings, carts, bowls and fences. Clay became cooking pots and storage vessels. Iron ore required fuel, furnaces and skill before it became a knife, plough fitting or sword. The metalwork in museums depended on miners, charcoal burners, smiths and long apprenticeships whose practitioners rarely received the burial of the patron wearing the finished torc.

Specialisation increased around major centres. Hallstatt's salt mines organised underground labour. Heuneburg drew traders and craftspeople into a dense settlement. Oppida such as Bibracte contained workshop areas, coin production and markets linking town to countryside. These places required decisions about walls, streets, storage, water, production and access. They were not Roman cities, and they were not primitive camps.

Society was unequal. Elite graves contained wagons, imported wine vessels, gold, weapons and feasting equipment. Followers gathered around powerful households through kinship, debt, patronage and the expectation of reward. Caesar's portrait of Gaul divides society too cleanly, but his nobles with clients and armed retainers fit wider evidence for competitive aristocracies. Wealth was displayed because rank had to be recognised.

New evidence has made kinship less abstract. A 2025 study of 57 genomes from Durotrigian burials in southern Britain found an extended group centred on one maternal lineage, while unrelated incomers were predominantly male. Comparisons with other British Iron Age cemeteries found unusually dominant maternal lines. The evidence supports matrilocal residence as a recurring British pattern, with women often remaining near their birth communities and men moving towards them.

A separate 2024 study linked elite Hallstatt burials across three centres in southern Germany, including a probable maternal-uncle and nephew relationship between two exceptionally rich graves. The authors argue that this may indicate matrilineal dynastic succession in that narrow elite setting. Put beside the British evidence, it makes one limit clear: residence, descent and succession are different institutions. None by itself proves matriarchy, equal power or a common Celtic kinship system.

War remained important. Raiding could produce cattle, captives, prestige and land. Weapons marked rank and masculine identity in many regions, though not in identical ways. Gallic and Celtiberian forces fortified, negotiated and adapted tactics rather than charging without thought. Captivity and slavery moved people as property and distributed labour unequally through households, estates and retinues.

The warrior image is therefore neither false nor sufficient. These were agrarian societies in which violence, craft, kinship and status were tied to control of surplus. The sword survives because somebody made it, somebody paid for it and many others produced the food that allowed both of them to stop farming long enough to care.

Gods, Druids and Offerings

At Llyn Cerrig Bach on Anglesey, people placed weapons, chariot fittings, tools, chains and decorated metalwork into a lake over several centuries. Many objects had been bent or broken before deposition. Valuable things were removed from ordinary use and given to a place from which they were not meant to return.

That action tells us more about Iron Age religion than a modern drawing of a white-robed druid. Sacred practice attached power to springs, rivers, lakes, groves, pits, boundaries and human remains. Communities deposited weapons after conflict, animal bones after feasts, jewellery, coins and tools. A gift could thank a deity, fulfil a vow, mark a transition or secure favour. The exact words are lost, but the sacrifice of value is visible.

The gods were numerous and often local. Roman-period inscriptions preserve hundreds of divine names, many appearing once or in one region. Some travelled; others belonged to a spring, mountain or community. Under Roman rule local gods were paired with Roman ones, producing Sulis Minerva at Bath and many less famous combinations. Such pairings were acts of translation, not proof that the gods had always been identical.

Druids are historical, but almost everything precise about them comes through outsiders. Greek and Roman writers describe specialists in Gaul and Britain concerned with ritual, teaching, memory, law and disputes. Caesar says their education could last twenty years and that they transmitted learning orally. His account is valuable and strategic. He was describing societies he intended to conquer in forms useful to Roman readers.

The safest model is a learned class or set of roles in parts of Gaul and Britain, influential without forming a single European church. Caesar mentions druids mainly in an ethnographic digression and gives them little direct part in his campaign narrative. That silence is evidence about Caesar's book before it is evidence about Gallic politics. Their organisation, regional reach and role in resistance remain harder to recover than their existence.

Archaeology can identify sanctuaries, offerings and unusual treatment of the dead. It cannot usually identify the profession of the person supervising them. Llyn Cerrig Bach may belong to the sacred importance of Anglesey, where Tacitus later placed druids, but the objects do not carry job titles. Ireland has no contemporary text proving a druidic hierarchy, however prominent druids became in later literature.

Ritual killing belongs in the account because some cases are hard to explain otherwise. At Winterborne Kingston in Dorset, a young woman was killed by a cut to the neck, possibly while bound, and placed face down over a carefully arranged layer of animal bones inside a storage pit. That is powerful evidence for a ceremonial killing in one later Iron Age community. It does not turn every violent burial into the same rite.

Bog bodies, displayed skulls and disarticulated remains require the same discipline. Warfare, execution, punishment, mortuary practice and ritual could overlap. Classical writers turned northern violence into proof of barbarism, while older archaeology often converted every unusual death into druidic sacrifice or a Europe-wide cult of the head. The evidence now supports varied regional practices rather than one Celtic explanation.

Stonehenge supplies the cleanest correction. Its main phases were built roughly fifteen centuries before the first historical references to druids. Early modern antiquaries assigned the monument to the available name for Britain's learned pagan past. The historical druids are more interesting once they are removed from a monument they did not build.

Iron Age religion lives in damaged weapons beneath water, local gods, feasts, severed heads, strange burials and obligations whose words have vanished. Druids belonged inside that world. They did not own all of it.

Rome Brought an Empire into Local Wars

When Caesar entered Gaul in 58 BCE, he did not face one state. He faced the Helvetii moving through territories linked to the Sequani and Aedui, while Aeduan leaders disagreed over whether Rome was rescuer, rival or route to personal power. One brother, Diviciacus, worked closely with Caesar. Another, Dumnorix, cultivated popular support, marriage alliances and armed followers while resisting Roman influence. The family contained one part of the political problem in miniature.

Rome did more than exploit division. Caesar brought an experienced army backed by a state that had conquered the Mediterranean. Roman forces could build roads, bridges, camps and siege lines at speed, move supplies through provinces and allied territory, replace losses, and punish a defeated community severely enough to alter the calculations of its neighbours. Caesar also had a private reason to keep moving. Command in Gaul produced wealth, reputation and soldiers whose loyalty would later decide Roman politics.

Local rivalry still mattered because imperial power entered through it. The Aedui had long claimed the title of brothers and kinsmen of Rome. The Arverni had competed with them for influence. The Sequani invited German support against rivals, then discovered that an armed helper could become a new threat. Communities supplied grain or cavalry, accepted hostages, changed sides and tried to preserve rank. Caesar did not invent these interests. He converted them into a sequence in which every settlement justified the next intervention.

This was not evidence of defective peoples unable to recognise a common interest. Each polity faced its own balance of danger. Joining a coalition might protect the wider region while placing troops under a rival aristocrat. Supporting Rome might defeat an immediate enemy and preserve local privilege. Waiting could look prudent until the neighbouring state fell. The same decision could be rational locally and ruinous collectively.

Once a community accepted Roman protection, the decision changed its neighbours' options. Hostages, grain demands and garrisons could turn a temporary alliance into infrastructure. A local bargain accumulated at imperial scale: each agreement supplied men, food or access for the next campaign.

Vercingetorix came closest to reversing the process. In 52 BCE he assembled a broad coalition, enforced a scorched-earth strategy and defeated Caesar at Gergovia. The coalition was negotiated, contested and late, yet it mobilised on a scale that forced Rome into its hardest campaign. At Alesia, Caesar enclosed the defenders with one fortified line and built another against the relieving army. Gallic political plurality helped him reach that moment. Roman engineering, discipline, cavalry, supply and command won it.

Britain repeated part of the pattern without copying Gaul. Caesar's expeditions achieved submissions but no conquest. A century later Roman armies combined battle with client rulers and dynastic alliances. Cartimandua of the Brigantes worked with Rome and handed over Caratacus. The Iceni remained a client kingdom until Roman seizure and abuse helped produce Boudica's revolt. Rome concentrated professional force against regional coalitions and maintained garrisons, roads and forts after victory.

Conquest did not end identity on contact. Roman Gaul remained full of Aedui, Arverni and other communities operating through towns, councils and cults. Latin spread unevenly. Local gods acquired Roman partners. In Britain, imperial power reached farther than cultural uniformity and never crossed into permanent rule of Ireland.

Rome held Gaul for centuries and much of Britain for nearly four hundred years. It won through its own resources and through the political worlds it entered. Removing either half produces a comforting myth: unstoppable legions on one side, people defeated only by disunity on the other. The harder model keeps Roman capacity, coercion, local choice and contingency in the same campaign.

The Celts Were Rebuilt After Antiquity

By the early Middle Ages, the ancient map had been transformed. Gaulish had faded under Latin. Celtic speech survived most strongly in Ireland and western Britain, while Brittonic migrants carried their language to Brittany. Galatian disappeared in Anatolia. The people involved did not normally call themselves Celts. They were Irish, Britons, Welsh, Cornish, Bretons and members of smaller kingdoms whose concerns belonged to their own time.

These survivals followed different routes. Ireland remained outside Roman conquest, though never outside contact. Western British communities retained Brittonic speech through and after the end of Roman government. Migration carried related speech to Armorica. Geography, state power, prestige and population movement mattered more than the absence of a single Celtic capital. Political fragmentation did not preserve Gaulish, and Roman unity did not prevent Latin from later dividing into many languages.

Christian scribes preserved bodies of Irish and Welsh story that became central to modern ideas of Celtic mythology. The survival is remarkable and requires a warning against false purity. The manuscripts were written centuries after the Iron Age by authors inside Christian literary cultures. They preserved names, places, themes and narrative structures from older traditions while reshaping them through later politics and belief. Cú Chulainn and the Mabinogi are inheritances, not transcripts from a druidic archive.

Art followed another route. The interlace, illuminated manuscripts and stone crosses now sold as timeless Celtic design belong mainly to the early medieval Christian world. They combined local metalworking traditions with Germanic, Mediterranean and ecclesiastical influences. La Tène curves and Insular knotwork can be placed in a history of repeated adaptation without becoming one style carried unchanged through a thousand years.

The modern Celtic category took shape through language. In 1707 Edward Lhuyd compared Welsh, Cornish, Breton, Irish and Scottish Gaelic and established their relationship with unprecedented clarity. The linguistic connection was real. Eighteenth and nineteenth-century writers then built more around it: common racial ancestry, national character, ancient wisdom and a shared artistic temperament. A family of languages became the foundation for a family of peoples.

Some additions were creative to the point of fraud. James Macpherson's Ossian poems drew on Gaelic traditions while presenting extensive composition as translations of an ancient bard. They became an international sensation and helped create the misty, melancholy Celt admired across Romantic Europe. Their doubtful authenticity did not reduce their political or artistic force.

The revival served different needs. Welsh scholars and cultural societies defended language and literature. Irish nationalists used Gaelic history against colonial accounts of backwardness. Scottish writers negotiated Highland identity after military defeat and social destruction. Brittany, Cornwall and the Isle of Man developed their own revivals. The six territories commonly grouped as the modern Celtic nations never shared one programme, but the label created solidarity and visibility across borders.

Modern archaeology then separated language, Hallstatt, La Tène, ethnicity and medieval art more sharply. The correction was necessary and sometimes turned into its own excess. The inability to show ancient Britons or Irish people routinely calling themselves Celts does not make modern Welsh or Irish Celtic identity fraudulent. Identities can be built from linguistic descent, historical memory, political choice and later reconstruction.

The causal loop closes more carefully than the legend allows. Ancient Europe supplied related languages, names, objects and practices without one central owner. Roman conquest, non-conquest, language shift, migration, Christian writing and modern revival then selected different parts. No one event preserved the whole because there was no whole in that form to preserve.

The modern Celts are neither an untouched ancient people nor a nineteenth-century fantasy. They are a living inheritance assembled from materials of unequal age. The reconstruction is part of the history, and it deserves the same scrutiny as the ancient past from which it draws.

How It Actually Works

Salt and the first centres

Around 800 BCE, miners were cutting salt from the mountain above Hallstatt in the Austrian Alps. Salt preserved what ordinary soil destroys: wooden tools, clothing, food remains and equipment used underground. The cemetery in the valley preserved another part of the system. Some people were buried with swords, wagons, bronze vessels and ornaments brought across long distances. A remote valley had become wealthy because its mountain contained something every household needed and few places could supply.

Hallstatt gave its name to a broad phase of the central European Iron Age, but there was no Hallstatt state. Power gathered unevenly around routes and resources. Salt, iron, amber and livestock moved between regions. Wine, bronze vessels and decorated pottery came north from the Mediterranean. Local rulers converted access into prestige, then displayed it through feasts, fortified settlements and burials whose scale made hierarchy visible after death.

At Vix in Burgundy, around 470 BCE, a high-ranking woman was buried with a wagon, gold jewellery and an immense Greek bronze wine vessel. The krater had travelled from southern Italy into a society that was neither Greek nor isolated. Mediterranean drinking equipment had become evidence of northern authority, while the wagon and jewellery placed that authority within a local language of rank.

The centres did not form one kingdom. They depended on routes, harvests and alliances that could shift. Most people remained in farmsteads and villages beyond the great mounds. Many communities within the connected zone probably spoke early Celtic languages, but the objects cannot identify which ones. They show a Europe organised by exchange, craft and political competition before classical writers supplied broad ethnic names.

A new style and a moving world

From about 450 BCE, the material pattern changed. Archaeologists call the later phase La Tène, after a lakeside site in Switzerland where weapons, tools and ornaments were deposited. Curving plants, animals and human faces replaced much of the earlier geometric decoration. New centres appeared from the Marne to Bohemia. The style travelled because objects, craftspeople, marriages and tastes travelled.

People moved as well, though never in one continuous wave. Some crossed a neighbouring valley. Some joined a war leader or served abroad for pay. Some communities relocated in large numbers. The movements recorded into Italy and the Balkans belong inside this wider mobility rather than proving that one nation left one homeland.

By the late fifth and early fourth centuries BCE, Celtic-speaking groups had crossed the Alps into the Po valley. The Insubres, Boii, Senones and others occupied land among Etruscan, Venetic and Ligurian populations. Around 390 BCE, an army associated with the Senones defeated the Romans at the Allia and entered Rome. Later Roman memory enlarged the sack into a national trauma, adding polished speeches and gestures long after the event.

The migrants founded or occupied settlements, farmed, traded, fought one another and served in Mediterranean armies. They did not remain a raiding population suspended outside ordinary life. Rome spent generations dealing with Celtic powers in northern Italy. By the early second century BCE it had broken the strongest, planted colonies and drawn the Po valley into its expanding system. Migration had created regional societies. Conquest later ended their independence without reversing the movement that formed them.

The road east

Celtic warriors appeared in Greek warfare before Celtic communities settled in the east. In the fourth century BCE, rulers hired them as soldiers. After 280, larger groups moved through the Balkans. One army killed Ptolemy Keraunos, king of Macedon. Another reached central Greece and attacked the sanctuary region around Delphi in 279. Resistance, winter conditions, losses and disputes fractured the expedition, while part of the movement continued across the Hellespont.

In 278, Celtic forces entered Asia Minor as allies in a dynastic struggle and remained. The region around Ancyra became Galatia. The Tolistobogii, Tectosages and Trocmi divided territory and preserved leaders, councils and military followings inside a Greek-speaking and Phrygian world. They raided, collected payments and served as soldiers, while farming and governing a region that acquired their name.

Attalus I of Pergamum defeated Galatian forces and commemorated the victory through monuments that made the defeated Gaul a recognisable Hellenistic figure. Sculptures associated with those programmes, known chiefly through later copies, could present the enemy as both barbarian and courageous. Rome defeated Galatian forces again in 189 BCE. Neither victory erased the communities. Galatian rulers later became Roman allies, and Celtic speech persisted there into late antiquity. Migration had produced another local combination, not an unchanged culture transplanted east.

Towns before Rome

The migrations hide the larger population that stayed home. Across Gaul and central Europe, most people farmed, raised animals, made cloth, repaired tools and moved goods between households and markets. During the last three centuries BCE, parts of this world became more concentrated. Coinage spread. Craft production grew more specialised. Political centres acquired ramparts, streets, workshops and populations large enough to look urban.

The largest are called oppida. At Bibracte, the Aedui enclosed more than 200 hectares behind double ramparts. Metalworkers made tools, weapons, brooches, coins and enamelled ornaments. Italian wine arrived in amphorae. Merchants, clients, farmers, artisans, nobles and religious specialists met within a political centre that existed before Roman conquest.

The Aedui did not live under a king. Aristocratic families competed through councils, elections, patronage, debt and armed followers. Other peoples organised authority differently. An oppidum still depended on farms beyond its walls for grain, cattle, timber and charcoal. Its importance lay in gathering production, exchange and political decisions, not in replacing the countryside.

Southern Gaul became a Roman province in the late second century BCE, while merchants, migrants and diplomacy extended Roman influence farther north. Italian wine crossed the frontier in immense quantities. Gallic elites acquired imported goods and Roman relationships before they acquired Roman governors. Caesar would encounter communities already connected to Mediterranean markets and to one another. Their roads, granaries, councils and rivalries gave a conqueror assets to seize and choices to manipulate.

Atlantic worlds

Britain and Ireland belonged to the wider Iron Age without forming a western copy of Gaul. Britain contained enclosed farmsteads, roundhouses, hillforts, open settlements and, in some regions, growing concentrations of power and exchange. Coins spread chiefly in the south and east. Cross-Channel trade moved wine, pottery, metals and people. Political names recorded later by Rome describe regional communities, not one British state.

At Snettisham in Norfolk, people deposited large quantities of gold and silver alloy torcs during the late Iron Age. The Great Torc alone weighs 1,084 grams. Its scale, twisted construction and decorated terminals reveal specialised craft and an extraordinary concentration of wealth. The hoards may have been offerings, political stores or both. Their meaning was local even while their art belonged to a wider La Tène conversation.

Ireland followed another path. It had no oppida on the continental scale and left almost no contemporary writing. Its metalwork still connected it to Britain and Europe. The Broighter hoard, deposited near the old shore of Lough Foyle during the first century BCE, contained a gold collar, necklaces, a bowl and a miniature boat complete with oars and fittings. The objects show Irish craftspeople selecting and remaking outside forms. Later mythology cannot securely identify the deity or ritual involved.

Bodies now reveal social arrangements that objects concealed. Genomes from Durotrigian cemeteries in southern Britain show a large kin group connected through women, while most newcomers were men. Other British Iron Age cemeteries display unusually concentrated maternal ancestry, consistent with matrilocal residence beyond this one community. The direct kinship reconstruction remains local, and residence does not establish inheritance or political authority. What it reveals is a household map unlike the patrilocal pattern documented at many earlier European sites.

The same research finds fine-grained regional clusters and links across the Channel. Britain was neither sealed off nor socially uniform. Britain and Ireland matter before Rome arrives because they supply their own households, burial practices, art and political geographies. Calling them Celtic can be useful through language and later inheritance, provided the label does not turn their differences into delay or deficiency.

Iberia takes its own route

In central Iberia, peoples whom ancient writers called Celtiberians built fortified towns, minted coins and wrote a Celtic language using an adapted Iberian script. They lived among Iberian, Lusitanian and other communities and borrowed forms from them. Their language links them to the Celtic family. Their politics and material world belong to Iberia.

Rome entered the peninsula during the war with Carthage in 218 BCE and stayed through decades of alliance, rebellion and conquest. Celtiberian communities fought Rome, fought beside it and pursued local rivalries while successive Roman commanders sought glory, tribute and secure roads. Numantia became the best-known centre of resistance. In 133, Scipio Aemilianus surrounded it with camps and a circumvallation, cut access to the countryside and starved it into surrender.

Many Celtiberian inscriptions date from the age of Roman conquest. They use an Iberian script or, later, Latin letters to record names, hospitality agreements, coin legends and public texts in a Celtic language. Political independence narrowed while the language became more visible to history. Conquest destroyed one kind of record and helped create another.

Caesar turns interventions into conquest

By 58 BCE, Rome already controlled Mediterranean Gaul. Julius Caesar arrived with legions, debt, political enemies and a command that could repair all three if it produced victories. The rest of Gaul contained alliances and threats that allowed each campaign to be presented as a response.

The Helvetii were preparing to move west. Caesar blocked them, defeated them and forced the survivors home. Aeduan leaders then appealed for help against Ariovistus, a Germanic king invited across the Rhine during an earlier rivalry. Caesar defeated him. The following year he attacked the Belgae, arguing that their mobilisation threatened Rome. Intervention solved one local problem and placed Roman force inside the next.

In 56, the Veneti on the Atlantic coast detained Roman officers and relied on tidal harbours and heavy ships difficult for Mediterranean galleys to catch. Caesar built a fleet, defeated them when calm conditions and Roman devices removed their advantages, executed their council and sold much of the population. Rome answered maritime resistance with punishment designed to be remembered.

Crossings of the Rhine and two expeditions to Britain enlarged Caesar's reputation without completing either conquest. In 54, Ambiorix of the Eburones destroyed a legion and part of another after drawing them from camp under a promise of safety. Caesar answered with campaigns intended to ruin the communities that supported him. Roman capacity did not prevent serious defeat. It allowed defeat to be followed by another army and a broader retaliation.

In 52, attacks on Roman traders at Cenabum helped ignite the largest coalition of the war. Vercingetorix, an Arvernian noble, overcame opposition at home and imposed a scorched-earth strategy on allies. It required communities to burn their own settlements and stores for a coalition whose future remained uncertain.

Avaricum was spared after its inhabitants argued that the town could hold. Caesar besieged it, captured it and allowed a massacre. At Gergovia, Vercingetorix then defeated a Roman assault, and the Aedui changed sides. Caesar recovered, forced his opponent into Alesia and built a ring of fortifications around the town. When a relief army assembled outside, he built another line facing outwards.

Alesia was not won by Gallic disunity alone. The relieving coalition represented a huge mobilisation and attacked repeatedly. Roman engineering enclosed the defenders, cavalry helped defeat the relief force, and a disciplined army held two fronts while hunger worked inside. Vercingetorix surrendered. Fighting continued into 51, followed by enslavement, executions and the destruction of strongholds.

Caesar's narrative presents conquest as a chain of necessary responses. The accumulation tells another story. He entered as an ally, punished resistance, widened the zone requiring protection and acquired an army whose loyalty would decide Roman politics. Local rivalries helped him begin. Roman institutions, personal ambition and sustained force explain why the interventions did not end.

Gaul entered Roman rule and was later organised into provinces. Its inhabitants did not cease at once to be Aedui, Arverni, Treveri or members of smaller communities. Rome removed independent war and foreign policy first. Language, cult and daily practice changed over generations.

Britain, revolt and accommodation

Caesar crossed to Britain in 55 and 54 BCE, won submissions and hostages, and left no occupation. The expeditions increased Roman knowledge of an island already tied to Gaul by trade and dynastic politics. During the next century, rulers in south-east Britain minted coins, competed for territory and used Roman links against rivals.

Claudius invaded in 43 CE with an army able to remain after the initial victories. Roman forces advanced through battle and alliance, building roads and forts as they moved. Some rulers resisted; others accepted client status and retained authority under imperial protection. Caratacus led resistance after defeat in the south. When he fled to the Brigantian queen Cartimandua, she handed him to Rome, preserving her alliance while deepening conflict inside her kingdom.

The arrangement broke among the Iceni in 60 or 61. After the death of their client king Prasutagus, Roman officials moved to absorb the kingdom, seized property and abused the ruling family. Boudica led the Iceni and allies against Camulodunum, Londinium and Verulamium. The towns were destroyed before a smaller Roman force defeated the revolt. Other rulers remained neutral or loyal, while the Roman army retained the training and supply needed to concentrate against the coalition.

Conquest continued through Wales and northern Britain. Agricola's campaigns reached into what is now Scotland, but Rome never converted the advance into permanent control of the Highlands. Hadrian's Wall, begun in 122, marked a managed military zone with gates, forts and movement across it rather than a border between two peoples. Ireland never became a province.

Roman rule changed regions unevenly. Towns, villas, forts and roads spread most densely in some areas, while others remained more rural. Latin served government and status without erasing Brittonic speech. Local gods survived under translated names. Rome held much of Britain for nearly four centuries, but it never turned the island into one uniform society.

Different endings

Gaulish did not vanish when Caesar finished his account. It continued beside Latin for centuries, leaving words and place-names after everyday speech shifted. Latin offered access to law, office, trade and an empire-wide culture, so language change could advance through advantage as well as coercion. By late antiquity, most descendants of conquered communities were Roman provincials, Christians, soldiers and landowners who could hold several identities at once.

Imperial government in Britain broke down during the early fifth century. Brittonic-speaking kingdoms emerged in the west and north, while Germanic-speaking settlers gained territory in the east and south. Language boundaries moved through war, settlement, intermarriage and political prestige rather than one replacement event. Brittonic survived in Welsh and Cornish and travelled with migrants to Armorica, helping create Breton.

Ireland produced the clearest late ancient writing in a surviving Celtic language. From roughly the fourth to sixth centuries CE, names and descent were cut along standing stones in ogham. The inscriptions are brief, but they give local people a written presence unavailable for Iron Age Britain. Far away in Anatolia, Galatian speech was fading under Greek.

There was no single Celtic fall. Northern Italy, Celtiberia, Gaul and Britain entered Roman power at different times. Ireland did not. Some languages disappeared; others moved and divided. Political systems ended before every inherited practice did. The ancient name faded from ordinary use long before the inheritance it now describes was assembled again.

How we know

Before Roman conquest, most of the story comes from mines, graves, farmsteads, settlements, coins, workshops, hoards and deposits in water or earth. Scientific analysis can identify diet, movement, kinship and ancestry. The 2025 genomic evidence identifies close kinship and male-biased mobility directly at Durotrigian burial sites; the argument for a wider British pattern comes from comparative mitochondrial evidence. Both describe sampled cemeteries, not every household or the distribution of power.

Radiocarbon dating fixes objects and burials within ranges rather than supplying a perfect calendar. Isotopes can show that a person grew up elsewhere without naming the language spoken there. Ancient DNA can identify relationships and population links, but identity still has to be argued from context. Each method answers a narrower question than the word Celt seems to ask.

Inscriptions reveal Gaulish, Lepontic, Celtiberian and early Irish, while names extend the linguistic map. Greek and Roman writers supply events and labels from outside. Polybius wrote near the wars in Italy. Caesar was commander and political beneficiary of the conquest he narrated. Tacitus shaped British history around arguments about empire and freedom.

Archaeology tests parts of those narratives. Bibracte confirms large pre-Roman centres. Alesia preserves siege works. Burnt layers fit the Boudican destruction. An object still cannot state its owner's language. The broad sequence is secure. Early language spread, exact migration routes, self-identification, druidic organisation and the meaning of many ritual deposits remain less certain.

What People Get Wrong

“The Celts were one people”

There were ancient people called Celts, but the singular picture is a modern enlargement. Greek and Roman authors applied overlapping Celtic and Gallic labels to populations they considered connected. Some inhabitants of Gaul called themselves Celts in a broad sense. Political life still belonged to communities such as the Aedui, Arverni, Helvetii, Iceni and Brigantes.

Those communities had separate rulers, councils, alliances, territories and enemies. They did not answer to a Celtic capital, share one army or treat every related speaker as a member of one nation. Language supplies the strongest common thread, but a language family does not create a state. Nor does political separation prevent wider affinities from being recognised when trade, migration or conflict brings people together.

The national picture became persuasive because maps prefer clean colours, Roman authors grouped outsiders broadly, and later movements wanted an ancient population equal in scale to Rome. The opposite claim, that the Celts never existed, became attractive once those maps were criticised.

Both flatten the evidence. Ancient names, related languages and connected artistic worlds were real. Their borders did not coincide. Rome conquered local polities whose interests often conflicted, not a united civilisation that had forgotten to unite. Their political plurality was normal for the period, not evidence of a country they had failed to invent.

“Hallstatt and La Tène are maps of Celtic ethnicity”

Hallstatt and La Tène are names taken from archaeological sites, then applied to recurring patterns in burials, settlements, weapons, ornaments and art. They organise time and material evidence. They do not mark the borders of two successive Celtic countries.

Objects travel without carrying identity intact. A bronze vessel can be traded, a brooch copied and a style adopted by people who speak different languages. Iron Age elites borrowed from Greek, Etruscan, steppe and neighbouring traditions, then changed what they borrowed. The Snettisham and Broighter goldwork belongs to a wider La Tène history while remaining distinctively British and Irish.

The reverse problem also applies. Celtic speech existed where the expected material package is weak or absent. Inscriptions in northern Italy and Iberia preserve Celtic languages, while Britain and Ireland adopted La Tène forms selectively. A distribution map of swords or torcs cannot settle every speaker's language.

The old model joined archaeology to migration in one tidy picture: a people, a homeland and an outward spread. Separating the evidence produces a better account. Hallstatt and La Tène reveal exchange, production and prestige. Language must be traced through inscriptions and recurring structure. Ethnicity requires another argument. One artefact cannot prove all three.

“They were primitive tribes without towns or writing”

The word tribe often arrives carrying its conclusion. It suggests small wandering groups outside urban organisation until Rome supplied roads, markets, literacy and administration. Iron Age Europe contained far more variation.

Most people farmed from settled households. They grew cereals, raised livestock, worked wood and iron, made textiles and exchanged goods. Salt mining at Hallstatt required organised labour and technical knowledge. Heuneburg drew a dense population and Mediterranean goods. Later oppida such as Bibracte contained ramparts, roads, workshops, storage, coin production and political gathering places.

Writing existed too. Lepontic, Gaulish and Celtiberian appear in inscriptions using borrowed scripts. Coins carried names and titles; dedications and public texts survive. Ogham later supplied the earliest substantial written evidence for Irish. What is missing is a continuous native history comparable to Caesar or Livy.

The myth persists because Roman writers contrasted civilisation with barbarism, while timber buildings, cloth and routine records disappeared more readily than stone cities and manuscripts. The correction does not make every hillfort urban or every community literate. It removes a false sequence in which complexity arrives with conquest. Rome captured and redirected existing societies rather than creating them from scratch.

“Druids built Stonehenge and led a Celtic church”

Stonehenge was ancient before the first historical references to druids. Its principal construction belongs to the Neolithic and early Bronze Age, while written accounts of druids begin in the last centuries BCE. The gap is roughly fifteen hundred years. Antiquaries made the association because a learned pagan priesthood seemed the available explanation for Britain's most famous monument.

Druids themselves are historical. Greek and Roman writers describe learned figures in Gaul and Britain who dealt with ritual, teaching, memory, law and dispute settlement. Caesar's omission of druids from his campaign narrative, despite discussing them in his ethnography, may have helped frame resistance as political rebellion rather than sacred war. That is a serious modern interpretation, not a settled fact.

What cannot be recovered is a pan-Celtic church. There is no evidence for one doctrine, central hierarchy, uniform costume or authority extending from Ireland to Galatia. Archaeology identifies sanctuaries, offerings and sacred places more confidently than it identifies the profession of the person conducting a rite.

The familiar druid combines classical description, medieval story, antiquarian speculation and modern ceremony. Remove Stonehenge and the universal priesthood, and the historical role becomes narrower, less visible and more credible.

“Celtic warriors all fought naked and painted blue”

Two ancient passages produced an entire costume. Polybius describes the Gaesatae fighting naked at Telamon in 225 BCE, while the Insubres and Boii beside them wore trousers and cloaks. Nudity distinguished one contingent within the same army. It was a display of fearlessness, ritual confidence or status, not standard equipment for every Celtic-speaking warrior.

Caesar claims that Britons stained themselves with vitrum and appeared blue in battle. Translators often rendered vitrum as woad. The identification is disputed, the archaeological evidence for body paint is inconclusive, and Caesar did not describe Gauls, Celtiberians or Galatians doing it. Nor did he provide the full-body designs supplied by later artists.

Roman art benefited from a visually recognisable barbarian. Early modern antiquaries, painters, nationalist imagery and film kept the costume because blue skin and nakedness identify a Celt in one glance. The image then looped back into popular history as though every part came from archaeology.

Some warriors used nudity, body colouring or conspicuous appearance. The correction concerns scale and certainty. One named war band and one ambiguous British term do not amount to a continental uniform.

“Rome could never hold the Celts”

Roman rule lasted about five centuries in Gaul and nearly four across much of Britain. It absorbed Celtic-speaking communities in northern Italy, Iberia and Galatia as well. The picture of an unconquerable people beyond every frontier confuses incomplete conquest with failed rule.

The limits were real. Roman armies never established permanent rule in Ireland. Rome never secured permanent control of the Scottish Highlands. Vercingetorix assembled a major coalition, Boudica destroyed several Roman towns, and resistance in Celtiberia continued for generations. None prevented long government in the conquered regions.

Rome did not win because local communities were incapable of unity. It brought manpower, engineering, supply, professional command, political endurance and severe coercion, then worked through alliances and rival elites. Political plurality supplied openings. Roman capacity converted openings into roads, forts, sieges, taxes and provinces.

Conquest changed authority first. Independent warfare and foreign policy disappeared, while military demands and provincial law followed. Language and identity changed more slowly. Gaulish survived beside Latin; Brittonic continued through and after Roman rule. Political control can be durable without producing complete cultural replacement. That is a more accurate measure of what Rome held.

“Modern Celtic identity is either ancient and unchanged or modern fiction”

The argument usually offers two choices. Modern Celtic identities are direct survivals from the Iron Age, or eighteenth and nineteenth-century enthusiasts invented the whole thing. Both demand a purity identities rarely possess.

The inheritance is real. Irish, Scottish Gaelic, Manx, Welsh, Cornish and Breton belong to the Celtic language family. Medieval writers preserved stories containing older names, places and themes. Communities maintained languages and local traditions through conquest, migration and political change.

The reconstruction is real as well. Ancient people in Britain and Ireland did not commonly use Celt as their collective name. Medieval Christian art is not identical to La Tène art. Scholars identified the language family, while antiquarians, artists and national movements combined language, archaeology and literature into a wider past. Some selections were tendentious, and Macpherson's Ossian involved extensive fabrication.

Construction does not make an identity fraudulent. A modern community can build itself from inheritances that were never previously organised under one label. The useful questions concern which elements descend, which were recovered, which were adapted and which were invented. Modern Celtic identity is neither a fossil nor a forgery. It is an inheritance made historical by continued use.

Use It

This subject offers a discipline before a conclusion: broad identities may be real, but the evidence supporting them has to be named. The method is precision without paralysis.

State what kind of connection you mean

Language, ancestry, artefact style, political allegiance, self-description and later memory are different kinds of connection. They may reinforce one another or produce contradictory maps. Celtic languages link populations across ancient Europe. La Tène art links objects, makers and patrons. Caesar's labels reveal Roman classification. None silently answers the questions belonging to the others.

Use the distinction whenever a group is described as one people. What is the claim based on? Are speakers being turned into a race, objects into a nation, or an outsider's name into self-identification? The correction need not destroy the category. It may narrow it into something more accurate and more useful.

Breton gives the method in miniature. It is a Celtic language spoken in France, but it descends chiefly from Brittonic carried across the Channel rather than from Gaulish standing still. Language, place and continuity tell different parts of the history. The broad label works once the mechanism has been specified.

Then ask what power does with the label. Caesar's divisions of Gaul were descriptions written by a conqueror whose categories helped explain his wars to Rome. Once law, taxation or territory follows a classification, the name stops being passive and becomes part of the history it claims to record.

Read what preservation has selected

Museums make the Iron Age look crowded with swords, torcs, shields and princely graves. Those objects survived because they were durable, valuable and deliberately deposited. Grain was eaten. Timber rotted. Cloth disappeared. Daily labour leaves a quieter record than ceremonial death.

Whenever evidence is visually impressive, ask whether it was common or preservable. The Vix burial can prove that one woman received extraordinary treatment. It cannot describe every woman's life. The Snettisham Great Torc reveals craft and concentrated wealth. It does not prove that most people wore gold or that its deposit had one agreed meaning.

The same bias appears outside archaeology. Archives favour governments, crises and people able to leave records. A vivid source is not a representative sample. Look for the systems that made it possible: farms behind feasts, workshops behind treasures, labour behind a warrior's equipment.

The missing evidence matters too. We know Caesar's reasoning because he wrote it. We do not possess Vercingetorix's account. The silence can tempt historians to fill the space with the conqueror's categories. Preservation is never neutral merely because no editor is visible.

Break movement into mechanisms

Old accounts imagined a Celtic people leaving a central homeland and spreading a complete cultural package across Europe. The reaction sometimes denied substantial movement at all. Both compress several processes into one word.

People move as individuals, spouses, households, armies, settlers, captives and traders. Languages move through prestige, intermarriage and institutions. Objects travel as gifts, loot, commodities and models for imitation. Genes record biological ancestry, not the language spoken at dinner. Each mechanism leaves a different trace and operates at a different speed.

British burial genetics supplies one mechanism: male-biased movement into communities organised around local maternal lines. Galatia supplies another: armed groups crossed into Anatolia, settled and formed a regional society. Breton supplies a third, with language carried from Britain into a continental region where related Gaulish had already receded. The traces differ because the movements differed.

A single arrow on a map often compresses centuries and mechanisms into one wave. Before accepting it, ask what moved, in what numbers, over what time and with what effect on the people already there.

Put outside capacity beside local choice

The conquest of Gaul is often explained from one side. Rome won because its opponents were divided, or Rome won because its legions were superior. Both remove the interaction that made the outcome possible.

Local communities had reasons to ally, wait, resist or use Rome against a rival. Those choices mattered. Rome also possessed experienced forces, engineers, supply, replacement capacity and a commander rewarded by continued conquest. Local politics opened doors that imperial capacity could force wider. After a Roman defeat, another force could be raised. After a Gallic community lost its harvest, hostages or leading families, its range of choices narrowed.

Carry both levels into any account of collective failure. Reconstruct what each participant could see and what the outside power could do. A shared interest may lose to immediate incentives without anybody being foolish. An external actor may turn small differences into control because it can concentrate resources none of the local participants can match.

Blaming disunity alone mistakes vulnerability for sufficient cause. Praising superior organisation alone erases the alliances, betrayals and political calculations through which power entered. The outcome belongs to the relationship between local decisions and unequal capacity.

Audit inheritance without a purity test

Modern Celtic identity combines language descent, medieval literature, local memory, scholarship, political revival and invention. Its reality does not depend on every element being the same age.

When an identity claims the past, separate transmission from recovery and recovery from creation. A revived language or ceremony can be new in form and rooted in inherited material. A nineteenth-century invention should not be passed off as Iron Age fact. Later invention should not be used to dismiss the attachments, losses and communities organised around it.

Macpherson's Ossian makes the distinction sharp. He drew on Gaelic tradition while presenting extensive compositions of his own as translations from an ancient bard. The claim was false, the cultural effect was real, and the poems helped shape how Europe imagined the Celtic past. Historical influence does not authenticate a source. Fraud does not make the surrounding tradition imaginary.

The test is accuracy, not purity. Living identities are constructed because every generation selects what to carry. The responsible question is whether the construction tells the truth about its materials and leaves room for people whose version of the inheritance differs.

The limits

Category scepticism can become another simplification. Once language, objects and political identity have been separated, it is tempting to conclude that no broad group can ever be named. That would ignore the real linguistic and historical connections preserved by Celtic.

The layers should be distinguished, then related. Ancient people could recognise wider affinities while caring more about local allegiance. Modern people can share a Celtic identity without claiming that an Iron Age federation existed. No category is made false because its edges are untidy.

The evidence remains uneven. Most continuous writing comes from Greek and Roman outsiders. Archaeology cannot tell us every language or self-name. Genomics can recover kinship, ancestry and movement without identifying political identity, inheritance rules or the distribution of power. The matrilocal pattern in sampled British cemeteries should not be stretched into a social constitution for Europe.

The method produces better questions, not one definition valid from Hallstatt to modern Wales. It is least useful when caution becomes an excuse never to state what the stronger evidence does show.

The one thing to keep

Keep one follow-up question: Celtic in what sense?

It forces the claim to show its evidence. The answer may be linguistic, archaeological, political, genealogical, religious or modern. Several answers may apply, and they need not coincide. That does not empty the category. It prevents the category from doing work the evidence cannot support.

Return to the Battersea Shield. It is reasonable to call it Celtic because it belongs to a recognised Iron Age artistic world and to the later history built around that word. The label becomes misleading only when it is made to prove a single nation, language or self-name without further evidence. The object does not become less interesting when the claim becomes more precise. It becomes usable.

The question also changes how any one-colour map of a people should be read. The colour is not a finding. It is a proposal about which connection matters most. Ask who supplied the name, what evidence fixes the boundary, which local histories have been hidden and what changes if another layer is mapped instead.

The Celts are clearest when the broad inheritance and the local worlds remain visible together. Their connections never made them uniform, and conquest did not give them one ending. What survived came from different ages. Identity need not be pure to be real. It does need to say what kind of reality it claims.

Terms

Celt or Keltoi. Keltoi was a Greek name for certain western peoples and the ancestor of Celt. Its ancient range shifted, so it cannot be stretched into a precise map of one nation.

Gaul or Galli. The Roman label for peoples in Gaul and northern Italy. Gaul overlaps with Celt but is not an exact synonym, and most inhabitants identified more immediately with local communities.

Galatian. A member of the Celtic-speaking groups that settled in central Anatolia during the third century BCE. Galatian identity developed inside a Greek-speaking and later Roman provincial world.

Celtic languages. A branch of Indo-European once spoken across much of Europe and Anatolia. Linguistic relationship provides the strongest defensible connection among the societies called Celtic.

Continental Celtic. The extinct Celtic languages attested on mainland Europe, including Gaulish, Celtiberian and Lepontic. Galatian is known mainly through names and short references rather than surviving texts.

Insular Celtic. A common scholarly grouping for Celtic languages of Britain, Ireland and their descendants. It is a linguistic category and should not be mistaken for proof of one ancient island ethnicity.

Goidelic. The branch containing Irish, Scottish Gaelic and Manx. Older books call it Q-Celtic because an ancestral consonant remained k-like where related Brittonic languages often developed a p sound.

Brittonic. The branch containing Welsh, Cornish and Breton. Breton descends chiefly from speech carried from Britain to Armorica, not from Gaulish surviving unchanged in France.

Gaulish. The Celtic language once spoken across much of Gaul, known from inscriptions, names and words preserved in Latin. It retreated gradually rather than disappearing when Caesar declared victory.

Celtiberian. A Celtic language of central Iberia, written first in an adapted Iberian script and later in Latin letters. The name does not mean a simple mixture of Celtic and Iberian languages.

Lepontic. One of the earliest attested Celtic languages, written in northern Italy from around the sixth century BCE. Its inscriptions help separate linguistic evidence from Hallstatt and La Tène artefact maps.

Ogham. An alphabet of strokes cut mainly along stone edges, used for the earliest substantial written evidence of Irish. Most surviving inscriptions name people and date from late antiquity.

Hallstatt. An archaeological label for much of central Europe's earlier Iron Age, named after the Austrian salt-mining site. It describes material patterns and chronology, not a Hallstatt kingdom or nation.

La Tène. The later Iron Age material tradition named after a Swiss lakeside site, conventionally beginning around 450 BCE. Its curving art spread farther than any single language or political identity.

Oppidum. Latin for a fortified town, used by archaeologists for large late Iron Age settlements such as Bibracte. The plural is oppida, and not every hillfort qualifies.

Hillfort. A settlement or refuge protected by earthworks, walls or ditches, often on high ground. Hillforts varied enormously and are neither automatically urban nor uniquely Celtic.

Torc. A rigid neck ring made from metal, often associated with high status, ritual deposition or hoards. Its visual power made it a modern Celtic emblem beyond its ancient distribution.

Fibula. A brooch used to fasten clothing. Changes in fibula shape help archaeologists date graves and trace contact, but a brooch style cannot identify its wearer's language by itself.

Matrilocality. A residence pattern in which partners tend to live with or near the woman's family. Genomic evidence directly supports it at sampled Durotrigian cemeteries and is consistent with a wider British Iron Age pattern. It does not establish matrilineal inheritance, matriarchy or one Celtic rule.

Druid. A learned religious, legal or teaching specialist described in parts of Gaul and Britain. The evidence does not support one Europe-wide priesthood, doctrine or central authority.

Nemeton. A Celtic word preserved in inscriptions and place-names for a sacred grove or sanctuary. It helps identify religious places without proving that every grove was supervised by druids.

Votive deposit. An object deliberately removed from use and offered in water, earth or a sanctuary. Bent weapons and valuable metalwork at sites such as Llyn Cerrig Bach fit this category.

Clientage. An unequal relationship linking an aristocratic patron to dependants, followers and armed retainers. Client networks converted wealth and reputation into political support and military force.

Polity. An organised political community capable of making collective decisions. It is often safer than tribe when describing Iron Age peoples whose institutions were local but far from primitive.

Tribe. A common label for named peoples such as the Iceni or Aedui. It can be useful, but it often smuggles in assumptions of simplicity, kinship and weak government.

Vitrum and woad. Caesar claimed that Britons used vitrum, giving them a bluish appearance. The common translation as woad is disputed, and the passage does not describe a uniform practice across Celtic-speaking Europe.

Vercingetorix and Alesia. Vercingetorix led the largest Gallic coalition against Caesar in 52 BCE. His defeat at the siege of Alesia ended the last campaign capable of reversing the conquest.

Boudica. The Iceni ruler who led the revolt of 60 or 61 CE in Roman Britain. She commanded a regional coalition, not a united army of all British Celts.

Romanisation. Shorthand for cultural and political change under Roman rule. Used carefully, it describes uneven adoption, coercion and adaptation rather than one-way replacement by a uniform Roman culture.

Celtic Revival. The modern recovery and reconstruction of Celtic languages, literature, art and national identities. It includes serious scholarship, political renewal and creative inventions such as James Macpherson's influential Ossian poems.

Go Deeper

One hour can separate the language family, the archaeological record and the legend. The next step is one broad synthesis, one conqueror's account, one sustained challenge to the category and one continent-wide archaeological reference. Read them together because no single map can carry the whole subject.

The overview

Barry Cunliffe, The Ancient Celts, second edition (2018). This is the most inviting full-length continuation: illustrated, wide-ranging and written with enough narrative movement to carry the reader from the Iron Age into later inheritances. Cunliffe gives trade, migration, art, religion and settlement as much attention as war. The warning concerns scale. He draws a broader Celtic world and a longer account of origins than some linguists and archaeologists accept, and parts of that model remain debated. Read it for range and physical detail, while keeping language, artefacts and political identity separate.

The primary account

Julius Caesar, The Gallic War, translated by Carolyn Hammond (2008). Begin with Books 1, 6 and 7: the interventions that became conquest, the ethnographic account of Gaul, and the coalition led by Vercingetorix. Hammond supplies maps, dates, notes and a glossary, which remove much of the friction. Caesar is lucid because he wants Roman readers to understand his decisions. He is also the commander, political beneficiary and chief surviving narrator of the war. Read him for events, local rivalries and Roman self-presentation, never as the voice of the conquered.

The argument about the name

John Collis, The Celts: Origins, Myths and Inventions (2003). Collis follows the word from classical ethnography through archaeology, antiquarian scholarship and modern nationalism, showing how language, La Tène art, Britain, Ireland and later revival were bundled into one apparent people. The book is argumentative, strongest on Britain and the history of scholarship, and at times presses scepticism harder than later work would. Its lasting value is disciplinary. Every use of Celtic has to state what evidence it means and what the label is being asked to prove.

The evidence

Colin Haselgrove, Katharina Rebay-Salisbury and Peter S. Wells, editors, The Oxford Handbook of the European Iron Age (2023). This is the reference work: fourteen regional chapters and twenty-six thematic studies covering settlements, agriculture, craft, migration, burial, gender, ritual, literacy and art across the continent. It replaces the old central-European-to-Atlantic storyline with many regional histories connected at different strengths. The book is large, expensive and designed for selective use. Start with the chapters on migration, urbanisation, identity and whichever region interests you, then follow their bibliographies into specialist work.

Notes and Sources

The peoples grouped under the word Celtic left short inscriptions, names, coins, settlements, graves, offerings and related languages, but no continuous native history of the Iron Age. Greek and Roman writers provide events and labels from outside. These notes therefore keep four bodies of evidence distinct: language, archaeology, ancient classification and later identity. Dates are BCE unless marked CE.

The Whole Thing in One Page

The Battersea Shield. The British Museum dates the bronze facing broadly to 350-50. It once covered a wooden shield whose organic core has disappeared. The object supports discussion of La Tène-style metalwork and possible votive river deposition, but no inscription identifies its makers or owners as Celts. Calling it Celtic is a modern archaeological classification, not a surviving self-description.

Ancient Celts without a Celtic nation. Caesar, Gallic War 1.1, reports that one part of Gaul was inhabited by people who called themselves Celtae and whom Romans called Galli. Greek authors had used Keltoi and Galatai for changing sets of western and northern peoples. D. W. Harding and Patrick Sims-Williams support retaining a defensible ancient and linguistic category while rejecting a single political nation extending across Europe.

Language as the strongest thread. Anders Richardt Jørgensen's survey defines Celtic as a branch of Indo-European and includes Goidelic, Brittonic, Gaulish, Celtiberian, Lepontic and Galatian. Patrick Sims-Williams reviews competing models for the early spread of Celtic speech and concludes that archaeology, genetics and language still do not yield one agreed homeland or route. The evidence establishes related speech more securely than ancestry, artefact style or political loyalty.

Atlantic objects and communities. The British Museum records the Snettisham Great Torc at 1,084 grams. The National Museum of Ireland dates the Broighter Collar and related hoard to the first century BCE and locates the find near the ancient shore of Lough Foyle. Both objects belong to wider La Tène histories while retaining regional construction, imagery and depositional contexts.

Hallstatt, La Tène and ordinary life. The Oxford Handbook of the European Iron Age supplies the broad archaeological framework. Hallstatt and La Tène are chronological and material classifications covering varied regions. They do not form the borders of two ethnic states. Farming, animal husbandry, textiles, mining, ironworking, exchange and settlement carried most daily life beneath the weapons and jewellery that survive most conspicuously.

Druids and sacred practice. Caesar and Tacitus describe druids in Gaul and Britain. Miranda Aldhouse-Green uses archaeology to test rather than illustrate those texts. Sacred waters, groves, sanctuaries, offerings and human remains show varied religious practice. They do not establish one Europe-wide church, one doctrine or a druid behind every deposit.

Mobility and conquest. Polybius, Caesar, Tacitus and the migration chapter by Andrew Fitzpatrick support movements into Italy, the Balkans and Anatolia, followed much later by Roman conquest in Gaul and Britain. Mobility ranged from individuals and marriages to organised armed migration. The book rejects one population wave without denying the larger movements jointly supported by text and archaeology.

The reconstructed inheritance. Nancy Edwards, James Harris, John Collis and Thomas Curley trace the roles of Edward Lhuyd, antiquarian scholarship, language comparison and the Ossian controversy in forming modern Celtic categories. Medieval languages and literature supplied real inheritances. The later label gathered them into a shared identity that had not existed in the same form in the Iron Age.

Why You Should Care

The Vix burial. The official museum dates the grave to about 470-460. A high-ranking woman was buried with a wagon, gold jewellery and a Greek bronze mixing vessel 1.64 metres high with a capacity of about 1,100 litres. The burial shows long-distance exchange and exceptional female status. It does not prove the office, title or general freedom of Iron Age women.

The older linguistic map. Celtic languages were once spoken in Gaul, Iberia, northern Italy, the Balkans and central Anatolia as well as Britain and Ireland. Jørgensen provides the linguistic framework; inscriptions and names establish the continental languages unevenly. The modern concentration on Europe's Atlantic edge is the result of language loss, migration and survival rather than the original geographical centre of the family.

Three maps that do not coincide. Harding, Sims-Williams and the European Iron Age handbook support the distinction among the map of Celtic languages, the map of Hallstatt and La Tène material, and the map drawn by Greek and Roman ethnonyms. The overlaps are historically meaningful. Treating them as one boundary creates claims the evidence cannot carry.

The survival bias of weapons and graves. Elite metalwork and burials dominate museums because durable and deliberately deposited objects survive. Settlement archaeology, environmental evidence and production debris restore households, farms and workshops. The manuscript uses the bias as a correction to the warrior stereotype, not as a claim that warfare lacked social importance. Snettisham and Broighter are used as evidence for craft and regional elite display rather than as representative possessions.

The modern name. Modern Celtic identity is treated as a later historical formation with genuine linguistic and cultural foundations. Collis is valuable for exposing invented continuity; Edwards, Harris and Laing show the scholarship and archaeology behind later classifications; Curley documents how Macpherson's Ossian poems shaped the revival despite the fraud at their centre.

The Core Ideas

Celtae, Galli and local names. Caesar's opening distinction supplies rare evidence for a broader self-name in part of Gaul. It cannot be projected automatically onto every population later called Celtic. Named communities such as the Aedui, Arverni, Sequani, Helvetii, Iceni and Brigantes were the principal political actors in the surviving narratives.

Polity rather than failed nation. Polity is used for an organised political community able to make collective decisions. Tribe is retained where it matches common usage but does not imply political simplicity. The absence of a Celtic state is not treated as an institutional deficiency. It describes a world in which authority remained local.

The Celtic language branch. Jørgensen's chapter provides the main classification. Continental Celtic includes the extinct ancient languages of mainland Europe; Insular Celtic groups the languages of Britain, Ireland and their descendants. The precise family tree within Celtic remains debated, especially the relationship of Brittonic to Gaulish and the force of an Insular Celtic grouping.

Q-Celtic and P-Celtic. The familiar contrast describes one recurring sound development, illustrated by Irish cúig and Welsh pump. It is useful as a memory aid for Goidelic and Brittonic, but it is not a complete account of the family and should not be turned into two ancient races.

Lepontic, Gaulish, Celtiberian and Galatian. These languages survive through unequal records. Lepontic and Gaulish were written in scripts adapted from neighbours; Celtiberian used an Iberian script and later Latin letters; Galatian is known mainly from names and external references. The linguistic classification rests on recurring structure, not on one alphabet or one artefact package.

Breton and Gaulish. Breton descends chiefly from Brittonic speech carried from Britain to Armorica during late antiquity and the early Middle Ages. Alderik Blom's study of late Gaulish allows the possibility of surviving Gaulish pockets, but Breton is not treated as Gaulish continuing unchanged in France.

Hallstatt chronology and salt. Hallstatt conventionally covers much of the central European earlier Iron Age, especially the eighth to fifth centuries. The Austrian site became wealthy through salt mining. The mines and cemetery preserve labour, textiles, tools and unequal burial wealth, while the name now covers regions that never answered to Hallstatt politically.

Heuneburg and Vix. The Heuneburg's sixth-century mud-brick wall reflects Mediterranean building knowledge adapted on the upper Danube. Vix shows Greek craft, wine culture and local elite display meeting in Burgundy. Both sites demonstrate connection without cultural surrender to the Mediterranean.

La Tène art and the type site. La Tène begins conventionally around 450 and takes its name from the Swiss lakeside site. Curving plant, animal and human forms spread through objects and local variants. The style travelled beyond any securely mapped language boundary, which is why a La Tène object cannot identify its owner's speech by itself.

Oppida and Bibracte. Caesar's Latin word oppidum is used by archaeologists for large late Iron Age settlements, though definitions vary. Bibracte enclosed more than 200 hectares and contained ramparts, workshops, coinage, imported amphorae and political functions. It demonstrates substantial pre-Roman urban development without making every hillfort a town.

Farming, craft and hierarchy. The handbook and Bibracte evidence support an agrarian base with specialist metalworking, textile production, mining, trade and increasing concentration at some centres. Elite graves and feasting equipment reveal hierarchy more clearly than the lives of ordinary households. Captivity and slavery are retained as parts of warfare and labour rather than treated as uniquely Roman arrivals.

Kinship, residence and succession. Cassidy and colleagues analysed 57 genomes from Durotrigian burial sites and found an extended group centred on one maternal lineage, with unrelated incomers predominantly male. Comparative mitochondrial evidence is consistent with a wider British matrilocal pattern. Gretzinger and colleagues analysed elite Hallstatt burials in southern Germany and found a probable maternal-uncle and nephew relationship between two exceptionally rich graves, which may indicate matrilineal dynastic succession in that narrow setting. The studies concern different regions, periods, social levels and institutions. Neither directly explains female political authority or establishes one Celtic kinship system.

Offerings and local gods. Llyn Cerrig Bach and comparable deposits show valuable objects deliberately removed from ordinary use and placed in sacred settings. Roman-period inscriptions preserve many local divine names, often paired with Roman gods. Pairing is treated as translation and adaptation, not proof that all communities shared one pantheon.

Druids. Caesar's Gallic War 6 and Tacitus' accounts of Mona are central texts. Tyler Creer argues that Caesar deliberately omitted druids from most of his campaign narrative to reduce the apparent organisation and ideological force of resistance. This is one interpretation of Caesar's literary strategy, not direct evidence that druids organised the revolt or were systematically suppressed. The manuscript retains druids as learned religious, legal and teaching specialists in parts of Gaul and Britain while refusing a central hierarchy or uniform doctrine.

Stonehenge, ritual violence and sacrifice. Stonehenge's main construction predates the earliest historical druids by roughly fifteen centuries. Ian Armit shows how classical descriptions distorted older interpretations of bog bodies, decapitation and a supposed Europe-wide cult of the head. Russell and colleagues published a later Iron Age burial from Winterborne Kingston in which a young woman was killed by a cut to the neck, possibly while bound, and placed over a structured animal-bone deposit. The case makes ceremonial killing a strong interpretation in that context. It does not identify a druid or establish one rite for every unusual death.

Rome, local politics and imperial capacity. Caesar's campaigns show Roman expansion working through alliances, rivalries and defections. The final model also requires Roman replacement capacity, engineering, supply, experienced command, terror and Caesar's political incentives. Vercingetorix assembled a broad coalition in 52, and its defeat at Alesia cannot be reduced to disunity. Roman fortification, cavalry, discipline and logistics converted political openings into conquest.

Lhuyd, Ossian and the revival. Edward Lhuyd's comparative work helped establish relationships among Welsh, Cornish, Breton, Irish and Scottish Gaelic. James Macpherson then presented extensive modern composition as translation from the ancient bard Ossian. Scholarship and invention both shaped the modern category, which later supported distinct political and cultural revivals.

Narrative and chronology

Hallstatt and Vix. Natural History Museum Vienna material supports the emphasis on mining, preservation and the cemetery at Hallstatt. The Vix museum supplies the burial date and measurements. These sites are used as entries into regional systems of labour, exchange and hierarchy, not as capitals of an early Celtic state.

Movement into Italy and Rome. Polybius Book 2 is the main surviving narrative for Celtic communities in northern Italy and Rome's later wars against them. The sack of Rome is dated approximately to 390 because ancient chronological traditions differ. Migration, settlement, raiding and military service are kept distinct.

The Balkans and Galatia. Classical accounts and Andrew Fitzpatrick's archaeological synthesis support large movements into the Balkans during the early third century. Stephen Mitchell supplies the political history of Galatia. The expedition reached the sanctuary region around Delphi in 279; the book avoids the later claim that the sanctuary was successfully looted. Groups entered Asia Minor in 278 and formed communities that persisted under Hellenistic and Roman rule.

Bibracte and late Iron Age concentration. Bibracte's official archaeological summaries support the site's scale, double ramparts, craft districts, coinage and imports. Oppida appeared in some regions during the later Iron Age. Their development was uneven and did not empty the countryside on which they depended.

Atlantic Britain and Ireland. British Museum and National Museum of Ireland records support the Snettisham and Broighter descriptions. Cassidy and colleagues supply direct evidence for dominant maternal lines and male-biased mobility at Durotrigian cemeteries, plus comparative mitochondrial evidence consistent with a wider British matrilocal pattern. Their study also finds fine-grained regional structure and cross-Channel connections. The manuscript does not infer language, political identity, inheritance rules or matriarchy from the genomes.

Celtiberia and Numantia. W. V. Harris supplies the Roman expansion sequence and the fall of Numantia in 133. The Novallas bronze, published by Francisco Beltrán Lloris and colleagues, shows Celtiberian written in the Latin alphabet during the late first century. Roman conquest could narrow a language's future while generating inscriptions that make it more visible to modern readers.

Caesar's war. The Gallic War remains the main continuous source for 58-51. The manuscript follows its sequence while rejecting Caesar's presentation of every campaign as a necessary response to somebody else's threat. His debt, political position and the value of prolonged command explain why opportunity repeatedly became necessity. Force totals and casualty claims are omitted because they are partisan and difficult to test.

Vercingetorix and Alesia. Caesar Book 7 supplies the campaigns at Avaricum, Gergovia and Alesia. The archaeological site preserves Roman siege works consistent with a double line around the town and against the relief army. The famous image of Vercingetorix casting weapons at Caesar's feet is a later simplification of ancient accounts that do not agree on the scene.

Britain. Caesar's expeditions of 55 and 54 achieved submissions and hostages but no permanent occupation. Tacitus, supplemented by John Wacher, supports the Claudian conquest, Cartimandua and Caratacus, the Boudican revolt, and later campaigning. Stewart demonstrates how Roman writing constructed Britain through stock images as well as information. Britain is described through regional Brittonic-speaking polities. Rome governed much of the island for almost four centuries but never conquered Ireland or permanently held the Highlands.

Language retreat and survival. Blom supports a gradual and regionally uneven shift from Gaulish to Latin, with spoken survival outlasting public writing. Ogham inscriptions, mostly fourth to sixth century CE, provide the earliest substantial written record of Irish. Caroline Brett and colleagues support Brittonic migration into Armorica and the formation of Brittany. These are separate histories, not one unbroken Celtic polity.

How we know. Archaeology supplies mines, graves, settlements, siege works and ritual deposits. Linguistics supplies a relationship among poorly and unevenly recorded languages. Greek and Roman texts supply names and events through the interests of outsiders. No one evidence type can silently fill the gaps in another. The main uncertainties concern migration routes, population scale, self-identification, druidic organisation and the relation between artefact style and speech.

What People Get Wrong

One people. Harding and Sims-Williams support retaining ancient Celts and a Celtic language family while rejecting one trans-European nation. Caesar's Celtae is important evidence for a broad self-name in part of Gaul, not permission to project the term onto every later population.

Hallstatt and La Tène as ethnic maps. The European Iron Age handbook and Sims-Williams show why material distributions, language and identity must be argued separately. The correction denies neither migration nor cultural connection. It denies that one object supplies all conclusions at once.

Primitive and unwritten. Bibracte, Heuneburg, Hallstatt and wider settlement evidence establish farming systems, specialised production and large centres before Roman conquest. Lepontic, Gaulish, Celtiberian, coin legends and ogham disprove the literal claim of no writing. The narrower absence is a continuous native political history.

Druids and Stonehenge. The chronological correction is secure. Caesar, Tacitus and Aldhouse-Green support retaining historical learned and religious roles in Gaul and Britain while rejecting a Europe-wide church. Creer's claim of deliberate omission is noted as one interpretation of Caesar's narrative silence, not evidence for a documented suppression campaign.

Naked and blue warriors. Polybius 2.28 describes the Gaesatae fighting naked at Telamon while the Insubres and Boii wore trousers and cloaks. Caesar, Gallic War 5.14, claims that Britons stained themselves with vitrum and appeared blue. P. C. N. Stewart and the study by F. B. Pyatt and colleagues note that the identification of vitrum with woad and the archaeological evidence for body paint are uncertain. The passages do not establish a Celtic uniform.

Rome could never hold them. Roman government in Gaul and much of Britain lasted for centuries. The corrected account joins local political openings to Roman manpower, supply, engineering and coercion. Political control, language shift and cultural identity changed at different speeds.

Ancient survival or modern fiction. Edwards, Harris, Collis, Curley and Laing support a layered answer. Languages and inherited traditions are real; the modern collective category was reconstructed through scholarship, revival and invention. Later construction is part of the identity's history rather than evidence that every modern claim is false.

Use It

The analytical method now rests on five subject-specific operations. It distinguishes kinds of connection; corrects preservation bias; separates movements of people, language, objects and genes; combines local incentives with external capacity; and examines modern inheritance without a purity test. These lenses are diagnostic. They do not turn Iron Age Europe into a direct analogy for modern politics.

The final question, "Celtic in what sense?", is designed to narrow claims rather than dismiss them. Language, archaeology, ancient naming and modern identity can each justify the term for different purposes. The evidence becomes misleading when one purpose is made to prove all the others.

Terms

Celt, Gaul and Galatian. These terms overlap historically without being interchangeable. Celt derives from Greek Keltoi; Gaul from Roman Galli; Galatian identifies the Celtic-speaking communities established in Anatolia and the region named after them.

Celtic, Continental and Insular. Celtic is the language family. Continental Celtic groups extinct ancient mainland languages. Insular Celtic is a linguistic classification for the languages of Britain, Ireland and their descendants, not proof of one ancient island people.

Goidelic and Brittonic. Goidelic contains Irish, Scottish Gaelic and Manx. Brittonic contains Welsh, Cornish and Breton. The Q-Celtic and P-Celtic labels describe a useful sound contrast but do not replace the fuller linguistic classification.

Hallstatt, La Tène and oppidum. Hallstatt and La Tène are archaeological periods and material traditions named after sites. Oppidum is a Latin and archaeological term for a fortified town. None functions as an ethnic passport.

Matrilocality. The term describes residence with or near the woman's family after partnership. Cassidy and colleagues reconstruct it directly at Durotrigian burial sites and identify comparative evidence consistent with a wider British pattern. Residence should not be translated into inheritance or matriarchy without separate evidence.

Druid, nemeton and votive deposit. Druid denotes an attested learned and religious role. Nemeton is associated with sacred groves or sanctuaries. A votive deposit is an offering removed from ordinary use. The terms identify different parts of practice and should not be collapsed into one institution.

Vitrum and woad. Caesar's vitrum produced a blue appearance in his account of Britons. Woad is the conventional translation, but the identification and practical evidence for body painting remain disputed.

Polity, tribe and Romanisation. Polity is the least prejudicial term for an organised political community. Tribe remains common but can imply simplicity or kinship beyond the evidence. Romanisation is retained only as shorthand for coercive, selective and regionally uneven change under Roman rule.

Celtic Revival. The term covers scholarly classification, language recovery, artistic reuse, political movements and creative invention from the early modern period onwards. It does not name one organisation or ideology.

Bibliography

Primary sources in translation

Caesar, Julius. The Gallic War. Translated with an introduction and notes by Carolyn Hammond. Oxford: Oxford University Press, 2008.

Polybius. The Histories. Translated by Robin Waterfield. Edited by Brian McGing. Oxford: Oxford University Press, 2010.

Tacitus. Agricola and Germany. Translated by Anthony R. Birley. Oxford: Oxford University Press, 2009.

Tacitus. The Annals: The Reigns of Tiberius, Claudius, and Nero. Translated by J. C. Yardley. Introduction and notes by Anthony A. Barrett. Oxford: Oxford University Press, 2008.

Modern works

Aldhouse-Green, Miranda. Rethinking the Ancient Druids: An Archaeological Perspective. Cardiff: University of Wales Press, 2021.

Armit, Ian. "Ritual Violence and Headhunting in Iron Age Europe." In The Cambridge World History of Violence, edited by Garrett G. Fagan, Linda Fibiger, Mark Hudson and Matthew Trundle, 441-459. Cambridge: Cambridge University Press, 2020.

Beltrán Lloris, Francisco, Carlos Jordán Cólera, Borja Díaz Ariño and Ignacio Simón Cornago. "The Novallas Bronze Tablet: An Inscription in the Celtiberian Language and the Latin Alphabet from Spain." Journal of Roman Archaeology 34, no. 2 (2021): 713-733.

Blom, Alderik H. "Gaulish in the Late Empire (c. 200-600 CE)." In Languages and Communities in the Late-Roman and Post-Imperial Western Provinces, edited by Alex Mullen and George Woudhuysen, 129-154. Oxford: Oxford University Press, 2023.

Brett, Caroline, with Fiona Edmonds and Paul Russell. Brittany and the Atlantic Archipelago, 450-1200: Contact, Myth and History. Cambridge: Cambridge University Press, 2021.

Cassidy, Lara M., et al. "Continental Influx and Pervasive Matrilocality in Iron Age Britain." Nature 637 (2025): 1136-1142.

Collis, John. The Celts: Origins, Myths and Inventions. Stroud: Tempus, 2003.

Creer, Tyler. "The Suppression of the Druids in Caesar's Gallic War." Classical Quarterly 73, no. 1 (2023): 169-183.

Cunliffe, Barry. The Ancient Celts. 2nd ed. Oxford: Oxford University Press, 2018.

Curley, Thomas M. Samuel Johnson, the Ossian Fraud, and the Celtic Revival in Great Britain and Ireland. Cambridge: Cambridge University Press, 2009.

Edwards, Nancy. "Edward Lhuyd and the Origins of Early Medieval Celtic Archaeology." The Antiquaries Journal 87 (2007): 165-196.

Fitzpatrick, Andrew P. "Migration." In The Oxford Handbook of the European Iron Age, edited by Colin Haselgrove, Katharina Rebay-Salisbury and Peter S. Wells, 1137-1186. Oxford: Oxford University Press, 2023.

Gretzinger, Joscha, Felicitas Schmitt, Angela Mötsch, et al. "Evidence for Dynastic Succession among Early Celtic Elites in Central Europe." Nature Human Behaviour 8 (2024): 1467-1480.

Harding, D. W. "The Celtic Debate: History, Linguistics, and Archaeology." In Re-writing History: Changing Perceptions of the Past, 185-202. Oxford: Oxford University Press, 2020.

Harris, James. "Language, Historical Culture and the Gentry of Later Stuart Cornwall and South-West Wales." Historical Research 95, no. 269 (2022): 348-369.

Harris, W. V. "Roman Expansion in the West." In The Cambridge Ancient History, Volume VIII: Rome and the Mediterranean to 133 B.C., 2nd ed., edited by A. E. Astin, F. W. Walbank, M. W. Frederiksen and R. M. Ogilvie, 107-162. Cambridge: Cambridge University Press, 1989.

Haselgrove, Colin, Katharina Rebay-Salisbury and Peter S. Wells, eds. The Oxford Handbook of the European Iron Age. Oxford: Oxford University Press, 2023.

Jørgensen, Anders Richardt. "Celtic." In The Indo-European Language Family: A Phylogenetic Perspective, edited by Thomas Olander, 135-151. Cambridge: Cambridge University Press, 2022.

Laing, Lloyd. The Archaeology of Celtic Britain and Ireland c. AD 400-1200. Cambridge: Cambridge University Press, 2006.

Mitchell, Stephen. Anatolia: Land, Men, and Gods in Asia Minor. Volume I: The Celts in Anatolia and the Impact of Roman Rule. Oxford: Clarendon Press, 1993.

Pyatt, F. B., E. H. Beaumont, D. Lacy, J. R. Magilton and P. C. Buckland. "Non Isatis Sed Vitrum, or the Colour of Lindow Man." Oxford Journal of Archaeology 10, no. 1 (1991): 61-73.

Russell, Miles, Martin Smith, Ellen Hambleton, Paul Cheetham and Heather Tamminen. "Brutalised, Bound and Bled: A Case of Later Iron Age Human Sacrifice from Winterborne Kingston, Dorset." The Antiquaries Journal 104 (2024): 40-66.

Sims-Williams, Patrick. "An Alternative to 'Celtic from the East' and 'Celtic from the West'." Cambridge Archaeological Journal 30, no. 3 (2020): 511-529.

Sims-Williams, Patrick. "Bronze- and Iron-Age Celtic-Speakers: What Don't We Know, What Can't We Know, and What Could We Know? Language, Genetics and Archaeology in the Twenty-First Century." The Antiquaries Journal 92 (2012): 427-449.

Stewart, P. C. N. "Inventing Britain: The Roman Creation and Adaptation of an Image." Britannia 26 (1995): 1-10.

Wacher, John. "Britain 43 B.C. to A.D. 69." In The Cambridge Ancient History, Volume X: The Augustan Empire, 43 B.C.-A.D. 69, 2nd ed., edited by Alan K. Bowman, Edward Champlin and Andrew Lintott, 503-516. Cambridge: Cambridge University Press, 1996.

Archaeological and institutional resources

Bibracte EPCC. Bibracte throughout Time.

British Museum. The Battersea Shield.

British Museum. The Snettisham Great Torc.

Musée du Pays Châtillonnais - Trésor de Vix. The Vix Burial and Krater.

MuséoParc Alésia. The Archaeological Site of Alésia.

National Museum of Ireland. Ogham Code.

National Museum of Ireland. The Broighter Collar and The Broighter Boat.

Natural History Museum Vienna. Hallstatt Archaeological Research.

Amgueddfa Cymru - Museum Wales. Ancient Druids of Wales and Llyn Cerrig Bach.