Romania · Where the sources disagree

When Rome left Dacia, and who was left behind

The textbooks say 271. The coins say the province had already stopped working ten years earlier. And what happened to the people is the most bitterly argued question in the region's history.

Updated 30 August 2026 7 min read One of 9 disputes in the Romania record

In 106 AD the emperor Trajan finished a brutal two-war campaign, killed the Dacian king Decebalus, and turned the mountains and river valleys of what is now central Romania into a Roman province. The wars are carved in a spiral up Trajan's Column in Rome, which is why most people have heard of them.

A hundred and sixty-five years later, Rome left.

Two questions hang off that sentence, and they are not equally difficult. The first is when, and it has a good answer that is not the one in most textbooks. The second is what happened to everybody who lived there, and it has no settled answer at all — it is the most politically charged historical argument in this part of Europe, and it has been running for two hundred and fifty years.

Part one: the date is wrong, and interestingly so

Ask when Rome abandoned Dacia and you will be told 271, or sometimes 275, under the emperor Aurelian. The ancient writers are the reason for the wobble, and they do not agree with each other about anything much:

• Eutropius and Rufius Festus say Aurelian pulled out the population.
• The Historia Augusta and Jordanes say the army and the population.
• Jordanes, elsewhere in the same book, says only the legions went.

None of them was there. All of them were writing at least a century later.

So look at what people left in the ground instead. The Romanian archaeologist Liviu Petculescu laid the evidence out in 2017, following a conclusion Coriolan Opreanu had reached in 2004 — that the abandonment was "a process started by Gallienus and finished by Aurelianus". Gallienus was emperor from 253 to 268, which is to say before the date everybody quotes.

Here is what that evidence looks like:

• The province minted its own bronze coins from 246. The mint shuts in 257.
• Exactly one inscription from inside Dacia is later than 260. It was set up by the commander of the Thirteenth Legion — at Băile Herculane, on the road down to the Danube, and not at Apulum where his legion was actually based. A commander leaves his monument on the way out, not at home.
• Both Dacian legions turn up somewhere else. Detachments of the Fifth Macedonian and the Thirteenth are attested at Poetovio, in what is now Slovenia, after 260 — including the clerks. Armies leave their filing cabinets behind last.
• Coins of Gallienus are scarce in Dacia. Soldiers are paid in coin. If the emperor's money is not arriving, the emperor is not paying anyone there.

And the giveaway: Aurelius Victor, Orosius, Eutropius, Festus and Jordanes all record the loss of Dacia under Gallienus. Roman historians were happy to say so, because the senatorial class loathed Gallienus and enjoyed cataloguing his failures.

So what happened in 271? Aurelian marched to the Danube, formally gave up a province that had already gone, and founded a new one with the same name on the southern bank. The mint at Milan struck coins reading Dacia Felix — "Dacia the fortunate" — most likely as a bonus for the troops involved.

That is a rebrand. And Rome had done it before: after the catastrophe in the Teutoburg Forest in AD 9, the commands west of the Rhine were named Germania Superior and Germania Inferior, so that the province could look as though it had never been lost. Aurelian's Dacia is the same trick.

Part two: did anyone stay?

Now the hard question, and the one people still argue about in public.

Romanian is a Romance language. It descends from Latin, like French and Spanish and Italian, and it is spoken today in roughly the area Rome held for those 165 years. The obvious story is that the Latin-speaking provincials simply stayed, survived everything that came afterwards, and became Romanians.

The competing story is that they left, or died, or were absorbed — and that Romanian formed south of the Danube, among the far larger and much longer Romanised population of the Balkans, then moved north again in the Middle Ages.

Neither story can be proved. Here is what each side has.

The case that people stayed

• You cannot evacuate a province. Estimates put the provincial population between half a million and a million. The sources describe moving the army and the town-dwellers, not everybody. Precisely one refugee from Dacia is known by name: C. Valerius Iulianus, born at Sarmizegetusa, who ended his life at Durostorum on the Danube.
• Romanian's Christian vocabulary is inherited Latin, not borrowed Slavonic. Hard to arrange if the speakers only met Christianity later, in the Balkans.
• The big rivers keep their ancient names.
• Latin words survive for things found north of the Danube and not south of it — naphtha, gold, the bison.
• Two medieval chronicles put them there. The Gesta Hungarorum, about 1200, has "shepherds of the Romans" in the Carpathian basin when the Hungarians arrive; Simon of Kéza has Vlachs who "remained in Pannonia".

The case that they did not

Most of this is linguistic, and most of it is about things that are missing:

• The place names vanished. More than twenty settlements in England still carry names of Roman origin. In the former Dacia there are effectively none. Even the river names that did survive show signs of having been passed on through Dacian and Slavic speakers rather than directly down a Latin line.
• No Gothic loanwords. The Goths were in this area for three hundred years. Romanian took nothing from them.
• No early Slavic loanwords of the right kind either — nothing from the eastern or western branches you would expect from northern neighbours, but a heavy layer from South Slavic.
• Romanian shares its deep structure with Albanian and Bulgarian — the set of shared features linguists call the Balkan Sprachbund. Languages get that way by sitting next to each other for centuries.
• Romanians took up Old Church Slavonic as their church language, which is difficult to explain for a population living north of the Danube after 864.
• 165 years is not very long. Every other Romance language grew in a place Rome held for five centuries or more. Britain was held for 365 years and the local languages survived anyway.

Why this argument never ends

It is old and it is political. Franz Joseph Sulzer opened the sceptical case in the 1780s; Robert Rösler's Romänische Studien of 1871 made it the standard Austro-Hungarian position — and it was, not coincidentally, a position that suited a state governing Transylvania. A. D. Xenopol answered it, and Nicolae Iorga built the continuity account into the Romanian national history. Gottfried Schramm's Ein Damm bricht (1997) is the strongest modern version of the sceptical case. Dimitrie Onciul's middle position — a small population stayed, and was reinforced later by arrivals from the south — has been the reasonable compromise for a century.

None of them convinced the others, because the evidence never arrived.

The reference works say so outright. The Cambridge History of the Romance Languages (2013) concludes that the historical, archaeological and linguistic data "do not seem adequate to give a definitive answer". The Oxford Guide to the Romance Languages (2016) says where Daco-Romance originated remains uncertain. The Byzantine writer Laonikos Chalkokondyles had already given up in the fifteenth century: "I cannot tell which group migrated to the region of the other."

What would change anybody's mind

• One dated Latin text from north of the Danube between the 300s and the 800s. That five-hundred-year silence is the whole problem. A single inscription inside it would be worth more than another century of argument.
• Published cemetery populations from those centuries in Transylvania, with the burial rite and the grave goods described before anyone attaches an ethnic label — which is the step where this subject usually goes wrong.
• Ancient DNA at scale, which would at least narrow the demographic possibilities. It would tell you about ancestry, though, and this is an argument about language. People can change language without changing parents.

What this map does about it

It records the province as 106 to 271, marks it contested, and stops. There is no holding drawn for a Latin-speaking population after Rome left, because there is no attested authority to draw — and inventing one to win a continuity argument is exactly the thing this record exists not to do.

Sources

• Liviu Petculescu, "The abandonment of Dacia and the imperial propaganda", Ephemeris Napocensis XXVII (2017), 259–266.
• Coriolan H. Opreanu, "Dacia, Pannonia and Raetia in the time of Gallienus: a comparative approach", BAR International Series 1312 (Oxford, 2004), 13–18.
• Eutropius 9,15,1; Rufius Festus 8,2; Historia Augusta, Aurelian 39,7; Jordanes, Romana 217; Aurelius Victor, Liber de Caesaribus 33,3.
• Robert Rösler, Romänische Studien (1871); Gottfried Schramm, Ein Damm bricht (1997); A. D. Xenopol and Nicolae Iorga for the continuity case; Dimitrie Onciul for the middle position.
• The Cambridge History of the Romance Languages (2013); The Oxford Guide to the Romance Languages (2016).
• Origin of the Romanians — a survey of both sides with the literature attached.
• Roman Dacia.