Banat, on a 6.75-metre scroll
The only Roman map of Banat strings ten stations in a row. Not one kept its name.
A medieval copyist reproduced the Empire's road map. On it, the road from the Danube to Sarmizegetusa has ten stops, each with its name and the number of miles to the next. Follow them on the parchment, then lay them on the ground and see whether the miles fit.
What the Vienna scroll is
The Tabula Peutingeriana is the only surviving map of the Roman public roads, the cursus publicus on which the imperial couriers ran. The original is lost; what survives at the Austrian National Library, shelfmark Cod. 324, is a parchment copy from the late twelfth century.
Of the twelve sheets, eleven survive, joined into a strip 6.75 metres long and only 34 centimetres high. The old attribution — a monk of Colmar, in 1265 — is now disputed, and the copyist is unknown.
The map was found in 1494 by the humanist Conrad Celtis in a library at Worms and passed in 1508, after his death, to his friend Konrad Peutinger of Augsburg, who gave it his name. The Peutinger family kept it for two hundred years; sold in 1714, it entered the Imperial Court Library at Vienna after the death of Prince Eugene of Savoy in 1737. In 2007 it entered UNESCO's Memory of the World Register. The Roman original it copies belongs to the late Empire, the fourth or fifth century.
Like any ancient itinerary, the map looks nothing like an atlas: roads are straight red lines, stations are symbols and names, and the figure written after each name is the distance in Roman miles to the next station. Geography is crushed to fit the strip. Yet for Dacia, this strip is the only surviving ancient road network: some 49 places, as counted by the historian Florin Fodorean. Tiuisco appears twice, once on each road that touches it.
One clarification from the start: the Banat road leaves from Dierna, the Danube harbour on the site of today's Orșova — on the map, Tierua. Drobeta, on the map Drubetis, is not on this road: it stands on the Danube road below and heads a second Dacian road, the one crossing Oltenia toward Romula. It gets a place of its own, so the count stays clean: ten stations on our road, from Tierua to Sarmategte.
- Where
- Vienna, Austrian National Library, Cod. 324
- Size
- 6.75 × 0.34 m; eleven of twelve parchment sheets
- When
- Copy: late twelfth century (the “Colmar, 1265” attribution, disputed). Roman original: the late Empire, fourth–fifth century
- Dacia on the map
- Three roads: Dierna–Tibiscum–Sarmizegetusa–Porolissum, Lederata–Tibiscum, and Drobeta–Romula–Apulum
Ten stops on the parchment
Read each stop as the copyist wrote it, with the number of miles that follows. For each one, the page shows what the map says, what Ravenna says, in the other great tradition, and where the place is today.
Stop 1 of 10 · XI miles to the next
Dierna
It starts on the Danube. Dierna was a fort and a harbour on the site of today's Orșova. Its name lives nowhere: not even the Cerna river keeps it, since that name comes from the Slavic word for “black”. Tierua is the only stop missing from Ravenna's list.
- On the parchment
- Tierua
- At Ravenna
- absent
- Today
- Orșova
Stop 2 of 10 · XIIII miles to the next
Ad Mediam
“At the middle”, probably the middle of a road. Specialists now put it at Băile Herculane; older books looked for it at Mehadia.
- On the parchment
- ad media
- At Ravenna
- Medilas
- Today
- Băile Herculane
Stop 3 of 10 · IX miles to the next
Praetorium
The praetorium was the commander's residence. The walls of the fort survive at Mehadia, but the name has gone out: no village and no river bears it.
- On the parchment
- Pretorio
- At Ravenna
- Pretorich
- Today
- Mehadia (the fort)
Stop 4 of 10 · IX miles to the next
Ad Pannonios
“At the Pannonians”, probably after a unit of Pannonians stationed here. The fort is at Teregova, in the Timiș valley; nothing of the Roman name survives in today's.
- On the parchment
- ad Pannonios
- At Ravenna
- Panonin
- Today
- Teregova
Stop 5 of 10 · XI miles to the next
Gaganae
Also in the Timiș valley. The spelling differs from one manuscript to the other, but the word is the same.
- On the parchment
- Gaganis
- At Ravenna
- Gazanam
- Today
- Armeniș
Stop 6 of 10 · XIIII miles to the next
Masclianis
Two close spellings of the same name, a few kilometres from the previous stop.
- On the parchment
- Masclianis
- At Ravenna
- Masclunis
- Today
- Slatina-Timiș
Stop 7 of 10 · XIIII miles to the next
Tibiscum
Two roads meet here, the one from Dierna and the one from Lederata, on the Danube. That is why the copyist wrote the name twice, once on each. The town lost its name, but the river it was born on, the Tibiscus, still flows as the Timiș.
- On the parchment
- Tiuisco
- At Ravenna
- Tibis
- Today
- Jupa, near Caransebeș
Stop 9 of 10 · XV miles to the next
Pons Augusti
“Augustus's bridge”. Two villages dispute the place, Voislova and Marga, and the Barrington Atlas gives Marga with a question mark. The next leg, XV miles, is the longest on the road.
- On the parchment
- Ponte augusti
- At Ravenna
- Augusti, under Moesia
- Today
- Voislova or Marga (uncertain)
Stop 10 of 10 · the last on this road
Sarmizegetusa
The capital of Roman Dacia, marked on the map with a sign of two houses. The village on the ruins was called Grădiște until 1941, when the state gave it the ancient name.
- On the parchment
- Sarmategte
- At Ravenna
- Sarmazege, under Moesia
- Today
- Sarmizegetusa (since 1941)
Ten stops, nine numbers: 11 + 14 + 9 + 9 + 11 + 14 + 14 + 8 + 15 = 105 miles. On the parchment, every leg looks equally simple. On the ground, do they fit?
The road on the ground
On the map, the road is a broken red line. On the ground, it is the Cerna–Timiș corridor: from Orșova up the Cerna, past Băile Herculane, then over the watershed into the Timiș valley, through Teregova toward Caransebeș and Jupa, and from there up the Bistra valley, through Zăvoi and Voislova, into the Hațeg depression.
Here are the ten stops, placed where the Barrington Atlas and the Pleiades database find them today. The line joins them in order; it is not the route of the Roman road. Notice how some stops have drawn together and others apart: on the parchment, every leg looked equally simple.
From Orșova to Băile Herculane the copyist counts XI miles, 16 km, and the straight line measures 17: it fits. The next leg, XIIII miles (21 km) to Praetorium, would call for places 21 km apart; the proposed ones stand 5 km apart, or 7 if you take the fort at Mehadia rather than the village.
Then comes a leg of IX miles (13 km) to Teregova, across 29 km of straight line, which no road can do. The two mismatches cancel: from Băile Herculane to Teregova the parchment gives 23 miles (34 km) and the straight line measures 33 km. Either one stop is misplaced on the ground, or a number was spoiled in copying.
At Teregova, Armeniș and Slatina-Timiș the stops stand 5-6 km apart, but the legs between them count IX and XI miles (13 and 16 km). The leg from Slatina-Timiș to Jupa, XIIII miles (21 km), covers 25 km of straight line: at the edge of the pins' error, but not beyond it.
Taken together, from Teregova to Jupa the parchment gives 34 miles (50 km) and the straight line measures 35 km. A mountain road with bends can make that difference. The total is good; its split into stops, less so.
Three legs touch places with a question mark (Zăvoi, Marga or Voislova), so we treat them apart: they put the proposals to the test, they do not prove them. From Jupa to Zăvoi the parchment gives XIIII miles (21 km) and the straight line 18 km. From Zăvoi to Marga, VIII miles (12 km) against 9 km. From Marga to Sarmizegetusa, XV miles (22 km) against 21 km. All three fit.
With Voislova in place of Marga, the first leg (12 km) would join two places 4 km apart, too close for it, and the second (22 km) would only just manage: 25 km of straight line. The miles favour Marga, but a single miscopied number would change the conclusion.
The miles on the parchment add up: from Tierua to Sarmategte, the copyist strings 105 Roman miles — XI, XIIII, IX, IX, XI, XIIII, XIIII, VIII, XV — about 155 of today's kilometres, at 1.48 km to the mile. Each leg is a short day's march or a long day's cart ride: between eight and fifteen miles, depending on the climb. Some miles carry copyists' slips, though — a dropped X or I, as Fodorean shows — so the total should be read as the copyist's reckoning.
The 155 km on the parchment are 16 per cent more than the legs in straight line laid end to end (135 km). The total does not contradict the ground, but does not confirm it either: the errors between legs cancel. Between secure places, only two of the six legs fit.
Leg by leg: miles and kilometres
Each row sets the leg on the parchment (miles × 1.48 km) beside the straight-line distance between the proposed places. One rule for every leg: a road cannot be shorter than the straight line (we allow 4 km of error, 2 km at each point), and we did not accept one more than twice as long (a threshold we chose; the whole road comes out at 1.16). Rows marked † touch a place with a question mark: the verdict tests the proposal, it does not prove it.
| Leg | On the parchment | Straight line | Fit |
|---|---|---|---|
| Dierna → Ad Mediam | XI miles · 16 km | 17 km | fits |
| Ad Mediam → Praetorium | XIIII miles · 21 km | 5 km | the places are too close for so long a leg |
| Praetorium → Ad Pannonios | IX miles · 13 km | 29 km | the straight line is longer than the leg |
| Ad Pannonios → Gaganae | IX miles · 13 km | 6 km | the places are too close for so long a leg |
| Gaganae → Masclianis | XI miles · 16 km | 5 km | the places are too close for so long a leg |
| Masclianis → Tibiscum | XIIII miles · 21 km | 25 km | fits |
| Tibiscum → Agnaviae † | XIIII miles · 21 km | 18 km | fits |
| Agnaviae → Pons Augusti † | VIII miles · 12 km | 9 km | fits |
| Pons Augusti → Sarmizegetusa † | XV miles · 22 km | 21 km | fits |
The road was guarded by forts: Dierna, Praetorium, Ad Pannonios, Tibiscum, Pons Augusti. At Jupa, near Caransebeș, the ruins of Tibiscum form an archaeological reserve. At Mehadia, the walls of Praetorium emerge from the digs. Elsewhere, the Roman line fades beneath highways and fields.
Distances are straight lines between the points that Pleiades and the Barrington Atlas give for the proposed places; the points are good to about 1-2 km, and for Praetorium the point is the village of Mehadia, not the fort. The mile is 1.48 km.
Ten names, one by one
Each station is judged on two axes. First the name: does it appear only on the Tabula, or also in the other great tradition, the Geography of the Anonymous of Ravenna, from the seventh century? Then the place: is the site secure, or an attribution with a question mark? The filters below work together: pick one on each axis and read the combination.
10 stations · 9 names in both traditions · 8 secure sites
Ten stations.
-
Dierna
Orșova
On the map: Tierua · in Ravenna: absent
name: on the Tabula onlysite: secure
A Danube fort and harbour on the site of today's Orșova. It is the only station on the road whose name does not appear in Ravenna too: the map reads Tierua, while modern gazetteers list it under Dierna. The ancient name survives in no present-day toponym — not even in the Cerna, whose name comes, according to the DEX dictionary, from the Slavic word for “black”.
XI miles to Ad Mediam; the road to Sarmizegetusa starts here.
-
Ad Mediam
Băile Herculane
On the map: ad media · in Ravenna: Medilas
name: in both traditionssite: secure
“At the middle”, probably the middle of a road. The current consensus puts the station at Băile Herculane — so the Barrington Atlas has it, after the Tabula Imperii Romani, and the Heidelberg epigraphic database with it — while the Mehadia fort counts as the next station, Praetorium. Older literature placed both at Mehadia, as the English Wikipedia still does. The link between the name Mehadia and Ad Mediam is disputed too: the historian Xenopol defended it; most linguists derive today's Mehadia from the medieval Hungarian Michald.
XIIII miles to Praetorium.
-
Praetorium
—
On the map: Pretorio · in Ravenna: Pretorich · site: the Mehadia fort
name: in both traditionssite: secure
An empire-wide generic name: the praetorium was the headquarters, the commander's residence. The Mehadia fort, excavated by Doina Benea and the Banatica team, keeps its walls — the 1993 monograph fixed its identification — but the name died without a trace: no village and no river still carries it.
IX miles to Ad Pannonios.
-
Ad Pannonios
—
On the map: ad Pannonios · in Ravenna: Panonin · site: Teregova
name: in both traditionssite: secure
“At the Pannonians”, probably after a Pannonian unit stationed here. The fort stands at Teregova, in the Timiș valley; today's place name keeps nothing of the Roman one. The word itself reached us by two routes: on the map and in Ravenna.
IX miles to Gaganae.
-
Gaganae
—
On the map: Gaganis · in Ravenna: Gazanam · site: Armeniș
name: in both traditionssite: secure
The name appears in two traditions — Gaganis on the map, Gazanam in Ravenna — and the place is at Armeniș, in the Timiș valley (Barrington, after the Tabula Imperii Romani). The spelling differs, the word is the same.
XI miles to Masclianis.
-
Masclianis
—
On the map: Masclianis · in Ravenna: Masclunis · site: Slatina-Timiș
name: in both traditionssite: secure
The twin of its neighbour above: Masclianis on the map, Masclunis in Ravenna, the site at Slatina-Timiș (Barrington, after the Tabula Imperii Romani). The name appears almost unchanged in both traditions.
XIIII miles to Tibiscum.
-
Tibiscum
Jupa · Timiș
On the map: Tiuisco, twice · in Ravenna: Tibis
name: in both traditionssite: secure
A Dacian town recorded by Ptolemy (III, 8, 10), then a Roman fort and municipium at Jupa, near Caransebeș — a municipium since Septimius Severus. Here our road joined the one coming from Lederata on the Danube; that is why the copyist wrote Tiuisco twice, once on each road. The town's name died with the town, but the river it was born on, the Tibiscus, still flows as the Timiș.
XIIII miles to Agnaviae; the junction where the two Dacian roads meet.
-
Pons Augusti
—
On the map: Ponte Augusti · in Ravenna: Augusti, under Moesia · site: Voislova–Marga?
name: in both traditionssite: uncertain
“The bridge of Augustus”: an imperial bridge guarded by a small fort. The Voislova fort is assigned to it in the fort lists; Barrington notes “Marga?” instead — the two neighbouring villages dispute the bridge, in print. The name left no heir.
XV miles to Sarmizegetusa, the longest leg on the road.
-
Sarmizegetusa
Sarmizegetusa
On the map: Sarmategte · in Ravenna: Sarmazege, under Moesia
name: in both traditionssite: secure
Colonia Ulpia Traiana Augusta Dacica Sarmizegetusa, the capital of Roman Dacia. In Ravenna the name appears as Sarmazege, under Moesia, not Dacia — because, on Nistorescu's account, the area had passed under Moesia Prima. The medieval village on the ruins was called Grădiște; after scholars showed this was the ancient capital, the state officially changed the name to Sarmizegetusa in 1941. On the 1872 plan of the ruins, reproduced below, the village is still called Gredistie.
XIIII miles to Ad Aquas; the road goes on toward Apulum.
The new reckoning: nine out of ten names reached us by two routes — the map and Ravenna — and only eight out of ten places are secure. In other words, it is easier to read a station's name in two manuscripts than to find its place on the ground.
The second tradition deserves a full look. The Geography of the Anonymous of Ravenna, from the late seventh century, lists the stations without distances: Drubetis, Medilas, Pretorich, Panonin, Gazanam, Masclunis, Tibis — then, under Moesia: Sarmazege, Augmonia, Augusti (Pinder–Parthey edition, 1860). The only one missing is Tierua. For Dacia, Ravenna seems to have used an itinerary from after 168–170, as Fodorean argues from the presence of Macedonica; and the fact that the Anonymous cites a certain Castorius as a source for the East led Konrad Miller to propose him as the author of the map's prototype — a hypothesis Fodorean sets aside in the Plekos review (2011).
Two names on place signs
Sarmizegetusa and Drobeta are the only names on the strip you can read today on a place sign — although Drobeta stands on the Danube road, not this one. Both were restored by the modern state: Sarmizegetusa in 1941, replacing Grădiște, after archaeologists proved this was the capital of Dacia; Drobeta in 1972, glued in front of Turnu Severin. Their Transylvanian pair is Cluj-Napoca, from 1974.
The plan below, drawn by Richard Kiepert in 1872 for the Berlin Geographical Society, shows the ruins of Sarmizegetusa when the village was still called Gredistie. The ancient name lived only in books. Today it lives on the station, the town hall, and the signposts.
None of the ten names stayed untouched for nineteen centuries, but nine reached us through two manuscripts. The places fared worse: two stations are still looking for a secure site, and between secure places the miles fit the ground on only two legs out of six. The map holds them all together, in red ink, on 34 centimetres of parchment.
Sources
Map readings are checked three times: against Florin Fodorean's transcription, against a reading of the facsimile, and against the OmnesViae dataset, compared with the high-resolution reproduction. The Ravenna forms come from the Pinder–Parthey edition (1860), checked in full. Station-name renderings are by Marius Comper.
Sources and facsimiles
- Florin Fodorean, “Tabula Peutingeriana and the Province of Dacia”, Acta Musei Napocensis 39–40/I (2002–2003): the transcription of segments VII–VIII and the count of places (49). biblioteca-digitala.ro
- Tabula Peutingeriana, Konrad Miller edition 1887, in the Bibliotheca Augustana: the scans of segments VI–VII and VII–VIII, read station by station. tha.de/~harsch
- UNESCO, Memory of the World: the late-twelfth-century copy, 11 segments, 6745 mm, the only surviving map of the cursus publicus. unesco.org
- The Austrian UNESCO commission: the 2007 inscription. unesco.at
- The Anonymous of Ravenna, Cosmographia, Pinder–Parthey edition, Berlin, 1860 (scan): the Dacia (IV, 14) and Moesia (IV, 7) lists; Tierua's absence, checked against the full text; Castorius as a source for the East. archive.org
- Wikipedia: “Tabula Peutingeriana” (en) — the 1494/1508/1714/1737 chain, the 1591 and 1598 printings, the “Colmar, 1265” attribution as a disputed hypothesis. en.wikipedia.org/wiki/Tabula_Peutingeriana
Tools: gazetteers, databases, dictionaries
- The OmnesViae dataset (MIT, René Voorburg): the third witness for legs and readings — Tierua, Agnauie — and for the 105 total; the high-resolution fragments, checked by eye. codeberg.org/renevoorburg/omnesviae
- Barrington Atlas of the Greek and Roman World, ed. R. J. A. Talbert, Map-by-Map Directory, Map 21 Dacia–Moesia (Princeton UP, 2000): the modern identifications and question marks — Dierna–Orșova, Ad Mediam–Băile Herculane, Praetorium–Mehadia, Teregova, Armeniș, Slatina-Timiș, Jupa, Zăvoi?, Marga? (the Dacia entries are reprinted and discussed in Fodorean 2022, below).
- The Heidelberg epigraphic database (EDH), inscription HD044779: Ad Mediam at Băile Herculane. edh.ub.uni-heidelberg.de
- Smith's Dictionary of Greek and Roman Geography (Perseus): Tibiscum (Ptol. 3.8.10), Tibiscus (3.8.1). perseus.tufts.edu
- dexonline: Cerna, from the Slavic for “black” (Șăineanu). dexonline.ro/definitie/cerna
- Wikipedia: the fort list — Pons Augusti at Voislova; the Zăvoi fort, with no ancient name assigned. en.wikipedia.org/wiki/List_of_castra_in_Romania
- Pleiades (CC BY 3.0): coordinates of the proposed places — 207078 Dierna, 206884 Ad Mediam, 207364 Praetorium, 206890 Ad Pannonios, 207122 Gaganae, 207270 Masclianis, 207495 Tibiscum, 206897 Zăvoi, 207527 Sarmizegetusa. Pons Augusti: OmnesViae (Marga) and the Wikipedia list of forts (Voislova). pleiades.stoa.org
- Miller 1887 scan, 26,381 × 2,014 px, public domain, from which the zoom levels are cut. commons.wikimedia.org
- Relief: Mapzen Terrain Tiles on AWS Open Data (SRTM and other sources), shaded by us. registry.opendata.aws. Rivers: © OpenStreetMap contributors, ODbL. openstreetmap.org/copyright
- Calculation: straight-line distance (haversine) between the Pleiades points, set against miles × 1.48 km; the mile links are those of OmnesViae, checked against the scan.
Studies and documents
- Florin-Gheorghe Fodorean, “Dacia in the Barrington Atlas”, Journal of Ancient History and Archaeology 9.3 (2022): the two branches of the imperial road; the index critique (including Agnaviae/Acmonia, Zăvoi?). jaha.org.ro
- Laurențiu Nistorescu, “Geographia Ravennantis și Dacia”, Qvaestiones Romanicae VIII (2020): the two routes; the Drobeta–Tibiscum–Acmonia route, preserved under Moesia Prima. diacronia.ro
- Florin Fodorean's review of Talbert, Rome's World, Plekos 13 (2011): Ravenna's Dacia source, from after 168–170; the Castorius hypothesis, set aside. plekos.uni-muenchen.de
- Bryn Mawr Classical Review 2009.06.07: the Colmar copy, around 1200 — the cautious dating. bmcr.brynmawr.edu
- “A few Dacian toponyms in Tabula Peutingeriana”, Oltenia XVII (2010): the current consensus — Praetorium at Mehadia, Ad Mediam at Băile Herculane. biblioteca-digitala.ro
- Doina Benea and the Banatica team: the Praetorium fort at Mehadia. banatica.ro
- M. Macrea, N. Gudea, I. Motu, Praetorium. Castrul și așezarea romană de la Mehadia (1993), via Barrington.
- Adrian Ardeț, “The Limes Program and the Integration of the Tibiscum Archaeological Heritage into UNESCO” (Belgrade, 2022): Tibiscum, a municipium under Septimius Severus; museum and research base. doi.fil.bg.ac.rs
- Oliviu Felecan, “Oikonymic Transformations in Romania”, Names 65/2 (2017): Turnu Severin until 1972; Cluj-Napoca from 1974. ans-names.pitt.edu
- Wikipedia: “Drobeta-Turnu Severin”, the restoration of the Drobeta name on 15 May 1972. en.wikipedia.org/wiki/Drobeta-Turnu_Severin
- The Hunedoara county planning memoir: the renaming of Grădiște to Sarmizegetusa, 1941. cjhunedoara.ro
- M. Buza, “Toponimia geografică”, Revue Roumaine de Géographie 55/1: Dierna become Orșova, Agnaviae become Zăvoi (unqualified; Barrington adds a question mark). rjgeo.ro
- Wikipedia: “Mehadia”, the disputed etymology (Xenopol, Hungarian Michald). en.wikipedia.org/wiki/Mehadia