The stratigraphy of river and village names
Romania's ten longest rivers: eight had their names written down before AD 600, centuries before any present-day village is mentioned. Of the next 140 rivers I could assess, six. Water remembers longer than land.
The rule of length
A great river is known to everyone on its banks, whatever language they speak there. Its name passes from one people to the next like a tool too useful to throw away. A brook and a village belong to the people beside them, and whoever comes next names them again. So when you line up the 150 rivers on this map by length and colour each name by the stratum it comes from, the gold gathers on the left.
The data could not be loaded; the table below and the text still stand.
Each bar is a river, longest to shortest; the height is its length in kilometres inside the country. The Danube, at 1,075 kilometres, runs off the scale. Press a bar to read the attestation. The criterion favours big rivers: a brook with an equally old name may never have reached a Greek geographer's pen, so the gold here counts attestations, and the true age of the small names stays open.
| # | River | km | First written form, or etymon | Stratum |
|---|---|---|---|---|
| 1 | Dunărea | 1,075 | Ἴστρος; Danuvius (430 BC) | ancient |
| 2 | Mureș | 761 | Μάρισος; Marisia (AD 18) | ancient |
| 3 | Prut | 742 | Πορατα / Πυρετός (430 BC) | ancient |
| 4 | Olt | 615 | Ἄλουτας; Aluta (AD 150) | ancient |
| 5 | Siret | 559 | Τιάραντος; Ἱέρασος (430 BC) | ancient |
| 6 | Someș | 465 | Samus (Samum) (AD 200) | ancient |
| 7 | Ialomița | 417 | Slavic ilov- “clay” | Slavic |
| 8 | Argeș | 350 | Ὀρδησσός (430 BC) | ancient |
| 9 | Jiu | 339 | open question | disputed |
| 10 | Buzău | 302 | Μουσαῖος (AD 372) | ancient |
Lengths are those of the Romanian Water Cadastre Atlas, as given by the Wikipedia list of rivers by length; for rivers under 144 kilometres, the length is computed from OpenStreetMap geometry. The Tisza, 966 kilometres long in all, has only 65 on Romanian territory and lands in 41st place.
Continuity is no straight line from the riverbank to today. Between the last ancient text and the first medieval charter, the names lived in Slavic mouths, then Hungarian ones, and each language shaped them by its own rules. The form Olt sits closer to Jordanes's Aluta than to Ptolemy's Alutas, and Transylvania's Germans kept an Alt that shows the same route. Someș continues Samus with a shifted vowel, Mureș comes from Marisos either directly or through Hungarian Maros, and Hungarian linguists require exactly that passage for Someș and Mureș, on the strength of a fourteenth-century vowel shift. What survived is the name; its road through the intermediary languages is, for each river, a matter of debate that the core sample records.
The Ialomița carries a Slavic name, from the word for clay, and is the only river in the top ten with a certain christening after 600. Some look for it in Herodotus's Naparis, but the match is a guess. The Jiu has no ancient form to continue: Ptolemy maps a Rhabon roughly in its place, and Jordanes a Gilpil, but neither yields the form Jiu by the known rules of the language. An old name, very probably, but an unwritten one.
Beyond the top ten, ancient attestation is rare and, each time, comes with an explanation. The Timiș and the Crișul Alb are big lowland rivers, noted by Ptolemy and Jordanes. The Tisza is huge, it simply passes briefly through Romania. The Bârzava survives in the one sentence left of Trajan's own book on the Dacian wars: “inde Berzobim, deinde Aizi processimus”. The Ampoi kept the name of the Roman gold town, Ampelum, after the town itself became Zlatna, which is gold again, in Slavic. The Motru is linked to Ptolemy's Amutrium, an old identification that still cannot be checked.
Six strata
Each stratum has a recognisable pattern and a region where it thickens. The number on each card is what can be counted with a declared rule, over the 18,214 GeoNames places and the 150 rivers on the map. Press “Show on the map” to see where a stratum lies.
Names written by Herodotus, Strabo, Pliny, Ptolemy, Jordanes, in a martyr's passion and on a fort inscription. Nearly all belong to big rivers. Today's forms continue them directly or, in a few cases, through Hungarian.
The middle and small rivers, then the villages: Bistrița, Dâmbovița, Târnava, Prahova, Craiova, Târgoviște. The suffixes -ov, -ova, -ovăț and -ăuți are Slavic. Petrovici showed that h for g, as in Dorohoi and Horodiștea, betrays an East Slavic dialect, and Vâlcea a Bulgarian one.
A thin stratum, but stuck to the plain: the rivers in -ui and -lui, from the Bahlui to the Călmățui, the Teleorman, “the mad forest”, Caracal, the Bărăgan. And the surname Coman, which was the Cumans' own name, went on to make Romanian villages in -ești.
The Hungarian layer includes Romanian forms taken from Hungarian or built in parallel: Oradea from Várad, Timișoara from Temesvár; at Cluj, Kolozsvár stands alongside the older Clus/Cluș series, whose origin remains disputed. The saints in Sân- render Szent-, and of the 98 names in Sân- nearly all lie inside the Carpathian arc. Miercurea Ciuc and Baia Mare are translations.
The Saxons and Swabians rarely supplied the Romanian form of a name. More often they left a third name, living beside the other two: Hermannstadt beside Sibiu and Nagyszeben, Kronstadt beside Brașov and Brassó. Ghimbav, from Weidenbach, is one of the few exceptions.
The top stratum, the thickest: 2,631 villages in -ești, then -eni and -ani, Valea, Poiana, Lunca, Gura, Dealu. It is the stratum that fills the map, and it is often built with older materials: Curtea de Argeș puts a Latin word on an ancient river name.
All strata at once. Gold flows, green scatters everywhere, red keeps to the plain, violet stays inside the Carpathian arc, and the blue rings mark places with a parallel German name.
11,884 places, nearly two thirds, are left uncoloured: the project’s rules recognise no pattern in their names. “No recognised pattern” describes the counting rule, not the name’s origin. Origins are checked name by name; the core sample below keeps the etymon, source and uncertainty visible where research has not closed the discussion.
Core sample
Any of the 18,214 places and any of the 150 rivers. For checked names the core shows the ancient form or the etymon, the source, and the parallel Hungarian and German names where OpenStreetMap has them. For the rest it shows the pattern a rule recognised, or says plainly that there is none.
Parallel names
In Transylvania the strata did not settle one on top of the other; they stayed side by side. The same town has a Romanian name, a Hungarian one and a German one, and often the three share no root at all.
| Romanian | Hungarian | German | What is going on |
|---|---|---|---|
| Sibiu | Nagyszeben | Hermannstadt | Romanian and Hungarian continue the name of the river Cibin; German names a founder, Hermann. |
| Brașov | Brassó | Kronstadt | Two forms of the same disputed name, and a “crown town” unrelated to either. |
| Sighișoara | Segesvár | Schäßburg | Romanian comes from Hungarian; German has a name of its own. |
| Bistrița | Beszterce | Bistritz | Three languages, one name, Slavic: the fast water. |
| Sebeș | Szászsebes | Mühlbach | Romanian from the Hungarian for “fast”; German says “mill brook”. One river, two meanings. |
| Alba Iulia | Gyulafehérvár | Karlsburg | Three names of a white fortress and its masters, on the site of Roman Apulum; for centuries Romanians called it Bălgrad, in Slavic. |
| Târnava | Küküllő | Kokel | Slavic (thorn), Old Turkic (sloe), and German takes the Hungarian one. |
| Arieș | Aranyos | Goldfluss | Romanian from Hungarian; German translates: the golden river. |
The counties with the most patterned place names, for the three strata that have a strict pattern. The count is by rule, over the 18,214 GeoNames places.
How I counted
The map uses the 18,214 populated places in GeoNames for Romania, with coordinates, and 150 rivers with geometry from OpenStreetMap, extracted on 6 September 2026: every river relation in the country longer than 45 kilometres, plus the ways for rivers without a relation, plus a further 31 rivers between 35 and 45 kilometres, researched and added by hand on 8 September 2026 to bring the list to 150. Secașul Mare and Secașul Mic were added through a WaterwayMap/OSM supplement checked on 7 September; for Secașul Mic, the available line covers only the mapped segment. The lengths of the first 30 rivers are the official ones, from the Water Cadastre Atlas, as the Wikipedia list gives them; for the rest, the length is the sum of OpenStreetMap segments, which can fall short of the true figure where the map is incomplete.
“Attested before AD 600” means something precise: today's name continues a form written in a text or on an inscription from before the Slavs reached the Danube. The sources are Herodotus (4.48), Strabo (7.3.13), Pliny the Elder (4.79–80), Ptolemy (3.8), Jordanes (Getica 12.74 and 22.113), the Passion of Saint Sabas the Goth, the inscription CIL III 827 from Cășeiu and Trajan's sentence preserved by Priscian. Jordanes writes in 551, which is why the threshold sits at 600. The identifications remain the traditional ones in the literature; where they are disputed, the core sample says so.
The Slavic, Turkic, Hungarian and Romanian strata on the map are counted by declared patterns; the etymologies verified name by name live in the core sample. The patterns are these: the endings -ov, -ova, -ovăț, -ăuți for Slavic; the -ui rivers, Teleorman, Caracal, Bărăgan and the name Coman for Turkic; Sân-, -falău, -hida and “Miercurea” for Hungarian; -ești, -eni, -ani and the common words Valea, Poiana, Lunca, Gura, Dealu for Romanian. A Slavic pattern can be used by a Romanian speaker, as Petrovici showed for Bahna and for Romanați; that is why the cards say “patterned names”, and why the core sample separates names checked in dictionaries from names classified by rule. The 150 assessed rivers and the 89 places in the core sample each have a source: Iordan (1963), Petrovici (1960, 1965), Conea and Donat (1958), Bocșan and Ungureanu (2015), Kiss Lajos (1988), plus the academic and institutional sources listed in the ledger.
The parallel Hungarian and German names come from OpenStreetMap's name:hu and name:de tags and measure how complete the map is, and a missing German name may only mean that nobody added it.
This page continues Who Named Romania, which counted the patronymic suffixes and the Sân- saints on the same repertoire of places, and The Layers of a Word, on the same strata in the lexicon. The strata here can be recomputed from the project files.
Sources