How old is Romanian Christianity? What 379,734 Latin inscriptions show

How old is Romanian Christianity? The short answer#

The conversion of the Romanians cannot be given a year, with or without the inscriptions. What can be dated is the evidence: the martyrs of the years around 300 and the Christian inscriptions from the Danube, nearly all between 300 and 600. The words inherited from Latin carry no date of their own. Those inscriptions name soldiers no more often than the frontier’s other inscriptions do.

How old is the evidence? Around 300 there were Christians in the forts along the Danube: they are known from the calendars of martyrs and from accounts of their trials, some written later. Nearly all the inscriptions come after that: of the 176 dated Christian inscriptions on the frontier (seals included), 164 are later than 300. It is the same all over the West: at Rome 86.8% of dated Christian inscriptions are later than 300, in Gaul, on the Rhine and in Spain almost all; many are given such dates by editors because they are Christian. So the inscriptions cannot show how long there had been Christians on the Danube. They show that there were Christians there in the fourth century, and that they wrote in Latin. Romanian biserică (church) comes from basilica, a word that appears once in the Christian inscriptions from the Danube, at Sirmium, between 351 and 400.

Brought by soldiers? The inscriptions do not show it. Six of the 59 Christian epitaphs on the Danube dated after 300 name a soldier, several times the share in the rest of the empire. But among the frontier’s other epitaphs of those centuries soldiers appear at least as often: 25 of 116. Some soldiers were Christians; the inscriptions do not show that the army spread the faith.

And Orthodoxy? The Latin inscriptions of the Danube stop around 600, when the frontier fell. Most words of hierarchy and worship came later, from the Slavonic that the Church south of the Danube used from the ninth and tenth centuries on; the sources do not say exactly when they reached the Romanians. “Orthodoxy” in today’s sense is dated, by convention, from the break between Rome and Constantinople in 1054.

Did soldiers bring Christianity to the Danube? The test, with every number#

The count rests on the labels epigraphers have assigned in the Clauss/Slaby database: inscriptiones christianae (“Christian inscriptions”), milites (“soldiers”) and tituli sepulcrales (“epitaphs”). An inscription also “names a soldier” when the soldier set up the stone for a wife or child. “The other epitaphs” are those without the Christian label; not all of them belong to pagans.

The five provinces: Pannonia Superior and Inferior (Upper and Lower Pannonia), Moesia Superior and Inferior, Dacia. “After 300” means inscriptions whose date range begins in 300 or later. The 95% confidence interval shows how uncertain a share is because the inscriptions are few: the fewer, the wider. It takes no account of what has been lost or of errors in dating and labelling. Dedications are altars and other inscriptions offered to a named god; those from the Danube are mostly of the second and third centuries and cannot be set directly beside epitaphs later than 300. The “Heidelberg” rows repeat the test on the Heidelberg database’s own classification, made by a different team: its “religion” field (which puts Judaism and Christianity together) and each person’s “military personnel” status. The “text search” rows recognise soldiers by the ranks and units written in the inscription, not by the label.

The table counts all Christian epitaphs, dated or not. Of the Danube provinces only Pannonia Inferior (Lower Pannonia) has at least 20 Christian epitaphs on its own. Britannia’s single case is a stone from Cornwall reading “son of Tribunus”, where Tribunus is more likely a personal name than the rank of tribune.

What it shows. In Christian epitaphs written in Latin, soldiers are rare almost everywhere. The largest exceptions are the Danube and neighbouring Thrace, the lower Rhine, and the north-eastern corner of Italy, with Aquileia and Concordia. On the Danube, soldiers appear in Christian epitaphs no more often than in the others: 10.2% against 21.6% by the Clauss/Slaby labels, 9.6% against 15.6% by Heidelberg’s, 5 of 59 against 20 of 116 by the text search.

Where the gap lies. The gap does not run along the whole frontier. In the two Pannonias the shares are level, a little higher on the Christian side: 5 of 43 (11.6%), against 8 of 83 (9.6%). In the two Moesias and Dacia one Christian epitaph in 16 names a soldier, against 17 of the 33 others. Nor is it spread evenly over the centuries: 5 of the 6 Christian epitaphs that name a soldier, and 24 of the 25 others, are dated between 300 and 425, when 5 of 32 Christian epitaphs and 24 of 60 others name one; this is largely the lower-Danube gap again. In epitaphs dated later or more loosely one soldier remains on each side (1 of 27 and 1 of 56).

How sure it is. The epitaphs are few, so the confidence intervals overlap. If soldiers were equally common in the two groups, a gap at least as large as that between 6 of 59 and 25 of 116 would turn up by chance about one time in 11 (Fisher’s exact test, p = 0.09). By Heidelberg’s coding p = 0.35; by the text search, p = 0.17. Below the usual threshold of 0.05 are the years 300–425 (p = 0.02), the lower Danube (p = 0.004 by the label; by the text search, 2 of 16 against 12 of 33, p = 0.10) and all inscriptions after 300 (p = 0.03). In the Pannonias the two shares cannot be told apart (p = 0.76), and 43 epitaphs cannot rule out a real difference either way. The p-values are not corrected for the several splits that were tried.

What it does not show. An epitaph says who was buried as a Christian, not who converted them; an army that helped spread the faith need not leave more soldiers in later Christian epitaphs. The test measures epitaphs of the fourth and fifth centuries. For the time before, when the army was in Dacia, the count runs the other way: of the nine epitaphs labelled Christian whose date range begins before 300 (the Jewish one left out), six name a soldier, against 1,684 of 4,943 of the others (34.1%; p = 0.07). Nine epitaphs are too few to conclude from, but it is the one period in which Christian epitaphs name soldiers more often. In every century soldiers set up stones more readily than peasants did. The other epitaphs are dated after 300 by what they say, and the ranks of the late army are among the things that date a stone, so the comparison group may hold more soldiers than it otherwise would. Nor are the labels complete. Five Christian epitaphs later than 300 lack the database’s epitaph label; two name a soldier, a centurion at Savaria (now Szombathely in Hungary) and an archer at Tomis (militans inter sagittarios iuniores, “serving among the junior archers”). With them it would be 8 of 64, or 12.5%. If all inscriptions after 300 are counted and not only epitaphs, 10 of the 118 Christian ones, seals left out, name a soldier (8.5%), against 51 of the 302 others (16.9%). And the text search finds more soldiers in the rest of the empire than the label does: 119 of 5,129, or 2.3%, where the label gives 1.8%.

The martyrs, another count. In Oxford’s Cult of Saints database, three of the 30 martyr cults in the cities on the Danube carry the label “soldier”, all tied to Durostorum. In the rest of the database 6.3% of martyr cults carry it. With numbers this small no more can be said than that soldiers are among the martyrs, as are seven members of the clergy, one of them Irenaeus, bishop of Sirmium.

The 205 inscriptions labelled Christian on the Danube, one by one#

These are all the inscriptions the database labels Christian in the five provinces, in order of date: epitaphs, building inscriptions, rings, vessels and 46 lead seals from Tomis. The text is the editors’ restored reading in normalised spelling, without the editorial marks; dots show where text is lost.

Province names are in Latin, as in the databases: Sup. and Inf. are Upper and Lower. Three epitaphs are entered twice, and a fourth, a fragment from Sirmium, probably is too; the Apahida rings appear in four records. Two other inscriptions are not Christian: Septima Maria, “Iudea”, is a Jewish woman, and “Here lies Tullia, daughter of Cicero”, from Budapest, is a modern invention. Without the three certain doubles and these two, 82 epitaphs remain (one fewer if the Sirmium fragment is also a double), and 12 of them still name a soldier. Of the 205, 14 texts are in Greek.

The Christian words Romanian inherited from Latin#

The etymologies are those of DEX, the standard Romanian dictionary, as given by dexonline. For each Latin word the table gives the number of Christian inscriptions across the empire that carry it and, separately, the number from the Danube where it can be read on the object itself and not only in the editors’ restoration.

In the dictionary’s lines Lat. is Latin; Sl., Old Church Slavonic; Mod. Gk, modern Greek; Hung., Hungarian; an asterisk marks a reconstructed form. A dash means no count was made: the Latin word is too common in ordinary use (rogare, to ask; monumentum, monument), is also a personal name (Quadragesima) or is not a word used in inscriptions. The counts are of letter-strings: under sanctus they include the name Sanctus, under repausare mostly the simple verb pausare. For preot the 1998 edition of the dictionary writes “From Lat.”, its formula for a loan; the 2009 edition has “Lat.”, its formula for an inherited word.

Nearly all are the words of an ordinary Christian: the name of God, the cross, baptism, prayer, sin, Sunday and the great feasts. Some are the same as in the other Romance languages (the words for cross, baptism, angel, Sunday). Others, according to the Romanian linguist Alexandru Niculescu, are Romanian’s own: Crăciun (Christmas), Rusalii (Pentecost), Florii (Palm Sunday), rugăciune (prayer). In biserică Romanian kept basilica, as did Romansh and the extinct Dalmatian. Martor, from martyr, kept that word’s first sense, a witness. Niculescu calls Florii and Rusalii late Byzantine in origin, and supposes that some words, among them Rusalii, Paște (Easter), creștin (Christian) and perhaps Crăciun, came in later, in the seventh and eighth centuries.

A few etymologies are disputed and cannot be used for dating. Crăciun has no agreed etymology. For Rusalii, colindă (carol) and sâmbătă (Saturday), DEX gives a Slavonic origin; other dictionaries derive them from Latin. For one word there is an earliest date: păresimi, the old name for Lent, comes from quadragesima, “the fortieth”; its Greek equivalent first appears in 325, in a canon of the Council of Nicaea, as the name of a point in the calendar, and the 40-day fast is clearly described a few years later. In this sense the word can be no older than the fourth century.

Two everyday words are often linked to the army: bătrân (old), from veteranus, and sat (village), by the usual etymology from fossatum, the ditch of a camp. Of the words of the faith only păgân might be a barracks word. One theory of how paganus came to mean a non-Christian starts from the sense “civilian”, as opposed to a soldier, another from “villager”, a third from “outsider”; in 1977 none had won general acceptance. Either way the word belongs to all of Latin Christianity. Sărbătoare (feast day) was formed in Romanian from a serba (to celebrate; from servare, “to keep”), and priveghi (wake) from a priveghea (to keep vigil; from pervigilare); the verbs have relatives across the Romance world.

The words of the Latin Church Romanian did not inherit, and what replaced them#

Some of these words are in the Christian inscriptions from the Danube. None left an inherited descendant in Romanian.

A dash in the last column means Romanian today has only late loans (cleric, lector, exorcist, neofit). The counts are of letter-strings: lector includes the reader addressed by an epitaph, benedicere people called Benedictus. For episcopus, Romanian dialects also have the form piscup, “Catholic bishop”, whose origin the dictionaries leave open; on the Danube a fragment from Preslav reads PISCOPII, with the first letter restored by the editors.

The verb a mântui (to save) comes from Hungarian. Many Greek church words (liturghie, călugăr, mănăstire, evanghelie) reached Romanian through Slavonic.

There is a tendency, with exceptions: the words that survived are mostly those of everyday belief, preot (priest) and altar among them, and those lost are the words of the hierarchy above the priest, of the school and of the book. The usual explanation is that the Latin hierarchy on the Danube broke up with the frontier; the inscriptions show only that after 600 it leaves no trace. Slavonic then brought new words for the ranks of the clergy, for the services and for part of the theology: sfânt (holy) displaced the older sânt, which survives mainly in the names of folk feasts (Sânpetru, St Peter’s day; Sântămărie, St Mary’s day; Sânziene, midsummer); duh (spirit) is used alongside the inherited suflet (soul).

Britain: 305 inscriptions labelled Christian, no clear soldier’s epitaph among them#

The database holds 9,677 Latin inscriptions from Britain; soldiers are named on 5.6% of all those without the Christian label, and Hadrian’s Wall is red on the map. Of the total, 305 are labelled Christian. A minority are Roman, such as the silver of the Water Newton Treasure and rings wishing “may you live in God”. Most are later: the inscribed stones set up in Wales and south-west England after the legions had gone, and Anglo-Saxon stones from Northumbria. None is clearly a soldier’s Christian epitaph; the label also takes in two military texts that are not Christian, a centurial stone from Llanio in Wales and a building inscription from the signal station at Ravenscar in Yorkshire.

From 597 and through the seventh century, as the English became Christian, Old English took the words of the organised Church from Latin; some may have been borrowed earlier, before the English left the Continent: bishop (from episcopus), priest (ultimately from presbyter, the word that became Romanian preot), monk (monachus), minster (monasterium), mass (missa). Romanian inherited none of episcopus, monachus, monasterium or missa, unless the dialect word piscup is one.

What Pârvan said in 1911#

In 1911 Pârvan published Contribuții epigrafice la istoria creștinismului daco-roman (“Epigraphic contributions to the history of Daco-Roman Christianity”, that is, of the Romanised people of Dacia and the lower Danube). The book sets the inscriptions beside the words. In Pârvan’s reading there were Christians in Dacia before the army withdrew, but as an exception; the faith spread in earnest in the fourth century, carried by Latin-speaking missionaries from south of the Danube, and the vocabulary, he argued, proves it.

On the army, Pârvan noted, following the church historians, that in Illyricum (the Roman Balkans) the martyrdoms of soldiers were “especially numerous”, and he believed that Diocletian’s persecution made martyrs of “countless soldiers” of the Danube garrisons. The Oxford database can neither confirm nor refute that: it counts the cults that survived, not the men who were killed. It has three cults of soldier martyrs among 30 on the Danube, a share a little above that in the rest of the database.

His evidence for the third century, the veteran’s sarcophagus from Napoca (III 866 in the Corpus Inscriptionum Latinarum, the standard collection), is still labelled Christian in both databases.

Small objects with Christian signs are less secure still. The Romanian historian Nelu Zugravu, author of a monograph on the beginnings of Romanian Christianity, wrote in 2022 that some Romanian studies postulate “far too large a number of Christian or Christianised objects earlier than Constantine”. A potsherd from Porolissum, a fort on Dacia’s northern frontier, with signs and letters long read as Christian, was judged a forgery in a 2021 study. The Biertan tablet, an offering made to keep a vow, is genuine and Christian, but it is not known whether it was made and used there or arrived as loot or by trade.

What the map cannot show#

Two databases, Latin only. LIST joins the Clauss/Slaby and Heidelberg databases as they stood when it was compiled. Both hold Latin inscriptions. The cities of the Black Sea coast wrote mostly in Greek, and their Christians, numerous at Tomis and at Histria farther up the coast, appear on the map only where they wrote in Latin; a few Greek texts, mostly seals, are the exception. Romanian took its words of the faith from Latin. Whether the Romanian people took shape north or south of the Danube, an old dispute, cannot be settled from this map.

Undated inscriptions are off the year-by-year map. Of 366,035 inscriptions with a known findspot, only 190,331 also have a date. The others are counted in a town’s result but do not appear on the year-by-year map. Dates are ranges, sometimes two centuries wide: an inscription “of 301–500” glows faintly for the whole range.

The labels are editors’ judgments. “Christian” is a label set by the databases’ editors. It covers epitaphs that say “in peace” or carry the monogram of Christ, but also the sixth-century imperial seals from Tomis; that is why the soldier test is run on epitaphs. The two databases do not label alike: of the Danube inscriptions that both databases hold, 147 are counted as Christian by at least one and 78 by both. With either, soldiers appear no more often in the Christian epitaphs after 300 (10.2% against 21.6%; with Heidelberg’s coding 9.6% against 15.6%), but with neither is the gap statistically secure. Nor do they divide the provinces alike: Emona (Ljubljana) is in Upper Pannonia in one and in Italy in the other.

Inscriptions are not a census. Those who wrote on stone had the money and the habit: soldiers, townspeople, clergy. After 300 the habit weakened everywhere.

Positions are approximate. Coordinates are those of today’s town or village rather than the exact findspot, and some are wrong: 5,583 inscriptions were placed more than 60 km outside the province the database assigns them to and were taken off the map. Three of the 205 Christian inscriptions from the Danube had wrong coordinates, one of them in Bavaria; two were put back by hand at their findspots and the third stays off the map. The Biertan tablet had no coordinates and was placed at its findspot.

Method#

The data. LIST v1.2 (Latin Inscriptions in Space and Time; Kaše, Heřmánková, Sobotková, 2024, CC BY 4.0) joins the Epigraphik-Datenbank Clauss/Slaby and the Epigraphic Database Heidelberg: 525,870 records. Makers’ stamps on bricks and other goods, and military diplomas (soldiers’ discharge certificates), were set aside: 146,136 records, which say nothing about who lived in a place. That leaves 379,734 inscriptions.

The classes. Christian: the label inscriptiones christianae. Soldier: the label milites. Epitaph: tituli sepulcrales. The Danube frontier was defined before counting as the five provinces between today’s Vienna and the Black Sea.

Checks. The same count was run on the open data of the Heidelberg database (82,450 inscriptions, CC BY-SA 4.0), with its own codes for religion and military status. Soldiers were also recognised a second way, by searching the text for ranks and units; the search and the label each find soldiers the other misses, and the search sometimes takes for ranks words that only resemble them, so it is given as a second estimate. Confidence intervals are 95% Wilson intervals; p-values are from Fisher’s exact test, two-sided. The frontier was split in two, and the years into periods, only after the first count; these splits describe the data and are not tests planned in advance.

The year on the map. An inscription dated to a range of years is shared out among the quarter-centuries the range covers, in proportion to the years that fall in each. The numbers in the legend are sums of these shares, so they are not whole numbers of inscriptions.

The words. Etymologies are those printed in DEX; where dictionaries disagree, the disagreement is noted. The Latin word was searched for in the text of inscriptions labelled Christian; for the Danube only what can be read on the object is counted.

Your town. For each place, the inscriptions found within 50 km are added up, dated or not. Places are from GeoNames.

Sources#

• Kaše, V., Heřmánková, P., Sobotková, A., LIST (Latin Inscriptions in Space and Time), v1.2, Zenodo, 2024. CC BY 4.0.
• Epigraphik-Datenbank Clauss/Slaby (M. Clauss, A. Kolb, W. A. Slaby, B. Woitas) and the Epigraphic Database Heidelberg (open data, CC BY-SA 4.0).
• Vasile Pârvan, Contribuții epigrafice la istoria creștinismului daco-roman, Bucharest, Socec, 1911 (copy in the Central University Library, Bucharest), pp. 76–78, 81–83, 139, 195–197.
• Nelu Zugravu, “Quelques observations sur l’historiographie roumaine du christianisme ponto-danubien dans les IV^e–VI^e siècles”, CICSA Journal Online, 2022.
• Alexandru Niculescu, “Originile creștinismului românesc. Greco-latinitatea”, Dacoromania, 2000–2001.
• Coriolan H. Opreanu, “Corrigenda Porolissensia (II)”, Ephemeris Napocensis 31, 2021.
• James J. O’Donnell, “Paganus”, Classical Folia 31, 1977.
• Patriarch Daniel, address on the Sunday of the Romanian Saints, June 2022, as reported by the Basilica news agency, carried by Orthodoxia News Agency (in Romanian).
• Eutropius, Breviarium 9.15; Historia Augusta, Aurelian 39; Eusebius of Caesarea, Church History 8.1 and 8.4; Lactantius, On the Deaths of the Persecutors 10; Pope Nicholas I, Responses to the Bulgarians, 866.
• The Cult of Saints in Late Antiquity, University of Oxford: saint records, consulted on 10 October 2026.
• Dicționarul explicativ al limbii române (DEX ’98), through dexonline.
• The map: relief from ETOPO 2022 (NOAA), coasts and rivers from Natural Earth, places from GeoNames (CC BY 4.0).