Skip to content

Roman religion

The vow and the contract

Roman public religion ran on conditional promises: if you do this, we will do that. The formulae mattered more than the sentiment.

Early Republic – late empire

What is attested

The Roman votum was a conditional undertaking, and its conditionality is its defining feature. The person vowing named what he wanted from a god and named what he would give if he got it; if the condition was met, the debt fell due and had to be discharged. The Latin vocabulary for it is legal: a man who had obtained what he asked for was damnatus voti, condemned in respect of his vow, in the idiom of a judgment entered against a debtor. What that idiom shows directly is that the obligation was understood as enforceable; whether it licenses describing the relationship itself as a contract is one of the open questions set out in the disputes below. This structure is not reconstructed from literature. It is legible directly on the objects, in the very large numbers of dedications across the Roman world that record a vow paid — most compactly in the formula abbreviated V.S.L.M., 'he discharged his vow, willingly, deservedly'. It is one of the largest and most uniformly distributed bodies of evidence for Roman religious action — and, as set out below, one of the most systematically skewed.

Two moments were procedurally distinct: the nuncupatio, the public pronouncement of the vow with its terms, and the solutio, the discharge. State practice made this cycle annual and archival. The Arval Brethren assembled each 3 January to pronounce vows for the ruler's welfare, and their proceedings were inscribed on stone at their sanctuary outside Rome; the surviving Acta preserve the sequence year after year, including the discharge of the previous year's vow before the new one was undertaken. A military calendar on papyrus from Dura-Europos in Syria, of the early third century AD, lists the same 3 January vows as a fixed observance for a garrison on the eastern frontier. Between them these two documents show the same date prescribed at both ends of the empire — one as performed liturgy at Rome, one as a calendar entry for a frontier garrison.

Vows scaled. A private individual might promise a small altar or a modelled body part; a magistrate might promise games; a commander might promise a temple. Central Italian sanctuaries have yielded votive deposits containing thousands of terracotta anatomical models — feet, hands, eyes, ears, internal organs, heads — deposited as discharges of vows, presumably though not demonstrably relating to healing. Roman temples were routinely built on vows made in war and paid years later out of the spoils; the dedicatory inscription naming the vower and the occasion was itself part of the discharge. Augustus's own account of his achievements records the vows undertaken for his health and the monuments voted on his safe return. The system produced buildings, and this is why so much Roman monumental architecture is legally a debt payment.

Alongside this ordinary machinery, Roman literature preserves two extraordinary applications of the same contractual logic, and here the evidence changes character completely. In devotio a commander offered himself, together with the enemy army, to the gods of the underworld, dying in the charge so that the enemy should die with him. In evocatio a besieging general invited the tutelary deity of a city to abandon it and accept a better cult at Rome. Both are described as formal procedures with dictated wording. Both are known essentially from narrative and antiquarian sources rather than from any contemporary record of the act, and the celebrated instances — Decius Mus at the Veseris, Juno Regina called out of Veii — reach us as fully composed literary scenes in Livy, writing under Augustus about events some three and four centuries earlier respectively.

One document keeps evocatio from being purely a literary construct. An inscription found in 1970 at Isaura Vetus in southern Anatolia, connected with the capture of the town by the proconsul P. Servilius Vatia in 75 BC, records a dedication addressed to whichever god or goddess had the town under protection, and states that a vow was discharged. The 'whether god or goddess' formula is a known Roman device for addressing a deity whose name or gender the speaker does not know, and its appearance here, in a first-hand commemorative inscription at a captured city, is the closest thing we have to contemporary evidence that something in the family of evocatio was actually performed in the Republic. It is a single stone, it does not use the word evocatio, and it has been read in more than one way.

How the rite worked

The vow was spoken, and the wording mattered. Roman practice held that a formula had to be pronounced correctly and without interruption, and that a magistrate undertaking a public vow was prompted phrase by phrase by a priest — Livy shows a pontifex dictating for a consul, and shows the pontifex maximus supplying the wording for a state vow, including the qualifying clauses that protected the state against technical failure. Public vows were undertaken on behalf of the Roman people by a magistrate with the Senate's authorisation, and the terms were recorded. Private vows required no priest and no authorisation; the vower spoke, and if the condition was met he paid.

Discharge was the visible half, and it is the half that survives. The physical dedication carried, at minimum, the god's name in the dative, the dedicant's name, and a statement that the vow had been paid. The costs ranged across the whole social spectrum: a mass-produced terracotta model or a small altar at the bottom, a marble altar with a cut inscription in the middle, games or a temple at the top. Because the payment was owed rather than optional, the surviving dedications skew heavily toward successful outcomes — a vow whose condition failed generated no monument. This is the most important sampling bias in the entire corpus and it should shape any statement made from it.

State vows for the ruler's welfare operated on an annual renewal cycle. The previous year's vow was discharged by sacrifice, and only then was the following year's vow pronounced, so that the state was never simultaneously in default and re-obligating itself. The Arval Acts document this sequence as a repeated liturgical event with named officiants, named victims and recorded formulae; the Feriale Duranum shows the same date fixed in a garrison's calendar. Longer-term vows attached to reigns — vows for five or ten years of rule — worked on the same principle at a longer interval and are reflected in imperial coinage.

Devotio and evocatio are described by our sources as using the same grammar at an extreme setting. In the devotio narrative the commander is robed, stands on a spear laid on the ground, is prompted through the formula by a pontifex, and then rides into the enemy line; Livy adds the provision that if the man who devoted himself did not die, an image had to be buried and an expiatory sacrifice made, which is a legal contingency clause rather than a heroic detail. In evocatio the deity is promised cult and a temple at Rome as the consideration for abandoning the besieged city, and the ritual is placed before the final assault. Macrobius, writing in the first half of the fifth century and conventionally dated to about AD 430, transmits what he presents as the actual formulae, saying he found them in an earlier compilation which had them from an older book — a chain of at least two intermediaries between him and any Republican original.

The evidence and its limits

Each class of evidence below is followed by what it cannot show. That second half is not a disclaimer; it is the part most accounts of ancient religion leave out.

  • Inscriptions

    Votive dedications across the empire, overwhelmingly recording discharge with formulae such as votum solvit libens merito, give a mass sample of the practice across five centuries, across a wide social range among those who could commission inscribed stone, and across the whole geographical reach of Roman rule.

    What it cannot show — They record only successful vows — the discharge is what generated the object — so they cannot show failure rates, cannot show what was asked, and rarely name the circumstance; the formula's very standardisation flattens out whatever variety existed in what people actually said.

  • Documents

    The Acts of the Arval Brethren, inscribed year by year at their sanctuary near Rome, preserve the annual vow-and-discharge sequence for the ruler's welfare with dates, officiants, victims and formulae.

    What it cannot show — This is one college's record of one liturgical cycle at Rome, heavily concentrated in the first and early second centuries AD; it documents an imperial state ceremony and cannot be generalised to Republican practice or to private vowing.

  • Papyri

    The Feriale Duranum, a military religious calendar from Dura-Europos of the early third century AD, fixes the 3 January vows and other observances for a unit on the eastern frontier.

    What it cannot show — A calendar lists what was to be observed, not what was observed; it gives no evidence that any given ceremony took place, and it is a single document from one garrison.

  • Archaeology

    Votive deposits at Italian sanctuaries, including large assemblages of terracotta anatomical models, show the material scale and repetitiveness of discharge at the popular level.

    What it cannot show — The objects are mute about intent. A terracotta foot is compatible with a vow for healing, thanks for healing already received, or a formulaic offering with no bodily reference at all, and the deposits are usually cleared groups without individual context.

  • Literary sources

    Livy, Cicero, Pliny, Macrobius and Servius preserve vow formulae, the doctrine of the ver sacrum, and the two extreme rituals of devotio and evocatio.

    What it cannot show — Every account of devotio and evocatio postdates the events described by centuries, and the fullest formulaic texts come from Macrobius, conventionally dated to about AD 430, at two removes from any source he names.

What the sources do not record

  • There is no surviving record of a vow that went unanswered. The votive dedications are, as a class, records of debts paid — the discharge is what produced the object — so the practice's actual success rate, and what Romans did when a god did not deliver, are structurally invisible. The Arval Acts are the exception that proves the shape of the gap: they record vows pronounced as well as vows discharged, and never a vow that failed.

  • No contemporary document records any performance of devotio. The Decii are known from annalistic narrative written centuries later, and no inscription, no dedication, and no contemporary notice attests the act.

  • The word evocatio does not appear on the Isaura Vetus stone, the one near-contemporary document usually adduced for the ritual. What survives is a dedication to an unnamed protecting deity and a statement that a vow was paid.

  • Domestic and household vowing was essentially never written down. The dedications we have are the ones that could afford durable material and public display; the ordinary daily traffic of promises made at a hearth or a wayside shrine left no record at all.

  • The archives that would settle most of this — the pontifical books, the records of state vows, the Senate's authorisations — do not survive. Livy's formulae may be quotation from such records or may be his own composition, and there is no way to test which.

What the evidence supports

Evidence level: Documented

The vow-and-discharge cycle was an archived state liturgy with a fixed annual date, recorded in enough procedural detail to reconstruct the sequence of acts.

Inscribed marble tablets of the Acts of the Arval Brethren (From the sanctuary of Dea Dia by the Via Campana outside Rome; now largely in Roman collections, published by Henzen and re-edited by Scheid).

Evidence level: Probable

Near-contemporary evidence that a Roman commander at a captured city addressed its unnamed tutelary deity in vow language — the strongest documentary support for something in the evocatio family, though the term itself does not appear.

Latin inscription from Isaura Vetus recording a dedication to whichever god or goddess protected the town, and the discharge of a vow, in connection with its capture by P. Servilius Vatia in 75 BC (Found north of Bozkır, Konya province, Turkey, in 1970 by Alan Hall; published in the early 1970s).

Evidence level: Probable

A conditional vow followed by a recorded payment was a standard mode of individual religious action, uniform in structure across the empire.

Votive altars and plaques bearing the discharge formula, abbreviated V.S.L.M. and variants (Distributed across the Roman world from Britain to Syria and North Africa; thousands of examples in the epigraphic corpora). The corpus is by definition restricted to those who could commission inscribed stone, so it shows the range of that group rather than of the population.

Evidence level: Probable

The volume and standardisation of vow-discharge at the popular level.

Votive deposits of terracotta anatomical models — feet, hands, eyes, ears, internal organs, heads (Central Italian sanctuaries, including the large deposit at Ponte di Nona east of Rome; principally fourth to first centuries BC). That models of this kind were produced for the purpose is an inference from their mould-made repetition rather than something independently documented.

Evidence level: Probable

Annual vows for the ruler on 3 January were prescribed in the religious calendar of one unit on the eastern frontier — the only direct evidence that the Roman cycle was written into a garrison's year away from Rome, and a single document, not a demonstration of empire-wide uniformity.

Feriale Duranum, a papyrus military religious calendar (Dura-Europos, Syria; Yale collections; early third century AD).

Stories told about the rite

Ancient writers offered these as explaining the practice. They are evidence for what the tradition said about itself, and they are not the origin of anything: each is given with the author who tells it and the distance between the two.

  • Evidence level: Mythological

    Livy, book 8, chapter 9, writing under Augustus about an event placed in 340 BC — a gap of roughly three centuries; the repetition across three generations is itself a mark of family tradition rather than record.

    Publius Decius Mus, consul, faced with a collapsing line at the battle near the Veseris, has a pontifex dictate a formula, devotes himself and the enemy army to the Manes and to Earth, and rides into the enemy to die; the enemy line breaks. The scene is then repeated by his son, and by tradition his grandson.

    Offered as explaining — How a Roman defeat could be converted into victory by ritual means, and why the Decian family carried exceptional prestige. It supplies a mechanism for the otherwise unaccountable.

    This is an aition for a ritual category, not a report of a battlefield event. Its narrative details — the dictated wording, the spear underfoot, the contingency rule for a devotus who survives — document what Romans of Livy's generation understood the procedure to be. They are evidence for Augustan-era religious understanding, not for 340 BC.

  • Evidence level: Mythological

    Livy, book 5, chapters 21-22, writing about the traditional date of 396 BC — a gap of nearly four centuries. Livy himself flags the miraculous element as the kind of thing one may report without vouching for.

    At the siege of Veii, Camillus calls Juno Regina out of the city, promising her a temple at Rome; when Roman soldiers ask the statue whether she wishes to come, she is said to have nodded, or to have spoken. She is carried to the Aventine and a temple is dedicated to her.

    Offered as explaining — Why a foreign goddess had a major Roman cult on the Aventine, and how Rome could absorb the gods of the peoples it destroyed without impiety.

    The temple on the Aventine was real and its foundation needed an explanation; the story supplies one. The nodding statue is a narrative motif, and Livy's own distancing shows he knew it as such. The passage evidences a Roman theory of how conquest and cult fit together.

  • Evidence level: Mythological

    Livy, book 22, around chapters 9-10, on events of 217 BC — again some two centuries after the fact.

    In the crisis after Trasimene, on the direction of the Sibylline books, the Romans vow a ver sacrum — the sacrifice of all livestock born in a coming spring — with the pontifex maximus dictating the wording and building in clauses to cover irregularities in performance.

    Offered as explaining — Why an archaic-sounding institution could be revived at a moment of extreme danger, and how a vow of enormous scale could be undertaken without exposing the state to ruinous technical default.

    Whether the formula Livy gives is quoted from a record or composed by him cannot be determined. The qualifying clauses are the interesting part: they show that the tradition understood vows as legally exacting instruments, which is a fact about Roman religious thinking whatever the passage's documentary status.

Vocabulary

votum
A conditional promise to a god: if the request is granted, the named payment falls due. The word covers both the promise and the thing promised.
nuncupatio
The formal pronouncement of a vow, with its terms stated; for public vows, made by a magistrate with a priest dictating the wording.
solutio / votum solvit
The discharge of the vow once the condition was met. The abbreviation V.S.L.M. — votum solvit libens merito — is the commonest formula in Latin votive epigraphy.
damnatus voti
Literally 'condemned in respect of the vow': the state of a person whose request has been granted and who therefore owes. The idiom is taken directly from the language of legal judgment.
devotio
The offering of a person — in the famous cases a commander himself, together with the enemy army — to the gods of the underworld. Known essentially from narrative sources; no contemporary attestation survives.
evocatio
The calling-out of a besieged city's tutelary deity, offered better cult at Rome in exchange for abandoning its people. The term is Roman antiquarian usage; it does not appear on the one near-contemporary inscription associated with the practice.
ver sacrum
A 'sacred spring': a vow of everything born in a given spring, undertaken in extreme crisis, with elaborate qualifying clauses to limit the state's exposure to technical default.
sive deus sive dea
'Whether god or goddess': a Roman formula for addressing a deity whose name or gender the speaker does not know, so that the address cannot fail for misidentification.

Contested points

Evidence level: Disputed

Was evocatio a regular, repeatable instrument of Roman siege warfare, or a handful of exceptional acts generalised into a rule by later antiquarians?

The maximalist reading points to Macrobius's transmission of a set formula, to the recurring pattern of foreign deities receiving Roman cult after conquest, and to the Isaura Vetus inscription as showing the practice alive in 75 BC, concluding that this was a standing procedure. The minimalist reading observes that the securely named cases are very few, that the fullest formula comes from an author conventionally dated to about AD 430, reaching him through at least two intermediaries, that Isaura Vetus does not use the word, and that Roman antiquarians had strong incentives to systematise scattered precedents into a doctrine. The disagreement is really about whether Macrobius preserves Republican liturgy or Republican-sounding reconstruction.

Evidence level: Disputed

Did the devotio of a commander ever actually happen?

The sceptical case notes that the act is reported for a father, a son, and by some accounts a grandson of the same family — a pattern typical of gentilicial legend duplicated by annalists — that no contemporary source attests any instance, and that Livy's version is a masterpiece of composition with a hero's speech at its centre. The defensive case argues that the Romans plainly had a ritual category of devotio, that the surviving contingency rules for a devotus who survives are too technical and too unheroic to be invented for effect, and that at least the earliest instance may preserve a real event however heavily dressed. Both sides agree that Livy's text documents Augustan understanding of the mechanism; they differ on whether anything is recoverable behind it.

Evidence level: Disputed

Is 'contract' the right model for the Roman vow?

The transactional reading takes the Latin at its word: the vocabulary is legal (damnatus voti, solvere), the structure is conditional, and Roman religion is best understood as a system of reciprocal obligation between citizens and gods, often summarised as do ut des. The alternative, developed by John Scheid and others working from the sacrificial evidence, holds that this makes the relationship too mercantile and misses that vow and sacrifice constituted and displayed a hierarchical relationship in which the god's superiority was affirmed rather than bargained away, and that Roman legal metaphor should not be read as Roman religious theory.

Where this comes from

  • Livy, Ab urbe condita 8.9

    Describes the consul Decius, with his line giving way, calling for a pontifex to dictate a formula, pronouncing it in prescribed dress and posture, devoting himself and the enemy army to the gods below, and riding to his death; the following passage sets out rules for cases where a devotus survives.

  • Livy, Ab urbe condita 5.21-22

    Narrates the fall of Veii, the invitation to Juno Regina to leave the city for a better home at Rome, the reported assent of the image, and the transfer of the cult to the Aventine.

  • Macrobius, Saturnalia 3.9 (the evocatio formula at approximately 3.9.7-8; the devotio formula following)

    Transmits what it presents as the formula by which a Roman commander summoned out the protecting deities of Carthage, and a further formula for devoting a city and its army to destruction, with a note on the sources through which the wording reached the author.

  • Livy, Ab urbe condita 22.9-10

    Gives the vow of a sacred spring undertaken after Trasimene, with the wording dictated by the pontifex maximus and explicit clauses covering defective performance, unintentional error, and the ownership of the animals concerned.

  • Augustus, Res Gestae Divi Augusti 9, 11-12 (vows for the emperor's health; monuments voted on his safe return)

    Records that the senate decreed vows for his health to be undertaken every fifth year by the consuls and the priests, and that games were repeatedly held in fulfilment of them by the chief priestly colleges or by the consuls; and that on his safe return the senate consecrated an altar of Fortuna Redux and voted further honours — an official first-person statement of the vow cycle at the top of the state.

  • The Arval Brethren (collective priestly record), Acta Fratrum Arvalium the annual entries for 3 January; individual years cited by consular date

    Records the college assembling to discharge the previous year's vows for the ruling house and to pronounce new ones, naming the presiding brother, the beneficiaries and the sacrifices.

Covered elsewhere

How to read the evidence labels

Evidence level: Documented
Supported by a named primary text or by excavated material evidence.
Evidence level: Probable
Supported by strong inference that most specialists accept.
Evidence level: Disputed
Substantial specialist disagreement exists; competing readings are stated.
Evidence level: Literary
Belongs to poetic or mythological construction rather than to history.
Evidence level: Mythological
Belongs to shared religious or mythic tradition rather than to any single text.
Evidence level: Unknown
The available evidence is insufficient to decide.

These labels describe the status of the evidence, not our confidence in a conclusion. A claim marked Literary is not a claim we doubt; it is a claim about what kind of thing a poem is.