When a Roman officer met a god he had never heard of, he did not usually record the god's name. He recorded which of his own gods it resembled. That reflex — the ancient world's default response to somebody else's religion — is called interpretatio, and it is simultaneously the reason we have any evidence at all for several vanished pantheons and the reason that evidence is close to useless for reconstructing them.
It is worth understanding as a technology rather than a mistake. It was efficient, it was usually well-meant, and everybody did it. It was also lossy in a very specific direction: it preserved what a god did and erased who a god was.
Caesar's list and the names it swallowed
The most consequential single paragraph on Gaulish religion was written by a general in the middle of conquering Gaul. Caesar reports that the Gauls venerate above all a god he calls Mercury, inventor of crafts and guide of travellers, followed by ones he calls Apollo, Mars, Jupiter and Minerva, and he assigns each a neat portfolio: healing, war, sky, handicrafts.
He does not give a single Gaulish name. Not one.
The damage compounds. That passage was the standard reference for centuries, so later readers concluded that the Gauls had five main gods with Roman characters. Inscriptions and archaeology tell a very different story: hundreds of divine names across Gaul, most attested only once or twice, many local to a spring or a mountain. Caesar's "Mercury" is not one god. It is a category he built by matching function, and it may cover a dozen unrelated deities whose only shared feature is that a Roman found them useful to describe the same way. The Celtic pages here treat Continental and insular evidence as separate bodies for exactly this reason — an Irish text written a millennium later cannot repopulate Caesar's blanks.
What an equation actually claims
An ancient equation between gods is not a statement about mythology. It is a statement about cult: this is the one you sacrifice to for that purpose. Two gods can be equated while having incompatible parents, incompatible stories and incompatible personalities, because none of that was the criterion.
Herodotus, touring Egypt in the fifth century BCE, produces a whole table of these: the Egyptian craftsman-creator becomes Hephaestus, Thoth becomes Hermes, the hidden god of Thebes becomes Zeus, Isis becomes Demeter, and the disorder god becomes the monster Typhon. Some of these are genuinely illuminating — Thoth and Hermes really do share the writing, the messages and the escorting of the dead. Others are close to arbitrary. And the Typhon identification did lasting harm, because it recruited Set into a role as cosmic enemy that Egyptian religion had only partly assigned him, at a period when his reputation was already collapsing for internal Egyptian reasons.
He does the same in the Black Sea, giving a list of Scythian deities in Greek dress with the native names attached almost as an afterthought — which, by the standards of his century, counts as unusually good practice.
The stones that kept both names
The honest record is epigraphic, and it survives because provincials wrote their own dedications. A soldier or a landowner paying for an altar frequently kept both names, producing double-barrelled gods all over the western empire.
At a hot spring in Britain, the local goddess and the Roman one appear fused in a single dedication, and the resulting figure has the Roman name attached to a sanctuary whose curse tablets are written in a wholly un-Roman idiom of personal grievance. In Gaul and the Rhineland the war god appears under a string of local surnames, each tied to a place. These compound names are the best evidence we have that equation was not replacement. Two systems ran in parallel, in the same person's head, and the local layer was frequently the one that mattered when something actually needed doing.
Gods built to specification
Some equations were engineered. When the Ptolemies took Egypt, they had a governance problem: a Greek-speaking court ruling an Egyptian population with a formidable priesthood. The response was Serapis, promoted as a state cult under the first Ptolemy — an Egyptian underworld god of Memphis fused with elements of the Apis bull cult and given entirely Greek looks, a bearded enthroned man with a grain measure on his head. It worked. Serapis spread across the Mediterranean and outlived the dynasty that commissioned him.
The same era produced Hermanubis, Anubis with a herald's staff, and — over the following centuries — Hermes Trismegistus, a composite sage grown from the Greek reading of Thoth whose supposed writings drove alchemy and Renaissance occultism for a thousand years. A god can be a policy instrument, and a mistranslation can found a philosophy.
The habit is in your calendar
English speakers use interpretatio every week without noticing. The Roman planetary week named days for Mars, Mercury, Jupiter and Venus. When it spread north, Germanic speakers did not borrow the Roman names — they translated them, substituting their own nearest gods. Mars's day went to Tyr, Mercury's to Odin, Jupiter's to Thor, Venus's to Frigg.
Look at what that calque asserts. Mercury, the trickster-messenger, was matched to Odin, and the war god was matched to Tyr rather than to Odin — which tells us that whoever made the equation understood Odin primarily as the cunning traveller between worlds and Tyr as the god of the sworn contest. That is a genuine piece of evidence about Germanic religion in the Roman period, preserved by accident inside the word Wednesday. Roman writers describing Germanic peoples made the same match independently, which strengthens it.
Cultures do it to themselves as well
Interpretatio is not only what conquerors do to the conquered. Japan spent centuries working out a formal scheme in which local kami were understood as manifestations of Buddhist figures, with specific pairings argued over by specific temples — an internal, theologically elaborated version of the same operation. Continental deities arriving with Buddhist scripture were absorbed and rebuilt, so that an Indian river goddess of speech and music became a Japanese goddess associated with islands, water and fortune, with a different iconography and a different clientele.
The direction of borrowing there is documented, which makes it a useful control case. It shows what a slow, deliberate, well-recorded equation looks like — and by contrast, how little Caesar's one-line matches can bear.
Where the equations simply fail
The best-known equation set in the world is the Greek–Roman one, and it is the most misleading, because it is genuinely close in many pairs and genuinely false in others. Athena and Minerva converge respectably. Ares and Mars do not: one is a figure of literary contempt, the other a founding father of the Roman state with a month, a field and a priesthood, and treating them as one god makes Roman religion incomprehensible. Hera and Juno have separate cults, separate festivals and separate temperaments underneath the shared story cycle.
The practical rule when you meet any claim that god A "is" god B: ask who made the equation, when, and what they needed it for. An equation made by a conqueror describing an enemy, a merchant looking for a shrine, a court legitimising a dynasty, and a modern writer looking for a headline are four different kinds of claim, and only the middle two were ever meant to be taken as religion.
Questions this raises about the evidence
Is interpretatio Romana the ancient term or a modern one?
It comes from Tacitus, who uses the phrase in passing while describing a pair of Germanic divine brothers — noting the Roman gods they would be identified with, and then declining to treat that identification as their real nature. The phrase became a technical term in modern scholarship, which is a slightly ironic afterlife for a remark that was making a sceptical point about equations in the first place.
Did anyone in antiquity object to the practice?
Some did, on both sides. There were Greek and Roman writers who insisted that a foreign god's own rites and names mattered and could not be swapped, and Egyptian priestly tradition maintained its own naming with great tenacity through centuries of Greek rule. What is largely absent is the modern discomfort about it, because the underlying assumption — that gods are real, universal and available to everyone under various names — was not controversial.
Does comparative scholarship do the same thing today?
It risks it, and the discipline knows it. Reconstructing an ancestral Indo-European pantheon from correspondences between Vedic, Greek, Roman and Norse material is a serious linguistic enterprise with real results, but the framework has been criticised for sorting gods into a small number of predetermined social functions and then reading the evidence to fit. The safeguard is the same one the ancients lacked: state which language, which text, which century, and keep the local name.
Why does this site refuse to call a figure "the X of Y"?
Because that phrasing smuggles in an equation and then hides the argument for it. It is faster to write "the Norse Hermes" than to explain what Odin and Hermes share and where they part, and the shortcut has to be paid for later by the reader, who now believes something nobody has demonstrated. The comparison section exists to make those arguments explicitly rather than assume them in an epithet.