Iran’s grandest “temple of Anahita” rests its entire claim on one line a Greek road-surveyor wrote two thousand years ago. The stones are certain — a cut-limestone platform in the same family as Persepolis’s — but the goddess is a hypothesis. Archaeologists have called it her sanctuary, an unfinished palace of Khosrow II, and everything in between. The name, unproven and unkillable, has outlasted them all.
“Until detailed further excavations are carried out, no definite judgments may be declared on the function of Kangavar platform.”
Encyclopædia Iranica, “Kangavar” — the field’s official shrug, after half a century of digging
In the middle of Kangavar — an ordinary bazaar town on the Hamadan–Kermanshah road — a hill of dressed stone rises out of the shops and traffic. Great blocks of grey-white limestone, drums of fallen columns, and two monumental staircases climb a terrace measuring roughly 224 by 209 metres: about four and a half hectares of cut stone, wrapped around a natural rock that stands up to 32 metres above the plain. Persian references like to call it the second-greatest stone building of ancient Iran after Persepolis — a popular claim we report rather than certify, but standing under its southern wall, twenty metres of masonry overhead, you will not feel like arguing.
What you are looking at, everyone agrees, is enormous, ancient, and important. What it is, nobody has ever managed to prove. For most of the last century it has been Iran’s most famous temple of Anahita — the great goddess of the waters — and that identification hangs, almost entirely, on a single clause written around the turn of the era by a Greek surveyor of the Parthian road system. Isidore of Charax, pacing off the empire’s caravan route for Rome, passed a city he called Concobar and noted, in six words, a shrine:
That is the deed of ownership: one line, in translation, in someone else’s language, for someone else’s goddess. Everything since has been argument. Medieval Persian and Arab writers, seeing the ruin from the same road, told a different story — an unfinished palace of Khosrow Parviz, the last great Sasanian king; travellers’ accounts quoted in Persian references knew the place as Qasr al-Lusus, the “Castle of the Robbers.” Modern archaeology dug for decades and split down the middle: the man who excavated it longest died convinced it was the goddess’s sanctuary; the man who dug after him published that it was never a temple at all. The stones have worn at least four names in two thousand years. The one that stuck is hers.
That is the real subject of this hill. Not water, not kings — naming: what it takes for a name to attach to stone, and how little proof it needs to hold on. The platform is a fact. The goddess is a claim. And the claim has outlasted every empire that could have settled it.
Kangavar exists because of its road, and so does the confusion. The town sits astride the great east–west artery of the Iranian plateau — the route Persian references list under four names for four ages: the Royal Road of the Achaemenids, the Silk Road of the Parthians, the Sasanian highway between east and west, and the Great Khorasan Road of medieval Islam. Isidore was measuring exactly this line when he logged his temple of Artemis; the same paragraph of his road-book runs on from Concobar to Ecbatana itself, twelve schoeni up the road, where — note the wording — he records “a temple, sacred to Anaitis; they sacrifice there always.” When Isidore meant Anahita by name, he could say so. At Kangavar, he wrote Artemis. Even the founding document of the identification is a translation.
Most monuments accumulate damage. This one accumulates identities. What follows is less a construction history than a custody battle — the same stones, claimed and reclaimed by every age that passed on the road below.
To understand why the name refuses to die, you have to understand who it belongs to. Aredvi Sura Anahita — the epithets are usually glossed as something like the mighty, the immaculate, of the waters — is the great river goddess of ancient Iran: guardian of all flowing water, of fertility, of the increase of herds and lands. In the Avesta she is a world-river in person, plunging from the mythical peak of Hukairya down to the cosmic sea, “life-increasing, herd-increasing… who makes prosperity for all countries.”
Her hymn is the Aban Yasht — Yasht 5 of the Avesta, and at 131 verses the third-longest hymn that survives. Most of its middle is a roll-call of heroes and kings, each sacrificing to her to win some enormous boon. And it carries one liturgical rule that lands differently once you have stood at Kangavar: by tradition, the Aban Yasht is never recited before a fire, only within sight of water. If a place wanted to claim this goddess, it would need water in view. Kangavar’s plain has it — the Sarab-e Kangavar stream runs west of the town toward the Gamasiab, and springs stitch the district (Sarab-e Fash, a spring-fed village, lies 15 km north). The excavation-era case for the temple leaned partly on exactly this geography of water. It proves nothing. It merely keeps the claim breathing.
Anahita travelled well — better than almost any Iranian divinity. From the reign of Artaxerxes II (early 4th century BCE), who promoted her cult across the empire, her statues and shrines spread from Mesopotamia to Asia Minor, where her worship long outlived the Achaemenids. Greeks meeting her abroad translated her, as Greeks did: sometimes Aphrodite, often Artemis, sometimes simply Anaitis, a Grecised form of her own name.
Which is precisely the problem with the Kangavar deed. In the very same paragraph of the Parthian Stations, Isidore uses both options within twelve schoeni of road: a “temple of Artemis” at Concobar, then at Ecbatana “a temple, sacred to Anaitis; they sacrifice there always.” Scholars reasonably read the Artemis of Concobar as Anahita in Greek dress — but the founding sentence of the world’s most famous Anahita temple does not actually contain her name. It contains a translation of it. The goddess arrived at Kangavar the way she arrived everywhere: renamed.
And she did not leave the region. Ninety-odd kilometres down this same road, in the great grotto at Taq-e Bostan, she stands carved in stone at the investiture of Khosrow II — the very king whose unfinished palace this platform may be. Whichever side of the dispute is right, the goddess and the king meet on the road either way.
The modern argument is really the story of two excavators, working the same stones a decade apart and reading them in opposite directions.
Seifollah Kambakhsh-Fard, arriving in 1968, dug like a man confirming a text. The colonnade his first trenches uncovered — those twelve columns in sixty metres — looked to him like the peristyle of a great open sanctuary; the platform’s water-rich setting suited the goddess; Isidore supplied the name. Across seven seasons he cleared the mound, restored what he could, and published the site as the Anahita temple, a complex with roots he traced back to the Achaemenids and use running through Parthian into Sasanian times. He gave the rest of his career to the reading; the research library at the site today bears his name.
Massoud Azarnoush, digging in 1977–78, looked at the same masonry and saw an absence: no cella, no altar layout, none of what he considered “the necessary characteristics that could identify it as a temple” (as the Encyclopædia Iranica summarises his case). His counter-reading — a late Sasanian palace, unfinished, quite possibly the Khosrow Parviz project the medieval geographers had described a millennium earlier — was heresy against a beloved name, and he spent decades defending it, capping the argument in 2009 with that thermoluminescence-dated brick from beneath the western platform. Ali Akbar Sarfaraz of Tehran University sided with him.
The temple camp never conceded. Arthur Upham Pope had already canonised the site as a “columnar temple dedicated to Anahit”; Karim Pirnia classed it as Parthian work renovated under the Sasanians; and Warwick Ball calls it “one of the greatest works of Parthian architecture,” an eastern cousin of the Roman temple form built around a vast sacred enclosure — while conceding, in the honest middle, that the site probably saw numerous major phases: perhaps a 2nd-century CE colonnaded enclosure with a heavy Sasanian rebuilding inside it. Which would mean, deliciously, that both excavators were holding a real piece of the truth — a Parthian-era sanctuary and a Sasanian construction site, stacked.
Where does that leave the visitor? Exactly where the Encyclopædia Iranica leaves the field: no definite judgement is yet possible. Every camp has stones on its side; none has an inscription. The building is the rarest thing in a country of confident monuments — a colossus that declines to testify.
Strip away the argument and a hard core of certainty remains — the things the stone will admit regardless of whose name is on it. Five of them carry the visit.
The monument is really a hill in fancy dress: a natural rock outcrop, up to 32 m high, cased and extended in tiers of masonry — the classic Iranian soffeh (terrace) technique — until it reads as one man-made acropolis of about 4.6 hectares. The southern retaining wall stands some 20 m sheer. Walk the full circuit at street level first; the scale is the argument.
Two monumental lateral stairways climb the platform — the detail that convinced Pope, Ball, and others that whoever built this was quoting the Apadana at Persepolis. The Iranica calls the ramp idea “doubtless an Iranian inheritance” from the Achaemenid manner. Whatever the building was, its builders wanted you to ascend it like a Persian king.
The colonnade’s thick, close-set columns with plain square-topped capitals, riding a cornice cut with a Greco-Roman cyma reversa profile, are the site’s second argument — the “western” accents that made 19th-century Europeans see a classical temple, and that scholars now read as provincial echoes rather than pedigree. Rows of re-seated drums, some numbered by the restorers, line the upper terrace like a parts inventory for a building nobody can reassemble.
Geochemical fingerprinting (a 2020 peer-reviewed study) has matched the monument’s limestone to nearby quarries; Persian references name Chehel Maran, about two kilometres west, where unfinished column drums and cut blocks still lie on the hillside — a Bronze-Age-style snapshot of a build stopped mid-order. If Azarnoush is right about an unfinished palace, this is where the music stopped.
The monument does not sit in a desert of rope-lines; it sits in a working town that has always lived off the road — the Emamzadeh Ebrahim shrine stands 150 m away, the bazaar closer still. Until the excavators bought them out in the 1970s, the mud-brick houses of the Gach-kon quarter stood on the platform itself. People have never stopped using this hill; they have only kept changing what they called it.
Honesty, as ever on Untamed Iran: this is a knowledge-wonder, not a spectacle. There is no standing temple, no roofline, no carved goddess. What survives is a platform, staircases, wall-lines, and drums — plus the best unanswered question in Iranian archaeology, which you must bring the reading (or a good guide) to see. Come expecting Persepolis and you will leave in twenty minutes; come expecting an argument two thousand years old, laid out in twenty-tonne evidence, and the hill opens up.
Untamed Iran rates each destination on two dimensions — Adventure, the physical demands a place makes on you, and Legacy, the weight it carries in history, atmosphere, and culture. Kangavar is the collection’s purest knowledge-wonder: physically one of the easiest sites we cover — it is literally in a town centre on a main highway — and intellectually one of the heaviest, a colossal stone question mark wearing a goddess’s name. The wildness here is entirely in the argument.
Go up the southern staircase in the last hour of light, when the limestone goes honey-coloured and the town below starts its evening noise — motorbikes, the call to prayer from the Emamzadeh a hundred and fifty metres away, someone shutting a shop grille. From the top terrace the Hamadan road runs away east and west exactly where Schmidt’s 1930s aerial photographs show it, exactly where Isidore paced it with his schoenus-rope. You are standing on the answer to a question nobody can finish asking.
And here is what lands: everyone around you has already decided. Ask at the gate and it is ma’bad — the temple. The street signs say temple. The ticket says temple. The national register has said temple since 1931. Two thousand years of medieval kings’ names and robbers’ castles and unfinished palaces, one devastating excavation report, a thermoluminescence date — and the town has simply carried on calling the hill by the name of a goddess whose presence here rests on six translated words. Not out of ignorance; the dispute is printed on the site’s own information boards. Out of something more stubborn: a name, once it has held for twenty centuries, becomes part of the monument.
You came to see a building and you leave having watched a claim survive. The empires that could have settled it are gone — the Parthians who may have raised the colonnade, the Sasanian king who may have died before finishing it, the scholars still filing papers against each other. The platform stands. The goddess was never proven. And somehow, walking back down into the bazaar as the lights come on, you find you have taken a side too. Almost everyone does. That is her real temple: not the stone — the reflex.
A stone platform in Persepolis’s league, wearing a goddess’s name it cannot prove — deeded by six translated words in a Greek road-book, contested by kings, robbers, and archaeologists for two thousand years. Every empire that touched this hill is gone. The attribution — unproven, unkillable — is the one thing still standing.
The prime window. The Kangavar plain sits at ~1,500 m in the Zagros, so spring means mild days, green wheat around the town, snow still on the high ridges, and the Sarab streams running full — the goddess’s season, if she ever had one here. Nowruz fortnight (late March) is busy with Iranian families; early May is the sweet spot.
The equal-best alternative: dry, golden, and calm, with harvest light raking the limestone in late afternoon. Persian travel sources name spring and autumn as the season, and they are right — this is when the platform photographs like an acropolis instead of a quarry.
The workable shoulders. March can still bite (cold snaps, mud); June is hot by midday but fine early and late — the altitude takes the edge off; November turns cold and grey but the site is empty. In all three, aim for the first or last hours of opening.
Hard season. Kangavar’s climate is cold semi-arid: freezing nights, snow, and — the real issue — ice on worn stone stairs. The same freeze–thaw cycles that are slowly splitting the monument’s limestone will not do your ankles any favours either. Visit only with good boots and low expectations of comfort.
⏰ Time of day is half the visit. The colonnade and the great southern wall need raking light — first hour after opening or last hour before closing; midday flattens the stone into grey. Note the site closes on major religious mourning days, and hours shorten in the second half of the Iranian year — confirm locally.
The visit itself ends above. What follows is the planning detail — gear, logistics, and questions — tucked away so you can open only what you need.
Kangavar may be the easiest great site in this collection to reach: it sits on the main Hamadan–Kermanshah highway, mid-route, with the monument effectively in the town centre. The planning is mostly about which base city to use and what to loop in along the old road. Prices move with the rial, so treat figures as orders of magnitude.
In the centre of Kangavar, easternmost Kermanshah Province, right on the Hamadan–Kermanshah highway — ~90 km from Hamadan, ~94–95 km from Kermanshah, ~405 km from Tehran via Saveh. Any bus or shared taxi on that road passes it; the platform is minutes from the highway, beside the bazaar on Shohada Street.
Unproven. The identification rests chiefly on Isidore of Charax’s 1st-century note of a “temple of Artemis” at Concobar (Artemis being the usual Greek rendering of Anahita). Pope, Pirnia, and Ball read the site as a columned sanctuary; excavator Massoud Azarnoush argued (1981) it lacked a temple’s characteristics and was an unfinished late Sasanian palace, plausibly Khosrow II’s. The Encyclopædia Iranica’s verdict: no definite judgement is yet possible.
Aredvi Sura Anahita — roughly “the mighty, immaculate one of the waters” — the great river goddess of ancient Iran: waters, fertility, abundance. Her hymn, the Aban Yasht (Yasht 5, 131 verses), is by tradition recited only within sight of water. Promoted empire-wide from Artaxerxes II onward, she was still crowning kings in stone at Taq-e Bostan under Khosrow II.
Depends on the reading. Kambakhsh-Fard attributed the complex to Achaemenid, Parthian, and Sasanian phases; stylistic arguments long favoured a Parthian date (a ~200 BCE construction was once proposed); Azarnoush’s evidence — including a thermoluminescence-dated brick from under the western platform — points late Sasanian, around the turn of the 7th century CE. Finds span Achaemenid to early Islamic layers either way.
About 224 × 209 m (~4.6 ha) of terraced masonry over a natural rock rising up to ~32 m, with the southern face standing some 20 m. Often described as Iran’s largest ancient stone structure after Persepolis — a popular claim we report rather than certify.
Ticketed, open daily except major religious mourning days. Persian sources give roughly 9–19 in spring–summer and 9–17 in autumn–winter; both hours and prices change, so confirm locally. Foreigners pay a higher tariff, in cash.
Yes — for the argument. No standing temple survives; what you get is a colossal platform, twin monumental staircases, column drums, and the best unresolved question in Iranian archaeology. It pairs perfectly with the same-road giants: Bisotun (~57–60 km) and Taq-e Bostan toward Kermanshah, Hegmataneh toward Hamadan.
Kangavar makes no sense alone; it is a way-station that grew a monument, and its true companions are the other giants strung along the same east–west artery — Isidore’s own itinerary, drivable in a day. Start where his paragraph ends: at Hegmataneh (Ecbatana) under modern Hamadan, the Median capital where he logged the shrine of Anaitis by her proper name, ~90 km north-east. Then this platform, his “temple of Artemis.” Then south-west down the highway to Bisotun (~57–60 km), where Darius cut his words into the cliff Isidore knew as the mountain of Semiramis — the site his road-book lists one stop before Concobar. End in the grotto at Taq-e Bostan on Kermanshah’s edge, where Anahita herself stands carved at the investiture of Khosrow II — the goddess this platform may honour, crowning the king whose unfinished palace it may be. Do it as one journey, east to west, and the dispute resolves into something better than an answer: one road, four names, and every claimant to this hill standing along it in stone. (For the goddess’s other contested house, far to the south, see Bishapur.)
Untamed Iran prefers official, scholarly, and first-hand sources, and separates what is established from what is argued. The dimensions, the excavation history, and both sides of the dispute draw on the following:
Facts last reviewed July 2026. Established: the location, the ~224 × 209 m platform and its figures, the 1931 national listing, the excavation chronology, and the existence of Isidore’s line. Attributed: every identification — the Anahita/temple readings (Pope, Pirnia, Ball, Kambakhsh-Fard, with Herzfeld among earlier proponents) and the late-Sasanian-palace reading (Azarnoush, supported by Sarfaraz) are presented as their proponents’ claims, not facts; the equation Artemis = Anahita at Concobar is itself an interpretation, and the Iranica notes even the ruins’ identity with Isidore’s temple is questionable. Reported: opening hours, the “second stone building after Persepolis” popular claim, the 2022 site-museum plan, and the 2024 “gradual death” conservation warning. Approximate: coordinates, road distances, elevation, and the ~4.6 ha figure (sources vary slightly on the platform’s exact envelope). Confirm hours and prices locally.