Karahunj · Armenia · 39.5508, 46.0288 · PYR record in the living atlas
On a plateau in southern Armenia, at 1,770 metres, roughly 223 stones of basalt and andesite stand, lean and lie across about seven hectares. They run from half a metre to three metres above ground and the largest weigh around ten tonnes. About 80 of them are pierced: smooth, carefully finished holes near the upper edge, roughly five centimetres across and as much as half a metre deep, drilled clean through the rock. Published measurements of the holes vary between sources, and no agreed dimensional survey of them has ever been published. Only 37 of the holed stones are still upright, and between them they carry 47 holes, a count that circulates through Wikipedia rather than through any field report; Onnik Khnkikyan's survey in 1984 counted 223 stones with 84 of them holed. The spread of those figures is a fair indication of how much of this site has moved since it was built. Whoever made those holes meant something by them, and the argument about what has made a quiet Armenian hillside internationally famous.
The observatory claim, in its strongest form
The idea did not start with a showman. The astronomical reading entered the scholarly record in 1984, when the Armenian archaeologist Onnik Khnkikyan examined the site, counted the stones and their holes, and suggested it may have functioned as an observatory. The astrophysicist Elma Parsamian of the Byurakan Observatory began her own investigation shortly afterwards, measured the holed stones against an astronomical calendar, and reported in a 1985 paper, "Carahunge — an ancient observatory in Armenia" in Astronomicheskii Tsirkulyar, that several aligned with summer solstice sunrise and sunset. That was a modest, specific observation, and the first alignment study here with a described method behind it. She also proposed the name Karahunj. The claim that travelled, though, belongs to the radiophysicist Paris Herouni, who surveyed between 1994 and 2001 and concluded that this was the oldest astronomical observatory in the world. In his reading, 17 stones related to solstice and equinox positions and 14 to the extremes of the moon, and the shift of star positions over time dated the arrangement to around 5500 BC, thousands of years before Stonehenge. He set the case out in Armenians and Old Armenia (Tigran Mets, Yerevan, 2004), where the site is a temple, an observatory and a university at once, dedicated to a sun-god named Ari. He also read the name as speaking stones, from the Armenian words for stone and for sound, on the story that the holes whistle when wind crosses the plateau, and argued that the name Stonehenge itself descended from the Armenian. In 1999 he corresponded with Gerald Hawkins, whose Stonehenge Decoded had made the same kind of argument in the 1960s, and Hawkins wrote back that he admired the precise calculations Herouni had made. That endorsement is a large part of why the claim carried, so it is worth being exact about what it was: a letter. Hawkins published no independent check of Herouni's method, and the endorsement is known through the correspondence alone. Weight it accordingly.
Why the case does not hold
The most useful counter-argument is mechanical rather than statistical, and it comes from A. Cesar Gonzalez-Garcia in a chapter titled Carahunge: A Critical Assessment, written for the Springer handbook of archaeoastronomy. A hole five centimetres wide through half a metre of stone is a wide aperture, not a sight. To get the angular precision the observatory claim requires, you would need a narrow tube fitted inside the perforation, and no tube, no fitting and no trace of one has ever been found. Without it, he argues, the claimed astronomical significance of the orientations simply vanishes. His conclusion on the whole body of alignment claims is blunt: they are untenable.
The counting problem sits behind that. Around 80 pierced stones, each hole defining a line in two directions, offers well over a hundred candidate bearings, and once you are free to search several millennia of precession for a match, something will always land on something. Many of the stones have fallen, shifted or been re-erected, so the bearings measured today are not reliably the bearings of the builders. Clive Ruggles makes a different point about the framing: comparing the site to Stonehenge is unhelpful, because the mid-twentieth-century case for Stonehenge as an observatory has itself been largely dismantled. Borrowing the prestige of an argument that did not survive is not evidence.
What the digging actually found
Archaeologists were working here long before the observatory argument existed. The ethnographer Stepan Lisitsian gave the first recorded modern scholarly description in 1935, and thought the stones had been used to hold animals. In the 1950s the archaeologist Marus Hasratyan identified burial chambers dating to the eleventh to ninth centuries BC, which is what first put the funerary reading on an empirical footing rather than an impressionistic one.
In 2000 a team from the Institut für Vorderasiatische Archäologie at the University of Munich, led by Stephan Kroll, carried out a systematic field survey with associated excavation and identified the site as a necropolis, used mainly from the Middle Bronze Age into the Iron Age. Around thirty graves have been recorded, spanning roughly 2000 to 1600 BC down to the tenth century BC, with a low central mound of flat grey stones belonging to the early first millennium BC and ringed by menhirs. Kroll's reading of the long stone rows is that they are the remains of a wall of rocks and compacted soil with uprights plugged in to reinforce it, most likely from the Hellenistic period, when the hill served as a refuge; the Munich work placed that reuse somewhere between roughly the third century BC and the third century AD. Pavel Avetisyan, who has worked at the site, has said in print that the observatory reading is not supported by the archaeological record, and that the place is best understood as a multi-use funerary complex. There is material to look at as well as an argument to have: grave goods from the Bronze Age cist burials are held in the Sisian History Museum, and petroglyphs have been recorded on some of the stones.
That is a humbler account than an observatory, and it is the better supported one. It is worth being precise about how well supported, because the necropolis reading is often presented as a closed case and it is not. No comprehensive, stratigraphically controlled excavation of the whole site has been published. Work here has been limited and intermittent, constrained by funding, and Armenia's Institute of Archaeology and Ethnography has said that long-term systematic excavation is what the site needs and has not had. The site is not on the UNESCO World Heritage List. A survey with partial excavation behind it is a great deal more than Herouni ever had, and it is still not the same thing as a settled site.
The name is younger than the argument
Locally the place was Ghoshun Dash, Turkic for army of stones, rendered into Armenian as Zorats Karer. Villagers also called it Dik-Dik Karer or Tsits Karer, the upright stones. Karahunj, or Carahunge, is recent: proposed in the 1980s and popularised in the 1990s, and there is a village called Karahunj a few kilometres away that is the more likely source of the word than any Bronze Age acoustics. The oldest local mention that may bear on the site is by the thirteenth-century Armenian historian Stepanos Orbelian, who names a settlement called Carunge in the district of Tsluk, a name usually read as stone treasure or foundation stones. The newer name has since spread onto maps, signposts and tourist literature, which is how a hypothesis from the 1990s ended up as the name on the road.
Two claims attached to that name should be named rather than left standing. The first is that the holes whistle when the wind crosses the plateau, which is where speaking stones comes from. The second is that the English word Stonehenge descends from the Armenian. Neither has ever been set out with a measurement, an acoustic test or a published philological argument behind it by anybody; they circulate because they are attractive. We are naming them here in order to refuse them. The derivation from kar and hunge is one proposed reading of the word rather than the settled one, and it is contested among Armenian linguists.
What is still open
Nobody has dated the erection of the stones directly. The dates belong to the graves and structures around them, and the two need not be the same age. No published work has established that the uprights and the cist graves belong to a single phase at all. If they do not, then necropolis or observatory is a false binary, and two camps may have spent forty years dating different features of a multi-period site. A properly stratified excavation would have to answer that before it answered anything else. The same question hangs over the wall: whether it was built as a wall from the start, or whether it was a later adaptation that reused monoliths already standing on the hill, is unresolved.
The holes have no accepted explanation at all. They have been read as sights, as tether points for animals, as anchors for ropes used to raise or drag the stones, and as fixings for timber or textile structures that have not survived. They are smoothly finished, deliberately placed near the tops of the slabs, and expensive to make with Bronze Age tooling, which is exactly why the question refuses to die. There is also a plainer obstacle: much of the excavation literature is in Armenian and Russian, and very little of it has been published in English. That imbalance is a large part of why the observatory story travels further than the archaeology does. A hole can frame a star, but it can also carry a rope, a beam or an offering, and no one can yet say which was meant.
What has changed since Herouni
Two things, and both of them cut against the picture of a closed argument between entrenched camps.
The first is that the strong claim and the weak claim have come apart. A 5500 BC observatory built by an ancestral civilisation is rejected, and the mechanical objection above is why. A much narrower question, whether any of these Bronze Age stones were set up with solar or lunar events in mind, is still open and is still being measured. Researchers have been back on site since 2017 taking new measurements and combining them with skyscape modelling, reporting to the Journal of Skyscape Archaeology and to the European Society for Astronomy in Culture, and they find that some stones do show orientations consistent with solar and lunar events. Whether that was intended is precisely what they have not resolved, and they say so. That is a published result that nobody has yet replicated, and it should be read as one: an open question, not a finding, and not a rescue of Herouni.
A stronger recent claim needs holding further out. In 2025 a summary of laser-based surveys carried out since 2020 was presented at the International Conference on Astronomical Heritage of the Middle East, reporting what its authors describe as stone observational instruments, including platforms and angular stones, and arguing for deliberate astronomical functions across several chronological layers. It has not been through peer review and no one has independently verified it. A conference paper is not evidence until somebody else can repeat it, and we are recording this one as new and unreviewed rather than as a result.
The second change is institutional, and it is the most useful news out of this site in years. The Institute of Archaeology and Ethnography at Armenia's National Academy of Sciences and the NGO Bnorran Historic-Cultural have argued opposite sides of this for years. They have now agreed a joint research plan rather than continuing to publish past one another. Its results were not out as of mid-2026. If the phasing question is ever answered, that is the arrangement likely to answer it.
What our engines found
We ran the same test here as everywhere else, with the settings fixed in advance and identical at every one of the fifty sites we cover: a 200 km search radius, at least 12 km between sampled sites, 15 sites, a 2 per cent tolerance, and 2,000 Monte Carlo runs against a null model resampled from real catalogued sites in the same region. The settings are frozen because the answer moves if you tune them, and tuning them after seeing the answer is exactly the trick we object to elsewhere. Karahunj returned 12 golden-ratio pairs where chance predicts 13.2, giving p = 0.6357.
That is fewer than chance would predict, and it is not significant. The control points the same way: the arbitrary ratio 2.0 scored 1.20 times its expected count against phi's 0.91, so nothing here singles out the golden ratio.
We publish the misses because a tool that only reports its hits is not measuring anything. Every figure above is reproducible from the same coordinates, the same settings and the same seed.
One limit worth stating plainly, because it is the one most often glossed over. This test compares positions between monuments. It says nothing about any building's own proportions, its height, base, slope or perimeter. That is a separate question, and at several sites a genuinely open one.
That limit applies directly to the holes and their bearings. A test comparing distances between catalogued sites cannot examine what a single stone points at, and a null result here is not a verdict on the observatory question in either direction.
Explore it yourself
Open the Karahunj record in the living atlas to walk the rows in 3D, see the site among its catalogued neighbours and set the sky to any date you like over the stones. The Karahunj site-emblem tee carries the pierced stone, the site geometry and the exact coordinates in the Pyramason Chronicle style.
Sources worth reading
Parsamian, "Carahunge — an ancient observatory in Armenia", Astronomicheskii Tsirkulyar (1985), the citable original of the solstice observation. Herouni, Armenians and Old Armenia (Tigran Mets, Yerevan, 2004), where the strong claim is actually set out. Barnard and Avetisyan, Zorats Karer: field survey and excavation report, Institut für Vorderasiatische Archäologie, LMU Munich (2000). Gonzalez-Garcia, "Carahunge: A Critical Assessment", in the Springer Handbook of Archaeoastronomy and Ethnoastronomy, for the mechanical objection. Ruggles on why the comparison with Stonehenge does not help. Khnkikyan's 1984 survey, for the stone and hole counts that everything since has argued over.