MYS0D32CBK250HV·Mystery · site

Gunung Padang

Mohammad Fadli · CC BY-SA 3.0

Where: Karyamukti, Campaka district, Cianjur, West Java, Indonesia⁠[12]
Structure: Five stone-walled terraces reached by a stair of about 370 steps⁠[1]
Material: Columnar-jointed andesite blocks from the volcanic hill⁠[1]
Excavation dates: Charcoal under the terrace 1 wall about 117 BC; under terraces 2 and 4 about 50 BC⁠[1]
First recorded: 1914, N. J. Krom; rediscovered by farmers in 1979⁠[1]
2023 claim: Buried pyramid begun before 14,000 BC; retracted 18 March 2024⁠[6]
Status: Partly explained
Questions: 1 of 5 answered

People Lutfi YondriArchaeologist who has excavated the terraces since the late 1990s Ali AkbarArchaeologist; co-author of the 2023 paper; leads the 2025 study and restoration team Danny Hilman NatawidjajaGeologist, BRIN; lead author of the retracted paper N. J. KromDutch archaeologist; recorded the site in 1914

Catalog keyMYS0D32CBK250HV

Gunung Padang is a promontory on an extinct volcano in Cianjur, West Java, whose top was cut into five terraces held up by retaining walls and covered with rectangular enclosures made from prismatic blocks of andesite. A stair of about 370 stone steps climbs to it from the valley. The blocks are columnar joints, a shape the rock takes naturally as lava cools, and the builders used them for walls and enclosures⁠[1]⁠[2].

Archaeologists who have excavated the terraces since the 1980s date them by charcoal under the walls to the last centuries BC, about 117 BC under terrace 1 and about 50 BC under terraces 2 and 4, and read the site as a punden berundak, a terraced shrine to ancestral spirits that was used for centuries⁠[1]⁠[3].

In October 2023 the geologist Danny Hilman Natawidjaja and colleagues published a paper arguing that the hill is a buried pyramid built in stages, the oldest dated to 25,000 to 14,000 BC, with hidden chambers inside⁠[4]⁠[5]. In March 2024 the publisher and the editors retracted it: the dates came from soil with no artefacts or features that could be read as human work, so the inference of a pyramid 9,000 or more years old was unsupported. The authors disagree with the retraction⁠[6]⁠[7].

In 2025 the Ministry of Culture began a study and restoration programme led by Ali Akbar, a co-author of the retracted paper. The team excavated on all five terraces and began shoring the landslide-prone upper ones⁠[8]. In December 2025 it announced, without publishing data, that carbon from about 4 m under the fifth terrace gives an age of about 6000 BC⁠[9]⁠[10]. Of the 5 questions on this site, 1 are answered. The most likely answer to “Is there a human-built structure inside the hill older than the visible terraces?”, not yet confirmed: the core is a volcanic body with natural columnar jointing; the deep layers are weathered and slumped columns….

Key questionsset by the editors

Is there a human-built structure inside the hill older than the visible terraces?Most likely: the core is a volcanic body with natural columnar jointing; the deep layers are weathered and slumped columns…

When were the visible terraces built?Answered: the terraces were built over several generations from about the 2nd to the 1st century BC and used for…

All 5 questions

The site

The terraces step up a narrow ridge, the lowest and largest first, the fifth and smallest at the top. Each carries enclosures, paths and standing stones built from andesite columns, and a stone well at the foot of terrace 1 may have supplied water for ceremonies. One block on terrace 1, the batu kecapi, rings when struck⁠[1]. Yondri has studied how the columns were laid and stacked in the walls⁠[2].

The five stone terraces and stairBeeyan. CC BY-SA 3.0Artifact

The Dutch archaeologist N. J. Krom recorded the site in 1914. It was then ignored for decades until farmers from Karyamukti reported it in 1979, and the national archaeology institute excavated and exposed its layout in the 1980s⁠[1]. Between 2011 and 2015 a team of geologists, with government support, drilled and surveyed the hill in search of a buried structure, a project criticised at the time as driven by national pride⁠[4]⁠[11].

Map of the areaOpen the full map

Area Region

+−

Gunung PadangTerraced megalithic site on a hilltop

Within 100 m

West JavaProvince

A region

MappedA regionNatural Earth

Timeline

Timeline

Squeezed where little happened. Each mark is a published claim, by how it stands:AcceptedMost likelyStill possibleDisputedRuled out

What was found

Excavation on the terraces has produced charcoal under the retaining walls, pottery and a grinding stone, but no animal bone that would point to feasting. Yondri reads the spread of dates as building by several generations and use over many centuries⁠[1].

The 2023 paper reported four layers from the surface down: Unit 4, the visible megalithic terraces; Unit 3, columnar rock in a matrix of soil; Unit 2, coarse sand and pillar-like rock; and Unit 1, a lava core that it argued had been sculpted. Radiocarbon dates on organic soils from cores gave four phases: 25,000 to 14,000 BC, 7900 to 6100 BC, 6000 to 5500 BC, and 2000 to 1100 BC. Resistivity and seismic tomography showed a high-resistivity, low-velocity zone that the authors read as cavities, and drilling lost water at depth⁠[4]⁠[5].

First page of Geophysical surveys of the hill, 2011 to 2015 (ERT, GPR and seismic tomography) — Geophysical surveys of the hill, 2011 to 2015 (ERT, GPR and seismic tomography)Danny Hilman Natawidjaja et al., 2011. CC BY 4.0Artifact

In 2025 the government team reported buried terrace-like structures beside the stair, four upright stones embedded deep in the ground, and a possible source of columns at Pasir Pogor, about 5 km away. In December it announced carbon-bearing material at about 4 m under terrace 5 and a date of about 6000 BC, with no laboratory, sample list or calibrated ranges given⁠[12]⁠[10].

What is known

The visible terraces are a human construction. The radiocarbon dates from charcoal under their walls fall in the last centuries BC, and archaeologists place the main megalithic building around 2,000 years ago⁠[1].

The soil dates in the 2023 paper are not in dispute as measurements: critics accept that the samples are as old as reported. What is disputed is what they date. None of the dated soils contained charcoal, bone or artefacts, so they date the formation of the soil, not human work⁠[13]. That is the error the journal named when it retracted the paper⁠[6].

In the region, people lived in caves between about 12,000 and 6,000 years ago, and no stonemasonry of that age is known; Indonesian archaeologists are therefore sceptical of any large construction in the Pleistocene⁠[13]⁠[1].

The explanations

The buried-pyramid reading holds that the hill was shaped and built up in four phases over 20,000 years, with chambers inside⁠[4]. It predicts worked stone, tools, hearths or other human traces in the deep layers, dates on material that people made or burned, and voids of regular shape. None of these has been published; the retraction removed the dating basis⁠[6].

The natural-hill reading holds that the core is a volcanic body whose rock broke into columns as it cooled, and that the deeper "layers" are weathered and slumped columns and soil. Dibble notes that blocks moving downslope tend to align, which can mimic order⁠[13]. It predicts columns in their cooling orientation at depth and soils without anthropogenic inclusions.

The archaeological reading holds that people in the last centuries BC terraced the top of a natural hill and built a shrine from the columns lying on it, which they used for centuries⁠[1]⁠[3]⁠[14]. It predicts that every secure date for construction falls in the late first millennium BC or later.

The 2025 government team's position sits between: an earlier human construction phase in the early Holocene, about 6000 BC, below the visible terraces⁠[10]. It cannot be judged until the samples, their contexts and the laboratory results are published.

Published claims

Published claims and their status

Year	Who	Claimed	Status	Reason
2014	Lutfi Yondri⁠[3]	The terraces are a punden berundak, a terraced shrine to ancestral spirits laid out to suit the hill and its surroundings.	Accepted	The standard archaeological reading, consistent with the finds and with the other terraced shrines of Indonesia⁠[1]⁠[14].
2016	Lutfi Yondri⁠[2]	The builders laid and stacked the natural andesite columns in regular patterns to make the walls and enclosures.	Accepted	The visible walls are accepted as human work by all parties, including the authors of the pyramid claim⁠[4].
2018	Danny Hilman Natawidjaja and colleagues⁠[16]	A large pyramid-like structure inside the hill predates 10,000 years before present.	Disputed	The fuller 2023 version was retracted because its dated soils show no human activity⁠[6].
2022	Graham Hancock⁠[17]	Gunung Padang is an Ice Age pyramid with hidden chambers, built by a lost advanced civilisation.	Disputed	It rests on the soil dates the journal later found do not date human work, and no Ice Age stonework is known in the region⁠[6]⁠[13].
2023	Danny Hilman Natawidjaja and colleagues⁠[4]	The hill is a pyramid built in four phases, the oldest between 25,000 and 14,000 BC, around a sculpted lava core.	Disputed	Retracted in 2024 because the dated soils contain no artefacts or features that could be read as human work; the authors dispute the retraction⁠[6]⁠[7].
2023	Danny Hilman Natawidjaja and colleagues⁠[4]	The visible megalithic terraces were built between 2000 and 1100 BC.	Disputed	Charcoal sealed under three terrace walls dates to the last centuries BC, though those results are reported only in the press⁠[1].
2023	Danny Hilman Natawidjaja and colleagues⁠[4]	A zone of high resistivity and low seismic velocity inside the hill marks built chambers.	Still possible	No borehole or camera has reached the anomaly, which could equally be fractured or loose rock⁠[13].
2023	Lutfi Yondri, Flint Dibble and Bill Farley⁠[13]	The dated soils contain no charcoal, bone or artefacts, so they date the formation of the soil, not any building.	Accepted	The journal retracted the paper on the same ground in March 2024⁠[6].
2023	Flint Dibble⁠[13]	Columns sliding downslope tend to line up, so the apparent order in the buried layers could be natural.	Still possible	A plausible geological mechanism that no fieldwork on the hill has yet tested⁠[15].
2024	Archaeological Prospection and Wiley⁠[6]	The 2023 paper's radiocarbon dates come from soil with no sign of human activity, so they cannot support a pyramid 9,000 or more years old.	Accepted	No data published since show human-made material in the dated layers⁠[1].
2024	Lutfi Yondri⁠[1]	Charcoal under the terrace walls dates to about 117 BC on terrace 1 and about 50 BC on terraces 2 and 4.	Accepted	No secure construction date older than these has been published, though the laboratory details are not in print⁠[1].
2024	Harry Sofian⁠[14]	Gunung Padang is a terraced shrine of the punden berundak type, not a pyramid.	Accepted	Matches the excavated finds and the dates under the terrace walls⁠[1].
2025	Ali Akbar's government study team⁠[12]	Terrace-like structures lie buried beside the stair beyond the five visible terraces, with upright stones set deep in the ground.	Still possible	Announced to the press with no plans, sections or contexts published⁠[12].
2025	Ali Akbar's government study team⁠[10]	Carbon from about 4 m under the fifth terrace dates a building phase to about 6000 BC.	Still possible	No laboratory, sample list or calibrated ranges have been released, and the 2023 paper's dates of this kind were found unsound⁠[6].

What would settle it

Dated material that only people produce, such as charcoal from a hearth sealed under a built wall, or seeds and bone from an occupation floor, taken from the deep layers in a published excavation, would test any pre-Holocene or early Holocene phase. The 2025 samples need a published list with depths, contexts, materials and laboratory numbers.

The magnetic-susceptibility and orientation work begun by Universitas Indonesia in late 2025 can show whether the buried columns lie in the orientation they had when the lava cooled, which would mean natural, or were moved⁠[15]. An independent re-processing of the 2023 resistivity and seismic profiles, with drill cores logged by a second team, would test the cavities.