Krakatau Erupted On Sunday. Nine Sensors On Gunung Padang Heard A Room 22 Metres Down.

CIANJUR, Indonesia — At 05:38 on Sunday morning, Western Indonesian Time, the ground under Gunung Padang began to shake, and it went on shaking for seventy-one minutes.

The source was Anak Krakatau, two hundred kilometres to the west, whose eruption closed Jakarta's airport by lunchtime. On the terraces of a hill above Cianjur it did something else. It lit up nine seismometers that had been buried there since June, listening for three months to nothing louder than a road.

By 11:00 the array's principal investigator had an image. Between 22 and 26 metres below the top terrace there is a volume where the waves slow almost to nothing. It is fifteen metres long, four metres high, and its top and bottom are flat.

The floor is flat. That is the finding.

The peg

The Associated Press reported the eruption as an aviation story: flights cancelled at Jakarta and elsewhere. That is what it was for most of Java.

For nine instruments on a hill in Cianjur it was the loudest thing to pass through the ground in a century.

The hill

Gunung Padang is a terraced hill of columnar basalt, four levels stepping up a ridge, excavated to the terrace floors between 2012 and 2014. What is above the floors nobody disputes: stacked columns, some mortared, arranged by people.

In 2023 a paper in a geophysical journal said the hill went on being built for a long way down, in three units, the deepest dated to more than twenty thousand years ago. In March 2024 the journal retracted it. The dates, the retraction said, were on soil, and nothing tied the soil to anything a person had made.

The retraction was about years. It said nothing about whether there was structure at depth. The deep units stayed disputed and the excavation permit stayed suspended, as it has been since 2014.

What was on the hill already

Dr. Wisnu Adiputra-Halvorsen of Bandung and Dr. Femke Ruysdael-Okafor of Utrecht buried nine broadband seismometers on the terraces in June, under a permit that allows instruments and forbids blades.

The method uses whatever shakes the ground, traffic and surf and weather, and over months builds a picture of how fast waves travel through the rock. Slow means loose, or wet, or hollow. Three months of the road to Cianjur produced, in Dr. Adiputra-Halvorsen's word, a smear: a slow zone somewhere between fifteen and twenty-five metres, with no shape.

Then the volcano gave them seventy-one minutes of a source a thousand times louder.

What the tremor shows

The time the tremor took to cross from one station to the next is the whole of the data.

Under Terrace 1, between 22 and 26 metres, the waves slow to a speed that is not sand and is not wet basalt. Dr. Adiputra-Halvorsen's velocity model puts it at air. The volume is fifteen metres long on the axis of the terraces, about four metres high, and its upper and lower surfaces are sharp on the image to within half a metre.

A lava tube has a rounded roof. A collapse cavity has an irregular one. This has a flat one, and a flat floor under it.

Dr. Ruysdael-Okafor has the same data in an airport in Singapore and is running it through a different code. She will not say what it shows because she has not let herself look. Her reply, at the foot of this page, says why.

The man who swapped the cards

The stations write to memory cards and send a low-rate copy over a cellular modem. The full-rate data, which is what the image needs, was on the cards.

At first light on Sunday Pak Dadang Sumarna-Oyelaran, who has looked after the site since 2009, walked the terraces and swapped all nine, without being asked, because the investigators had told him in June that if the ground ever shook he should not wait. He had them in a plastic bag at the site office by 07:15. The image exists because of that walk.

The permit

The array's permit is for nine buried instruments. Excavation below the terrace floors has been suspended since 2014, and the directorate confirmed on Sunday afternoon that no application to resume it is on file.

The investigators will not make one. It is not their field. What they can give the ministry, as of Sunday, is a set of coordinates and a depth, to within a metre.

What this desk takes from it

For three years the argument about this hill has been about a number of years, and the number is no closer today than it was on Saturday.

What is closer is the hole. There is a volume of air fifteen metres long under Terrace 1, with a flat floor, and a volcano told nine instruments where it is. Somebody with a permit and a drill could put a camera into it in a week, as this newspaper did under another hill last month.

The second image is being run by a woman who has promised not to look until it is finished. When it is, we will print it, whether it shows the room or shows that we should not have called it one.