The sea south of Sri Lanka sits 106 metres low. Nobody agrees why.

Sail south from Sri Lanka into the open Indian Ocean and, over an area of roughly three million square kilometres, the surface of the sea sits as much as 106 metres lower than it would on a smooth, evenly weighted Earth.

Nobody on board would notice. The drop is spread over something like a thousand kilometres, which works out to around ten centimetres per kilometre. There is no cliff of water and no whirlpool. The sea is simply lower, because gravity there is fractionally weaker, and water flows away towards places where the pull is stronger.

Geophysicists call it the Indian Ocean Geoid Low. Newspapers call it a gravity hole. It is the deepest dip in Earth's gravity field by the standard way of measuring it, and its cause has been argued over for decades. In 2023 a team in Bengaluru published an explanation that was widely reported as the mystery solved. It was not.

What a geoid is

If you switched off the tides, the currents and the wind, the oceans would settle into a shape set by gravity and Earth's spin alone. Extend that surface under the continents and you get the geoid, the shape of Earth's gravity.

It is not a smooth ball, because the planet's mass is not spread evenly. Where there is more mass under the surface, gravity pulls slightly harder and water piles up. Where there is less, water drains away. The bumps and hollows run to tens of metres over thousands of kilometres, and satellites such as the European Space Agency's GOCE, which flew from 2009 to 2013, have mapped them in detail. Exaggerate them ten thousand times, as in the picture above, and Earth looks like a lumpy potato. The deepest blue bruise on that potato sits south of India.

A dip in the geoid means a shortage of mass somewhere underneath. Not a hole in the sea floor. Something deep inside the planet, below this patch of ocean or near it, is lighter than the rock around it.

Found from a submarine

The first measurements came from an unlikely place: the inside of Dutch naval submarines.

Measuring gravity precisely means timing a pendulum, and a pendulum on a ship rolling in the swell is useless. In the 1920s the Dutch geophysicist Felix Andries Vening Meinesz built an apparatus with pendulums swinging in opposite phases, so the motion of the vessel could be cancelled out mathematically. Even then a surface ship vibrated too much. A submarine at about 30 metres, below most of the wave motion and running on quiet electric motors, did not.

Between 1923 and 1939 he went to sea on a string of Dutch submarines. The best known was the K XVIII, which left Den Helder on 14 November 1934 and reached Surabaya eight months later, by way of South America, Cape Town, Mauritius and Fremantle. On that voyage alone the boat dived 140 times so he could take 240 gravity measurements. The sailors called the apparatus Het Gouden Kalf, the Golden Calf. It is now kept at the TU Delft Library.

His analysis of the whole programme was published in 1948, and the large circular low in the Indian Ocean is usually dated to it. You will often read that the low was found by a ship survey in 1948. The measurements were made from submarines, mostly in the 1920s and 1930s, and 1948 is when the results came out.

Something light is down there

For decades the leading ideas involved old ocean floor that had sunk into the mantle, and the arguments did not converge. A 2017 paper from the group of Attreyee Ghosh at the Indian Institute of Science put it bluntly in its first lines: several theories had been proposed, and there was "no general consensus regarding the source of this particular anomaly".

Ghosh's team built computer models of the mantle flowing under its own density differences, feeding them three-dimensional pictures of Earth's interior made from earthquake waves. Several of those pictures produced a decent match to the planet's geoid overall. Only four of them got the Indian Ocean low in the right place with the right shape. Those four pointed to the same culprit, which Ghosh described in the institute's announcement:

Our study explains this low with hotter, lighter material stretching from a depth of 300 km up to ~900 km in the northern Indian Ocean, most likely stemming from the African superplume.

The "African superplume" is one of two continent-sized regions of hot, possibly chemically different rock sitting on top of Earth's core, one under Africa and one under the Pacific. The 2017 idea was that hot material peeling off its edge had drifted north-east, helped along by the Indian plate moving the same way, and come to rest under the Indian Ocean.

The 2023 answer

Six years later the same lab went further. Debanjan Pal and Ghosh ran mantle models forward through time, starting 140 million years ago and moving the tectonic plates according to reconstructions of their history. They published in Geophysical Research Letters on 5 May 2023.

Before India crashed into Asia, an ocean called the Tethys lay between them. As India moved north, the floor of that ocean was pushed down into the mantle. Those cold, dense slabs sank all the way to the bottom and pressed against the African blob, which squeezed out plumes of hot, light material. The plumes rose into the upper mantle and spread under the Indian Ocean. In their models the geoid low starts taking shape about 20 million years ago.

It is an elegant result, and it went round the world. Several outlets reported it as the 75-year-old mystery solved. The paper itself is more careful. Its abstract opens by saying the origin of the low "has been controversial", and it is candid about method: the authors tuned the density and viscosity of the material at the base of the mantle, the behaviour of a key boundary 660 kilometres down and the strength of the sinking slabs, specifically to control how the plumes rose, and the plumes in turn set the shape and depth of the low. That is normal practice in this kind of modelling. It also means the models were adjusted until they produced the thing they were trying to explain.

The plume that made the Deccan

The sharpest objection came from Alessandro Forte, a geodynamicist at the University of Florida who was not involved in the work.

The most outstanding problem with the modeling strategy adopted by the authors is that it completely fails to reproduce the powerful mantle dynamic plume that erupted 65 million years ago under the present-day location of Réunion Island.

The Réunion hotspot is the one whose arrival around 66 million years ago poured out the Deccan Traps, the vast stack of lava flows that covers much of western and central India and that erupted around the time the dinosaurs died out. A model of mantle flow under the Indian Ocean that does not produce the plume behind the Deccan has, in Forte's view, missed one of the biggest events in that region's history. He also pointed out that the modelled geoid did not match the real one well outside the Indian Ocean, which he said may reflect shortcomings in the simulations.

And then it stopped being the deepest

In December 2025 Forte and Petar Glišović published their own study in Scientific Reports, running mantle convection backwards in time from present-day seismic pictures. Its main subject is not the Indian Ocean at all. It is Antarctica.

The "106 metres" is measured against a standard reference shape that geodesists use, a slightly flattened sphere. Measure instead against the shape a spinning ball of fluid would naturally take, which is what matters if you want to understand forces inside the planet, and the picture changes. In the paper's words:

Unlike geodetic reference frames that place the deepest geoid low in the Indian Ocean, a geodynamic perspective, relative to a hydrostatic ellipsoid, reveals the strongest nonhydrostatic geoid depression resides over Antarctica.

They trace that Antarctic low back at least 70 million years, with a major change in its size and position between 50 and 30 million years ago that lines up with an abrupt shift in Earth's rotation axis about 50 million years ago. Nothing in this makes the Indian Ocean low disappear. The sea there is still lower, the measurements still stand. What changes is the claim that it is the single most extreme dip in the planet's gravity. That depends on which yardstick you hold against it.

What is actually known

The low is real. Pendulums swinging inside submarines in the 1930s and satellites in the 2010s agree on it. It is caused by a shortage of mass inside the planet, not by anything at the surface. Both of the Bengaluru studies point at hot, buoyant rock in the upper and middle mantle, a few hundred to about a thousand kilometres down, connected one way or another to the great hot structure under Africa.

What is not settled is how that rock got there and when. One lab has offered two different mechanisms six years apart, the second built on plate history and tuned mantle properties. Alessandro Forte says the second model misses the plume that built the Deccan, and has since argued that the lowest point in Earth's gravity is over Antarctica, depending on how you measure.

A patch of ocean nearly the size of India sits lower than the rest of the sea, off the southern tip of India. Seventy-odd years after a Dutch professor first detected it from a submarine, nobody can yet say for certain why.