Earth’s outer core is an ocean of liquid metal whose volume roughly equals that of Mars. Seismic waves from repeated French nuclear tests at Mururoa have now revealed parts of it changing over only a few years: signals arrived up to 0.2 seconds early or late, suggesting that structures hundreds of kilometres wide are moving through the molten core on human timescales.

Earth’s outer core is not a ball but a shell: liquid metal beginning roughly 2,900 kilometres below the surface and ending at the solid inner core about 5,150 kilometres down. Put standard boundary radii into the formula for the volume of a spherical shell and the result is about 169 billion cubic kilometres. Mars occupies about 163 billion.

The comparison gives scale to a 2026 result that is otherwise easy to lose in fractions of a second. Seismic signals from French underground nuclear tests at Mururoa in French Polynesia did not always take the same time to cross the core and reach a station in Kazakhstan. Some arrived about 0.1 to 0.15 seconds early; by 1995, comparable signals were roughly 0.15 to 0.2 seconds late.

Those shifts do not show metal visibly sloshing beneath the mantle. They show that the effective seismic travel time along a core-sensitive path changed between 1977 and 1995. Geophysical modelling then asks what kind of moving or evolving structure could produce it.

The Mars comparison is geometry, not metaphor

The US Geological Survey describes a liquid outer core about 2,200 kilometres thick surrounding a solid inner core roughly 1,250 kilometres in radius. More precise reference models place the core-mantle boundary near a radius of 3,480 kilometres and the inner-core boundary near 1,221 kilometres.

Subtracting the inner sphere from the full core gives an outer-core volume slightly larger than Mars. It is mostly iron, with nickel and lighter elements, held liquid by extreme heat despite immense pressure. Seismic shear waves do not pass through it, while compressional waves slow, bend and reflect through the fluid.

That enormous volume is also difficult to observe. No drill can approach it, and the mantle hides it from every camera. Almost everything known about the core has to be reconstructed from waves that enter it and return to instruments at the surface.

Mururoa supplied unusually repeatable sources

The study in the Journal of Geophysical Research: Solid Earth, by Virginia Tech geophysicist Ying Zhou, examined short-period waves generated by underground tests in French Polynesia and recorded at the BRVK seismic station in Kazakhstan.

Earthquakes are the usual lamps used to illuminate deep Earth. The problem is repeatability. Two natural ruptures may begin kilometres apart, differ in depth or release energy in different directions. A changed arrival can then reflect the source rather than the material crossed by the wave.

The Mururoa tests carried a grim historical cost, but as seismic sources they had a useful property: many occurred in nearly the same area. Zhou analysed 112 pairs whose published epicentres lay within 0.05 degrees of one another. Each pair sent energy through approximately the same parts of Earth at two different times.

Similar is not identical. Event locations and test characteristics retain uncertainty, so the method used differences between multiple seismic phases rather than treating any single arrival as a stopwatch reading of the core.

A newly described phase reflected inside the outer core

The first-arriving PKP signal posed a puzzle. Conventional ray theory did not explain its timing and shape. Finite-frequency sensitivity calculations instead identified it as an internal reflection within the outer core, with unusual waveform behaviour produced around a caustic where neighbouring wave paths converge.

Zhou calls the phase PKrKP. In seismic notation, P denotes a compressional wave in the mantle and K denotes its passage through the outer core. The added lower-case r marks the newly interpreted internal reflection.

The analysis compared PKrKP with PP, a phase that reflects from Earth’s surface and travels mainly through the mantle without entering the core. It also used PKiKP, which reflects from the boundary of the solid inner core. Double differential times measure how the gap between two phases changes across a pair of tests. Shared errors near the source or receiver should influence both arrivals and partly cancel.

This separation is the study’s main strength. The unusual phase gives sensitivity to the outer core itself rather than forcing researchers to infer its behaviour from a wave dominated by the inner core.

Two tenths of a second records a change across years

The reconstructed outer-core travel times were about 0.1 to 0.15 seconds faster during 1982–1983 and 1988–1990 than the 1977 reference. In 1995 they were approximately 0.15 to 0.2 seconds slower.

That is tiny beside the time needed to cross the planet, but not tiny to a short-period waveform measured against a closely related reference phase. An early arrival indicates a higher average compressional-wave speed along the sensitive part of the path. A late arrival indicates a lower one.

The result is not a claim that the entire outer core accelerated and decelerated together. The measurement samples a limited corridor through an object almost the volume of a planet. Most of the metal lies outside that corridor, just as one medical scan does not describe every tissue in a body.

The date pattern also matters. The signal did not simply drift in one direction for eighteen years. It was faster in two intervals and slower at the end, behaviour more consistent with transient heterogeneity crossing or evolving within the sampled region than with a uniform long-term trend.

The moving structure is a model, not a direct image

A lateral anomaly more than 700 kilometres across and about 100 kilometres thick in the low-latitude South Pacific part of the outer core could produce a travel-time shift of 0.1 to 0.2 seconds, according to the modelling. Zhou proposes suspended solid material moving through the liquid as one possible cause.

That does not mean seismologists have identified a single rigid slab with mapped edges. “Structure” here means a region whose density, composition, phase mixture or elastic response differs enough from its surroundings to change wave speed. Its dimensions are inferred from the wavelengths and sensitivity of the seismic phases.

Other physical explanations remain possible. Temperature can alter material properties, although wave speed in liquid iron at core pressure may be relatively insensitive to modest thermal variation. Concentrations of light elements can change density. Crystals or slurry near the inner-core boundary may be carried into neighbouring fluid. The data constrain the required seismic effect more directly than they identify the material producing it.

A 2022 study by Zhou found a different outer-core path becoming about one second faster across two decades. That work modelled localised, light-element-rich flow with a density deficit of 2 to 3 percent. The newer nuclear-test result uses a different source type, frequency range and core phase, adding evidence that rapid variations are not restricted to one earthquake comparison.

A changing core matters because it is Earth’s dynamo

The electrically conducting liquid outer core generates most of Earth’s magnetic field. Thermal and chemical buoyancy drive flow; planetary rotation organises part of that motion; the moving conductor sustains electrical currents and the geodynamo.

Previous SpaceDaily reporting described a low-velocity equatorial structure near the top of the outer core. That torus was inferred from waves reverberating for hours after earthquakes. The Mururoa analysis probes a different question: whether properties along an outer-core path changed with calendar time.

The two results should not be merged into one object. Together, they show why the uniform-shell diagram in a textbook is only a starting point. The liquid core can contain regional variations in composition, phase and motion, some persistent enough to map and others transient enough to cross a seismic path within years.

Seismic travel time does not directly reveal how a particular anomaly changes the magnetic field. Linking small-scale material movement to the geodynamo will require more paths, more wave types and comparison with geomagnetic models. One route from Mururoa to Kazakhstan cannot establish how common the behaviour is across the full core.

It can establish something more limited and still striking. Signals crossing a Mars-sized ocean of metal changed measurably within an eighteen-year record, including reversals over intervals of only a few years. The outer core is inaccessible, but on human timescales it is not still.