What is inside the Earth?

Seismic waves from earthquakes refract, reflect and change speed at boundaries inside the Earth. Reading those arrivals at stations worldwide gives a layered model: crust, mantle, liquid outer core, solid inner core. The inner core was proposed in 1936 by the Danish seismologist Inge Lehmann from an analysis of P-wave data: inside the core, she wrote, “there is an inner core in which the velocity is larger than in the outer one”.

A reference model built from those arrival times, ak135, supplies the travel times the International Seismological Centre uses to locate earthquakes.

Convection in the liquid iron outer core, combined with Earth’s rotation, sustains the magnetic field by dynamo action. The field has reversed polarity many times, recorded in the magnetisation of seafloor basalt, and the striped pattern those reversals left became one of the main lines of evidence for plate tectonics.

A diamond from Brazil was found to contain an inclusion of ringwoodite carrying roughly one percent water by weight, direct evidence that the mantle transition zone is hydrous, at least locally. Nobody went down for it: it came up inside the diamond, and the authors infer that some kimberlite magmas come from that deep.

Beneath Africa and beneath the Pacific sit continent-sized regions where seismic waves travel unusually slowly. They are real, they are large, and their composition and origin are debated: proposals range from piles of ancient subducted oceanic crust to primordial material that never mixed, to the remains of an early planetary collision.

The inner core appears to turn at a slightly different rate from the rest of the planet, and the difference changes over decades. It is read from repeating earthquakes whose waves should travel the same path twice, and two groups reading South Sandwich Islands earthquakes recorded in northern North America agree that the rotation has lately been turning back and differ on what it is part of. One reads a pause and a reversal inside an oscillation of about seven decades. The other tracks a super-rotation from 2003 to 2008 and then a return along the same path, two to three times slower, through 2023. That group later found that some of the change comes from the inner core’s surface deforming, and proposes, tentatively, that it both turns and changes shape.

A review of the geophysical estimates puts about four hundredths of an ocean’s worth in the upper mantle, between a fifth of an ocean and a whole one in the transition zone, and less than two oceans in the lower mantle. It also says which of those numbers to trust: that the transition zone is a major reservoir is agreed, and the lower mantle is poorly constrained. Each figure extrapolates electrical-conductivity and seismic measurements from particular regions across a whole layer, and the transition-zone range assumes that about 30 per cent of it is wet.

Composition is inferred by matching seismic velocities against laboratory measurements and calculations for candidate minerals under extreme pressure. The same average profile has been matched by different mixtures: one group’s measurements pointed to a lower mantle more than 93 per cent perovskite by volume, another group’s calculations to a mantle of one composition from top to bottom.

Microbes have been recovered from about 2.5 kilometres below the sea floor, and the rock beneath the continents is estimated to hold 2 to 6 × 10²⁹ cells. Cells in these settings break down food ten thousand to a million times more slowly than microbes in a rich laboratory culture. Temperature is taken to set the lower limit, and it is not a clean line: in the Nankai Trough, vegetative cells fall a hundredfold above 45°C, yet hyperthermophiles are active at 100 to 120°C. Where exactly that limit falls has not been established.

The core formed early, within about the first 30 million years of the Solar System on the evidence of hafnium and tungsten isotopes. How much of the planet’s earliest makeup survives in the mantle is an open question, though some modern lavas carry tungsten signatures from domains formed in the first 50 million years or so.

One proposal identifies the low-velocity provinces with material from the body that struck the early Earth and formed the Moon, fragments that sank and never mixed. It comes from impact and mantle-convection simulations, which show that it could have happened, and it is not established.

Reaching the mantle by drilling through thin oceanic crust has been the goal since Project Mohole was proposed to the US National Science Foundation in the late 1950s. In 1961 its platform, CUSS I, was held on target in a strong current, and drilling from a ship became feasible. The last of the programmes that followed ran from October 2013 to September 2024, and in 2023 it drilled the deepest hole yet into oceanic mantle rock: 1,268 metres of serpentinised peridotite from the Atlantis Massif, more than six times the previous record, altered by hydrothermal fluid over its whole depth. Faulting had already lifted that mantle to the sea floor, so the drill did not have to pass through the crust. Only drilling gives a continuous, oriented section of the mantle rather than loose pieces.