Earth’s Interior: From Crust to a Wandering Core

Oct 10, 2026 · @Thomas Abshier

A Planet We Have Never Seen Inside

Thomas Lee Abshier, ND | 10 October 2026

No one has ever seen beyond Earth’s outer skin. The deepest hole ever drilled reaches about 12 kilometers, less than two-tenths of one percent of the way to the center. Yet we can say with some confidence that the planet has four main layers: a thin rocky crust, a slowly churning rock mantle, a liquid iron outer core, and a solid iron inner core about the size of Pluto. We know the depth of each boundary to within a few kilometers, and we know that the innermost ball does not turn in perfect step with the rest of the planet.

We learned all of this from the surface, almost entirely by listening to earthquakes. This essay follows the layers from the crust to the center, then traces how a century of seismology worked out what lies below. It ends with the recent finding that made headlines: the inner core appears to have slowed, paused, and begun drifting backward relative to the mantle. The physics behind that headline is real. The drama attached to it in popular retellings is mostly not, and a few simple calculations show why.

The Layers, From the Surface Down

The Earth’s mean radius is 6,371 kilometers. The figures below are the standard values from seismic reference models; temperatures deep in the core carry uncertainties of several hundred degrees.

Layer	Depth (km)	State	Density (g/cm³)	Temperature (°C)
Crust	0 – 7 (ocean) to 0 – 35+ (continent)	Brittle solid	2.7 – 3.0	15 at surface to several hundred at base
Mantle	Base of crust to 2,890	Solid rock that creeps	3.3 – 5.6	About 1,300 near the top to 3,500 – 4,000 at the base
Outer core	2,890 – 5,150	Liquid iron-nickel alloy	9.9 – 12.2	About 4,000 – 5,400
Inner core	5,150 – 6,371	Solid iron-nickel alloy	12.8 – 13.1	About 5,200 – 5,500 at its surface

The crust

The crust is a thin shell, proportionally thinner than an apple’s skin. Under the oceans, it is basalt about 7 kilometers thick. Under the continents it is lighter granitic rock averaging about 35 kilometers and reaching 70 kilometers beneath high mountains. Together, the crust and the rigid top of the mantle form the lithosphere, roughly 100 kilometers thick. The lithosphere is broken into about a dozen large plates. Where they collide, separate, or grind past each other, most of the world’s earthquakes and volcanoes occur.

The mantle

The mantle holds about two-thirds of the Earth’s mass. It is solid rock, as shown by the fact that shear waves cross it freely. Over millions of years, however, hot solid rock creeps like a very stiff fluid. The mantle convects at a few centimeters per year, about the rate a fingernail grows. That slow overturning carries the plates along and returns old ocean floor to the depths.

The mantle is layered in its own right. Minerals change to denser crystal forms at depths near 410 and 660 kilometers. At the very bottom, just above the core, two continent-sized regions of unusually slow seismic speed sit: one under Africa and one under the Pacific. Their density differences matter later in this essay, because their gravity can tug on the inner core.

The outer core

At 2,890 kilometers, the rock ends abruptly, and liquid metal begins. The outer core is mostly iron with some nickel. It is about 10 percent less dense than pure iron would be at that pressure, so it must also contain lighter elements such as oxygen, silicon, sulfur, carbon, or hydrogen; the exact mix is still debated.

This liquid convects vigorously at speeds of roughly half a millimeter per second, about 100,000 times faster than the mantle. A moving conductor in a magnetic field generates electric currents, and those currents sustain the field. This self-sustaining dynamo produces the Earth’s magnetic field, which deflects much of the solar wind and cosmic radiation.

The inner core

At 5,150 kilometers, the iron becomes solid again. The temperature there is still above 5,000 °C, comparable to the Sun’s visible surface. The iron is solid not because it is cool but because the pressure, about 330 gigapascals or 3.3 million atmospheres, raises its melting point above the local temperature. The inner core has a radius of about 1,220 kilometers. That is about 70 percent of the Moon’s radius and roughly one-third of its volume, nearly the same size as Pluto.

The inner core is growing. As Earth slowly cools, iron freezes onto its surface at a rate of about a millimeter per year. Freezing releases latent heat and leaves the lighter elements behind in the liquid. Both effects stir the outer core and help power the dynamo.

Where the Earth’s heat comes from

Popular accounts often call the inner core the planet’s boiler room. That overstates it. The Earth loses about 46 terawatts of heat through its surface. Roughly half comes from the radioactive decay of uranium, thorium, and potassium, mostly in the crust and mantle. The rest is primordial heat left over from the planet’s formation, slowly leaking out. Only an estimated 5 to 15 terawatts cross from the core into the mantle. The core’s heat drives the dynamo and feeds the deep mantle plumes, but the plates are moved chiefly by the mantle’s own heat and by the sinking of cold, dense ocean slabs.

How We Know What Is Down There

The missing mass

By the late 1800s, scientists knew Earth’s average density from gravity measurements: about 5.5 grams per cubic centimeter. Surface rocks average about 2.7. Something inside had to be much denser than anything at the surface.

In September 1896, the German geophysicist Emil Wiechert, then 34, proposed an answer at a meeting of German naturalists and physicians in Frankfurt. He argued that compression alone could not squeeze rock to the needed density. Instead, the interior must hold a different, heavier material. His candidate came from the sky: iron meteorites, denser than Earth’s average and thought to be fragments of shattered planetary bodies. Wiechert proposed a metallic iron core surrounded by a stony shell, and he backed the idea with a calculation of how the planet’s mass and its moment of inertia would be shared between the two. The idea of an iron interior was not new, but Wiechert gave it a quantitative form just as a tool to test it was arriving.

Three kinds of waves

The tool was the seismograph. By the 1890s, instruments sensitive enough to record distant earthquakes were spreading across Europe and Asia. In June 1897, a great earthquake struck the Shillong Plateau in Assam, India. Richard Dixon Oldham of the Geological Survey of India studied its records from distant stations and found three distinct arrivals.

1. Primary (P) waves arrive first. They are compressional: the ground moves back and forth along the direction of travel. They pass through solids and liquids.
2. Secondary (S) waves arrive next. They are shear waves: the ground moves sideways to the direction of travel. A liquid cannot resist shear, so S waves cannot cross it.
3. Surface waves arrive last. They travel along Earth’s outer shell and do most of the damage.

Because the waves travel at different speeds, the gap between their arrivals grows with distance. That gap tells a seismologist how far away an earthquake was, and the paths the waves take tell him what they passed through.

A core appears

In 1906, Oldham published the argument that made the core a measured fact rather than an inference. Waves that traveled to stations on the far side of the Earth arrived later than they should have if the interior were uniform. He concluded that they had passed through a central core where waves travel more slowly, and he estimated its radius at about four-tenths of Earth’s. He did not say what the core was made of; he stated only what his data showed.

In 1913–1914, Beno Gutenberg, a student of Wiechert’s at Göttingen, used the P-wave shadow zone to fix the depth of the core boundary at about 2,900 kilometers. That figure has barely changed in more than a century. In 1926, Harold Jeffreys showed from the Earth’s tidal response that the core must be far less rigid than the mantle. Together with the absence of direct S waves beyond about 104 degrees from an earthquake, this established that the core is liquid.

The shadow zone and Inge Lehmann

A liquid core bends P waves sharply as they enter and leave it, the way a lens bends light. The result is a ring-shaped shadow zone, from about 104 to 140 degrees from the earthquake, where direct P waves should not arrive.

Inge Lehmann was chief of the seismological department of the Royal Danish Geodetic Institute. Her duties included maintaining the Danish and Greenland stations and publishing their bulletins; she did her original research on her own. Studying records of the large 1929 Murchison earthquake in New Zealand, she found faint P arrivals inside the shadow zone, where a fully liquid core allowed none. In a 1936 paper titled simply P′ (“P-prime”), she showed that these arrivals were explained if the core contained a smaller inner body in which waves travel faster. Its surface reflected and bent some energy into the shadow.

Lehmann’s inner core was confirmed by later seismology. In 1940 it was proposed to be solid iron, and in 1971 the Earth’s free oscillations, the slow ringing of the whole planet after great earthquakes, supplied evidence that the inner core resists shear. Lehmann lived to 104 and continued publishing into her nineties. Near the end of her life, she remarked that the first estimates of the inner core’s properties were rough and that the subject was far from closed.

The Cold War network

The next leap came from an unlikely source. To detect Soviet underground nuclear tests, the United States launched Project Vela Uniform in 1959. Part of that effort was the World-Wide Standardized Seismograph Network of the 1960s: about 120 identical, well-calibrated stations that openly shared their records. Telling a bomb from an earthquake required precise timing and wave shapes. That same precision, along with decades of archived records, later made it possible to look for changes in the inner core over time.

Does the Inner Core Keep Time With the Planet?

Why it could turn on its own

The inner core is a solid ball floating in liquid metal. Nothing bolts it to the mantle. Two influences act on it. The first is electromagnetic: the magnetic field and currents in the outer core exert a torque on the conducting iron ball, and mid-1990s dynamo computer models predicted this would spin it slightly faster than the mantle. The second is gravitational: dense regions in the mantle and matching bumps on the inner core pull toward alignment, tending to lock the two together. If these torques aren’t balanced, the inner core speeds up or slows down relative to the rest of the planet.

Every statement about inner-core rotation is a statement about this difference. The inner core turns once a day along with everything else. “Super-rotation” means it gains a fraction of a degree per year on the mantle. “Sub-rotation” or “backtracking” means it loses a fraction of a degree per year. It never stops spinning, and it never spins backward in absolute terms.

The grain of the inner core

Measuring that difference needed a landmark inside the inner core. One was found in the 1980s. In 1983, French seismologists noticed that P waves crossing the inner core along the polar axis arrived early. In 1986, two Harvard groups, Morelli, Dziewonski, and Woodhouse (using travel times) and Woodhouse, Giardini, and Li (using the Earth’s free oscillations), showed that the inner core is anisotropic. Waves running parallel to the spin axis travel about 3 percent faster than waves crossing the equatorial plane. (In the source video’s transcript, this word came through as “an isotropic,” which means the opposite.) The likely cause is a preferred alignment of iron crystals, like the grain in a board, though how that alignment formed is still debated.

The first claim: 1996

In 1996, Xiaodong Song and Paul Richards of Columbia University’s Lamont-Doherty Earth Observatory published the first seismic evidence of differential rotation. They used earthquakes in the South Sandwich Islands in the far South Atlantic, recorded at College, Alaska, near Fairbanks. That station had operated for about three decades, and its path to the South Sandwich Islands runs nearly pole to pole through the inner core.

The inner-core wave’s travel time, compared with a neighboring wave that skirts the inner core, shortened by about 0.3 seconds between the late 1960s and the mid-1990s. Song and Richards modeled the fast axis of the anisotropy as tilted roughly 10 degrees from the spin axis. If the inner core turned, the tilted fast axis would swing gradually into better alignment with this path, and the wave would speed up. Their best estimate was about 1 degree per year faster than the mantle, within a range of roughly 0.4 to 1.8 degrees.

Doubts and doublets

The claim drew immediate challenges. One was that small errors in locating the old earthquakes could produce the same 0.3-second drift. A French study comparing pairs of nearly identical earthquakes found no detectable change on that path and placed any rotation below about 0.2 degrees per year.

The answer to the challenge was the same tool. An earthquake doublet is a pair of earthquakes so close in location and mechanism that their seismograms are nearly identical. If a doublet’s waves match everywhere except where they crossed the inner core, location errors cannot be the cause; the inner core itself must have changed between the two events.

In 2005, Jian Zhang, Xiaodong Song, Paul Richards and colleagues published in Science an analysis of 18 South Sandwich doublets recorded at 58 stations in and near Alaska. For one key pair, a December 1993 event and a September 2003 event, the waves that missed the inner core were essentially identical. The inner-core waves arrived about a tenth of a second earlier in 2003 and had changed shape. They estimated a super-rotation of about 0.3 to 0.5 degrees per year. Later estimates drifted lower still, with some near 0.05 degrees per year, and some researchers proposed that the rate fluctuates on a yearly scale.

2023: a pause and a turning point

In January 2023, Yi Yang and Xiaodong Song of Peking University reported in Nature Geoscience an analysis of repeating earthquakes going back to the 1960s. Paths that had changed steadily before about 2009 showed little change afterward. They concluded that the inner core’s differential rotation had nearly halted around 2009 and was turning back, as part of an oscillation with a period of about seven decades and a previous turning point in the early 1970s.

They noted that the Earth’s day length and the geomagnetic field both vary with a period of roughly 60 to 70 years. They proposed that the oscillation reflects electromagnetic and gravitational torques taking turns, and that the planet’s layers, from the inner core to the surface, are coupled. Song projected that, if the model holds, the inner core would keep rotating slower than the surface until about the mid-2040s. Not everyone agreed on the period. For example, Hrvoje Tkalčić of the Australian National University favored a 20- to 30-year cycle.

2024: retracing the same path

In June 2024, Wei Wang, John Vidale and colleagues at the University of Southern California published in Nature a different approach. Instead of asking how much the waves had changed, they looked for times when the waves had changed back. If the inner core oscillates, it should return to positions it occupied years earlier, and earthquakes from those different years should produce matching inner-core waves.

They compiled 121 South Sandwich earthquakes from 1991 to 2023, forming 143 repeating pairs, recorded on two seismic arrays: Eielson in Alaska and Yellowknife in northern Canada. The matches showed that the inner core super-rotated from 2003 to 2008, then sub-rotated from 2008 to 2023 back along the same path, two to three times more slowly than it had advanced. This is direct evidence of backtracking. The study did not by itself measure a 70-year period, as the source video implies; it confirmed the reversal and showed that the forward and backward rates differ, which current models do not explain.

2025: the surface may be deforming

In February 2025, Vidale and colleagues reported in Nature Geoscience that some of the same data cannot be explained by rotation alone. For certain pairs, the inner core was in the same orientation at both times, yet waves that grazed its surface still differed. The best explanation is that the shallow part of the inner core changed shape, probably by slow viscous flow driven by the turbulent outer core. Only one location showed the effect clearly, between 2004 and 2008. If confirmed, it means the outer core disturbs the inner core on a human timescale, and the picture of a perfectly rigid iron ball needs revising.

A Proportion Check

A few round-number calculations put the headlines in scale.

Wiechert’s missing mass. The core’s radius is 3,480 kilometers, 55 percent of the Earth’s, so it fills only 16 percent of the planet’s volume. If the mantle and crust averaged 4.45 grams per cubic centimeter, which is about right once deep compression is included, the core must average about 11 to bring the whole Earth to 5.51. That matches the seismic value of about 10.9, and it puts roughly one-third of the Earth’s mass in the core. If instead the mantle were as light as surface rock, the core would need a density near 17, more than iron can reach even at core pressures. Wiechert was right that compression alone could not do it, and right that the answer was iron.

The deepest hole. The Kola Superdeep Borehole reached 12.26 kilometers in 1989. That is 0.19 percent of the way to the center, or about 0.4 percent of the way to the core. The core is known only through waves and calculation.

How far the inner core moves. The inner core’s surface lies 1,221 kilometers from the axis. At its equator, one degree of arc is about 21 kilometers.

Differential rate	Motion at the inner-core equator
1° per year (Song and Richards, 1996)	About 21 km per year, 58 m per day
0.3–0.5° per year (Zhang et al., 2005)	About 6–11 km per year
0.1° per year	About 2 km per year, 6 m per day
0.05° per year	About 1 km per year

Even the largest estimate is a creep compared with the Earth’s own spin, which carries the inner core’s equator about 7,700 kilometers each day.

The effect on the length of day. The inner core’s moment of inertia is about 5.8 × 10³⁴ kilogram-meters squared. The whole Earth’s is about 8.0 × 10³⁷, so the inner core holds about 0.07 percent of the planet’s rotational inertia. Suppose the inner core’s differential rotation of 0.1 degree per year were entirely handed to the mantle. The day would change by about 0.05 milliseconds. At 1 degree per year, the change would be about 0.5 milliseconds. Observed decade-scale swings in the length of day are a few milliseconds, and most of that is attributed to exchanges with the much larger liquid outer core. The inner core’s share is a small fraction of an already tiny effect.

The mantle’s pace. At 5 centimeters per year, a piece of mantle rock would need about 58 million years to sink from the surface to the core. The inner core’s few-kilometers-per-year drift is fast by mantle standards and slow by any human standard.

Should We Worry?

No. If the 70-year oscillation is real, it has very likely been running through all of human history, with no effect anyone noticed until seismologists went looking. The changes it correlates with are a fraction of a millisecond in the day and slow drifts in the magnetic field that geophysicists already track. Yang and Song also noted parallels with multidecadal swings in global mean temperature and sea level. A correlation of that kind does not show a cause, and the sizes involved are far too small to account for the warming of the past century.

The real value of the finding is scientific. The inner core is the only part of the deep Earth whose motion we can watch change within a few years. Each turn of the cycle tests our models of how the core, the dynamo, and the mantle are coupled.

The Open Questions

The seismic observations are now solid. What they mean is not settled. Five questions remain.

Rotation or reshaping. Some researchers have long argued that the changing seismograms come partly or wholly from changes at the inner core’s surface rather than rotation of the whole ball. The 2025 USC result shows that both happen. How to separate them along each path is unresolved.

The period. Estimates range from about 6 years, from length-of-day and gravity signals, through 20 to 30 years, to about 70 years. The record of high-quality repeating earthquakes covers only about 60 years, less than one full cycle of the longest proposal. A clean test may require waiting until the 2040s.

The driving torques. Electromagnetic coupling to the outer core and gravitational coupling to the mantle are the leading candidates, but neither model yet predicts why the inner core backtracked two to three times more slowly than it advanced.

The inner core’s age and makeup. Estimates of when the inner core first froze range from about half a billion to more than a billion years ago. The light elements in the core, which control its melting point and the dynamo’s power, are still not pinned down. There is also seismic evidence for a distinct “innermost inner core,” roughly half the inner core’s radius, with a different crystal grain.

Geography of the data. Nearly all the key evidence comes from one source region, the South Sandwich Islands, and stations in Alaska and northern Canada. Other paths are needed to show that the behavior is global and not a feature of one patch of the inner core.

Notes on the Source

This essay began with an Astrum Earth video narrated by James Stewart. The video tells the story well, but several statements need correction.

• Depth. The inner core does not lie 1,000 miles down. Its surface is about 5,150 kilometers (3,200 miles) deep, and the center is about 6,371 kilometers (3,960 miles) deep.
• Size. “Nearly as big as the Moon” overstates it. The inner core’s radius is about 70 percent of the Moon’s, its volume about one-third. Pluto is the closer comparison.
• The boiler room. The inner core is not the main source of the heat that drives plate tectonics. Most of the surface heat flow comes from radioactivity and primordial heat in the mantle and crust; see “Where the Earth’s heat comes from” above.
• Stopped and backward. The inner core has not stopped spinning or begun turning backward in absolute terms. Its rotation relative to the mantle slowed, paused, and reversed by a fraction of a degree per year.
• Lehmann. She studied the New Zealand earthquake in 1929, but her paper appeared in 1936, and its title was P′, not P. She worked at the Royal Danish Geodetic Institute, not a university.
• The missing step. The video omits Gutenberg’s 1913–1914 measurement of the core’s depth and Jeffreys’s 1926 demonstration that it is liquid, which are the two results between Oldham and Lehmann.
• The 2024 study. The USC paper confirmed the backtracking; it did not by itself establish the 70-year period.

Conclusion

The structure of Earth’s interior is one of physics’ quiet triumphs. A density mismatch noticed in the 1800s, three wiggles on a seismogram in 1897, and a few faint arrivals where none should have been in 1929 were enough to map a planet no one can visit. The same method, refined by Cold War instruments and decades of patient record-keeping, now shows that Earth’s solid iron heart drifts back and forth against the mantle and may even flex at its surface. None of this threatens anyone. It shows how much we can learn by taking small, careful measurements seriously, and how much remains to be learned at the center of our own world.

---

References

• W. Wang, J.E. Vidale, G. Pang, K.D. Koper, R. Wang, “Inner core backtracking by seismic waveform change reversals,” Nature 631, 340–343 (2024). doi:10.1038/s41586-024-07536-4 · open-access copy
• Y. Yang and X. Song, “Multidecadal variation of the Earth’s inner-core rotation,” Nature Geoscience 16, 182–187 (2023). Summary: Sci.News, January 2023; further comment in CNN coverage.
• J.E. Vidale et al., “Annual-scale variability in both the rotation rate and near surface of Earth’s inner core,” Nature Geoscience (2025), doi:10.1038/s41561-025-01642-2. USC news release, 10 February 2025
• J. Zhang, X. Song, Y. Li, P.G. Richards, X. Sun, F. Waldhauser, “Inner core differential motion confirmed by earthquake waveform doublets,” Science 309, 1357–1360 (2005). Columbia Climate School summary · Physics World
• X. Song and P.G. Richards, “Seismological evidence for differential rotation of the Earth’s inner core,” Nature 382, 221–224 (1996). Alaska Science Forum account
• Poupinet, Souriau et al., “The existence of an inner core super-rotation questioned by teleseismic doublets,” Physics of the Earth and Planetary Interiors. HAL archive
• “Inner core super-rotation,” Wikipedia (range of published rate estimates). link
• I. Lehmann, “P′,” Publications du Bureau Central Séismologique International, Série A, 14, 87–115 (1936).
• R.D. Oldham, “The constitution of the interior of the Earth, as revealed by earthquakes,” Quarterly Journal of the Geological Society 62, 456–475 (1906).
• A. Morelli, A.M. Dziewonski, J.H. Woodhouse, “Anisotropy of the inner core inferred from PKIKP travel times,” Geophysical Research Letters 13, 1545–1548 (1986).
• Astrum Earth (narrator James Stewart), video on the Earth’s inner core and its changing rotation; the starting point for this essay.