An Embarrassing Fact
Humanity has sent probes 25 billion kilometers from Earth — to the edge of the Kuiper Belt. But the deepest we have ever drilled is just over 12 km. Soviet scientists spent nearly 20 years on the Kola Peninsula and reached 12,262 meters before giving up: temperatures in the deep crust were hot enough to soften drill bits, and pressure made rock flow like toothpaste.
25 billion kilometers outward, 12 kilometers inward. That is the gap in our exploration of "the far away" versus "right below our feet."
Earth's radius is 6,371 km. We have directly touched only 0.2% of it. The remaining 99.8% must be inferred indirectly — through seismic waves, magnetic anomalies, gravity variations, and laboratory simulations of high pressure and temperature.
And within that 99.8% of darkness, there are two continents.
Not the familiar continents of granite and basalt drifting atop the asthenosphere. These two lie 2,900 km underground, pressed against the core-mantle boundary, made of material that is hotter, denser, and chemically distinct from the surrounding rock. They have names: the one beneath Africa is Tuzo, the one beneath the Pacific is Jason.
They are about the size of continents on the surface.
The Discovery of Tuzo and Jason
The story begins with seismic waves — the most reliable messengers of Earth's interior. When a large earthquake strikes, waves penetrate the planet, refracting, reflecting, and changing speed at boundaries between the core, mantle, and crust. Global seismic stations record arrival times and intensities, letting scientists reconstruct the interior — like doctors using ultrasound to see internal organs.
By the late 20th century, as the global seismic network grew denser, an anomaly emerged: two regions at the very base of the mantle, just above the core, showed clearly slowed seismic waves. Shear wave speeds dropped by roughly 2–5%, over astonishingly large areas — each thousands of kilometers wide, together accounting for 3–9% of the mantle's volume.
They were named Large Low-Shear-Velocity Provinces (LLSVPs). Slow shear waves suggest material that is hotter, softer, or chemically different.
The two provinces sit nearly at Earth's antipodes: one beneath Africa, one beneath the Pacific. The African one is nicknamed Tuzo, honoring Canadian geologist J. Tuzo Wilson (a key contributor to plate tectonics who proposed the "Wilson cycle"). The Pacific one is Jason, honoring American geophysicist W. Jason Morgan (a founder of plate tectonics and mantle plume theory).
What Are They?
This is the most fascinating — and still unanswered — question.
Hypothesis 1: Primordial relics. When Earth formed 4.5 billion years ago from the solar nebula, heavy elements (iron, nickel) sank to form the core while lighter elements floated up to form the mantle and crust. But this separation was imperfect — some denser material may have gotten "stuck" at the base of the mantle, forming Tuzo and Jason. If so, they are fossils of Earth's infancy: they have sat there for 4.5 billion years, never stirred by mantle convection — older than any surface rock.
Hypothesis 2: Graveyards of ancient oceanic crust. Plate tectonics recycles oceanic crust: born at mid-ocean ridges, subducted at trenches. Most subducted crust is heated, mixed, and remelted in the mantle. But some may sink all the way to the mantle's base and pile up there. If so, Tuzo and Jason are the tombs of oceans long vanished — remnants of every ocean that opened and closed through supercontinent cycles.
Hypothesis 3: Both. The latest view leans toward LLSVPs being "thermochemical structures" — with both temperature and compositional differences. They may have a primordial core, progressively wrapped and accreted by subducted oceanic crust like a growing snowball.
Whatever the answer, one thing is certain: Tuzo and Jason are Earth's largest internal heterogeneities, two enormous obstacles in the mantle convection system — physical embodiments of some deep order in this planet's 4.5-billion-year history.
How They Shape the Magnetic Field
In February 2026, a University of Liverpool team published a study in *Nature Geoscience* revealing a new role for Tuzo and Jason: shaping Earth's magnetic field.
Earth's magnetic field is generated in the outer core — a ~2,260 km thick layer of liquid iron. Flowing liquid iron acts like a dynamo, generating currents and thus the magnetic field. This "geodynamo" is powered by latent heat and light elements released by inner-core crystallization, and by the core's cooling. The cooling rate depends on the temperature of the mantle directly above the core.
That is where Tuzo and Jason come in.
The Liverpool team's finding in one sentence: the temperature at the core-mantle boundary is not uniform.
Tuzo and Jason are hotter than the surrounding mantle. Directly beneath them, the top of the core is hotter than elsewhere. This temperature difference alters liquid iron flow: in cooler regions, iron churns vigorously, actively driving the geodynamo; in hotter regions (beneath Tuzo and Jason), buoyancy-driven flow weakens and stagnates.
Imagine a pot of boiling soup with two heat-insulating pads at the bottom — above those spots, the soup stops churning. Tuzo and Jason are two insulating pads beneath the geodynamo.
Combining paleomagnetic data (magnetic records preserved in ancient rocks) with supercomputer simulations, the team reconstructed the magnetic field's behavior over the past 265 million years. They found:
Some components of the magnetic field have remained relatively stable for hundreds of millions of years, while others changed dramatically.
The stable parts correspond to the locations of Tuzo and Jason — like two nails pinning certain magnetic features to fixed geographic positions. The changing parts correspond to regions between them, where liquid iron flows freely and magnetic patterns constantly reorganize.
The End of the "Perfect Bar Magnet"
This finding has a deeper implication: it overturns a long-standing assumption.
For decades, paleomagnetists assumed that if you average Earth's magnetic field over a long enough time (say, millions of years), it behaves like a perfect bar magnet — magnetic poles aligned with the rotation axis, a perfect dipole. This assumption is a cornerstone of paleomagnetism, underpinning reconstructions of ancient continental positions, plate motions, and paleoclimate zones.
But the Liverpool team showed this assumption may not be entirely correct.
Tuzo and Jason impose a long-term thermal asymmetry above the core, systematically distorting the field away from a perfect dipole — and this distortion is not random. It stays stable over hundreds of millions of years, anchored by the fixed positions of the two structures.
This means previously reconstructed ancient continental positions may need fine-tuning. Research depending on paleomagnetism — paleoclimate, paleobiogeography, natural resource distribution — could face cascading effects.
A rocky structure buried 2,900 km down can reshape our understanding of the surface climate 300 million years ago. This is the butterfly effect of Earth science — except the butterfly's wings are the size of Africa.
The Inverse Problem: From Effects to Causes
We cannot see Tuzo and Jason directly. We can only measure: seismic velocity changes as waves pass through them, subtle surface magnetic anomalies, tiny gravity deviations. Inferring the "cause" 2,900 km underground from these "effects" is a classic inverse problem.
Inverse problems are everywhere in science: doctors infer tumor locations from CT projections; astronomers detect exoplanets from tiny dips in stellar brightness (transit method); seismologists infer subsurface structure from waveform records.
AI researchers do inverse problems too. LLM interpretability research is essentially inferring a model's internal structure from its outputs — which neurons handle which functions, which activation directions correspond to which concepts.
The study of Tuzo and Jason offers a valuable paradigm: cross-validation from multiple independent data sources. Seismic waves reveal their shape and location; paleomagnetic data reveal their effect on the magnetic field; supercomputer simulations test whether those effects are physically self-consistent. Any single data source can be misread, but when three independent methods point to the same conclusion, confidence rises sharply.
This mirrors the "multi-probe" approach in AI interpretability: activation analysis, causal interventions, behavioral tests, and structural analysis all cross-validating the same conclusion. A single probe can deceive you; the probability that multiple independent probes all deceive you is far smaller.
Invisible Continents
Tuzo and Jason remind me of something: science's most thrilling moments often come not from seeing something new, but from realizing something always there had been overlooked.
Before continental drift was accepted, people looked at the matching coastlines of South America and Africa for centuries and saw only coincidence. Before plate tectonics, people knew earthquakes and volcanoes clustered in belts, but no one realized they were expressions of the same machine.
So it is with Tuzo and Jason. The seismic velocity anomaly appeared by the late 20th century, but interpreting it as continent-sized thermochemical structures — and then as long-term shapers of the magnetic field — required spanning seismology, mineral physics, geomagnetism, paleomagnetism, and computational geodynamics.
Each discipline saw only part of the elephant. The true picture emerges only when all the blind people piece their information together.
Epilogue: The Faraway Beneath Our Feet
Sometimes I wonder how long Tuzo and Jason have existed 2,900 km below. If they truly are primordial relics, they are older than the oldest surface rocks — older than the Jack Hills zircons of Australia (4.4 billion years), older than the Moon. They have watched Earth cool from a ball of magma into a planet, watched the atmosphere change from hydrogen and helium to nitrogen and oxygen, watched oceans form, life originate, every mass extinction and every great radiation.
And no one has ever touched them.
We send probes to Mars, to Jupiter, to the edge of the solar system, searching distant worlds for secrets. But Earth's greatest secret may have always been buried 2,900 km beneath our feet — two continent-sized bodies of hot rock, quietly shaping the magnetic field that protects all life on this planet.
25 billion kilometers outward, 12 kilometers inward. Perhaps the true faraway is not in the sky, but beneath our feet.
---
References:
- University of Liverpool (2026). "Scientists discover hidden deep-Earth structures shaping the magnetic field." *Nature Geoscience*.
- Garnero, E. J., McNamara, A. K., & Shim, S.-H. (2016). "Continent-sized anomalous zones with low seismic velocity at the base of Earth's mantle." *Nature Geoscience*, 9(7), 481-489.
- Shrestha, R. (2022). "LLSVPs: Mysteries in the Deep Mantle." EGU Geodynamics Blog.
- McNamara, A. K. (2019). "A review of large low shear velocity provinces and ultra low velocity zones." *Tectonophysics*, 760, 199-220.