Gravitational tug-of-war between Earth's core and mantle drives decades-long shifts in day length, study finds

Deep beneath the planet's surface, a slow tug-of-war between Earth's solid inner core and its rocky mantle is the primary reason the length of a day drifts by milliseconds over the course of decades, according to a study published Thursday in the journal Nature. The research, led by physicist Huifeng Zhang and geophysicist Mathieu Dumberry at the University of Alberta, identifies gravitational pull between the core and mantle as the dominant force behind rotational speed changes that scientists have measured but not fully explained for roughly three decades.

The finding does not mean humans need to reset their clocks. The variations involved are measured in milliseconds accumulated over decades, far too small to notice in daily life. But the mechanism the researchers describe offers geophysicists a rare, indirect way to observe motion occurring more than 3,000 kilometers underground, in a part of the planet no instrument can directly reach.

A tug-of-war a few thousand kilometers down

Earth's rotation is not perfectly steady. Averaged over long periods, the day is gradually lengthening as the moon's gravity slows the planet down, adding roughly milliseconds per century. But superimposed on that gradual drift are shorter, choppier oscillations on the scale of years to decades, in which the day speeds up for a while and then slows down again. Scientists have tracked these multidecadal wobbles for about 30 years using precise astronomical and satellite measurements of Earth's orientation, of the kind compiled by the U.S. Naval Observatory's Earth orientation program, but the underlying cause has remained contested.

Zhang and Dumberry's study zeroes in on the boundary between two of Earth's least accessible layers: the solid inner core, an iron-nickel sphere about 70 percent as wide as the moon, and the mantle, the thick rocky shell more than 3,000 kilometers deep that surrounds the molten outer core. According to Nature's own news coverage of the paper, the inner core does not rotate in perfect lockstep with the rest of the planet. For decades at a time it turns very slightly faster or slower than the mantle above it, a difference of a fraction of a degree of longitude per year that seismologists have reconstructed by tracking how earthquake waves pass through the core over time.

Because the inner core is not a perfect sphere and the mantle's base has its own irregular lumps and density variations, that mismatch in rotation generates a gravitational pull between the two layers, tugging them back toward alignment even as they never quite catch up with each other. The researchers modeled this "gravitational torque" alongside two competing forces: electromagnetic drag, caused by the core's molten iron dragging on Earth's magnetic field, and topographic or mechanical coupling, caused by physical friction where the jagged base of the mantle meets the churning liquid outer core. When they compared different combinations of these three forces against six decades of observed length-of-day records, the model that matched best was the one in which gravitational torque did most of the work, resisted by the other two.

What the numbers show

The inner core's rotation relative to the rest of the planet appears to have changed direction around 2010, part of a pattern that researchers say has an oscillation period of roughly 30 years, in which the core alternately gains and loses ground against the mantle. Across the observational record analyzed in the study, spanning the early 1970s to 2021, these core-related effects shifted the length of the day by several milliseconds. Zhang and Dumberry's model reconstructs the corresponding push and pull between the core and mantle using seismic measurements of the inner core's differential rotation combined with a separate model of core fluid flow derived from changes in Earth's magnetic field.

The competition between the three torques is close enough that small shifts in the balance produce the kind of decade-to-decade variability that has puzzled researchers. As Zhang put it, describing the surprise of the result: "Before we obtained the result, we didn't know they are competing with each other," according to an account of the findings reported by Interesting Engineering.

"Even tiny changes in Earth's rotation can provide valuable information about processes occurring thousands of kilometers beneath our feet," Zhang said.

Decades of competing explanations

The question of what drives Earth's decadal rotation changes is an old one in geophysics. Researchers have known since at least the 1990s that the length of day fluctuates on multidecadal timescales in a way that correlates with changes in Earth's magnetic field, which originates in the churning liquid outer core. That correlation led many scientists to assume electromagnetic coupling between the core and mantle was the main driver, since a magnetic-field link between the two layers was easier to observe and model than the more obscure gravitational interaction.

Dumberry has spent much of his career studying core-mantle dynamics and their surface effects, including how core flow influences sea-level change and Earth's magnetic field. The new study builds on a 2023 seismological reconstruction of the inner core's rotation history, using earthquake-wave data to track how the solid core's spin has sped up and slowed down relative to the mantle since the 1960s. Combining that record with independent estimates of core-driven fluid flow allowed Zhang and Dumberry to test, for the first time in a single statistical framework, how much of the observed length-of-day signal each of the three candidate forces could explain on its own and in combination.

Dumberry, describing the parts of the mechanism that remain uncertain, said of the underlying viscous behavior of the inner core: "This is the part that's unclear," according to the same reporting. The inner core is thought to deform slightly under gravitational stress over a period of roughly a decade, a process that is not directly observable and that the researchers had to infer indirectly through their modeling.

Why it matters, and what comes next

The practical stakes of the finding are modest. Timekeeping authorities already correct for Earth's uneven rotation using leap seconds and similar adjustments, and the millisecond-scale variations described in the study are far too small to affect GPS navigation, telecommunications, or daily life. The significance is instead scientific: the inner core is the most physically inaccessible part of the planet, sealed beneath roughly 5,000 kilometers of rock and molten metal, and scientists have no way to sample or directly image its motion. Subtle changes in Earth's rotation, translated through careful modeling, are one of the few windows available into what is happening down there.

A better accounting of the gravitational, electromagnetic, and mechanical forces acting on the core-mantle boundary could sharpen models of the geodynamo, the process that generates Earth's protective magnetic field, and help explain other puzzling core-related signals, including geomagnetic jerks and shorter-period wobbles in Earth's orientation. The authors note that the statistical fit favoring gravitational torque does not rule out meaningful contributions from the other two mechanisms, and further seismological and magnetic-field data covering additional decades will be needed to test whether the roughly 30-year oscillation pattern holds over a longer stretch of Earth's history. Other coverage of the paper, including from The Debrief, notes that independent geophysicists are likely to test the model against alternative reconstructions of inner-core rotation as those datasets are refined.