Earth’s deepest tug-of-war changes the length of a day

A day can gain or lose a few milliseconds without anyone missing lunch. Yet those imperceptible changes give scientists something extraordinary: a way to investigate material thousands of kilometres beneath our feet, where no instrument can visit.

Earth does not rotate at one perfectly steady rate. Winds and ocean currents move angular momentum around, rather as a skater changes a spin by repositioning their body. The Moon’s tides gradually alter the rotation over far longer periods. But after researchers account for seasonal atmospheric and ocean effects, long-term lunar friction, changes associated with ice loss and other known contributions, a slow pattern remains: the length of the day shifts by several milliseconds across decades.

The changes themselves are observations. The International Earth Rotation and Reference Systems Service publishes daily values for Earth’s orientation, including the rotation-based time UT1 and length of day, in a record extending back to 1962.

The explanation is less direct. A study published in Nature compared the residual day-length pattern from 1964 through 2019 with reconstructions of activity deep inside Earth. Its conclusion is that gravity linking the inner core and mantle most plausibly drives the broad rise and fall, while electromagnetic and pressure-related forces resist it.

Small enough to ignore, precise enough to investigate

For scale, take an illustrative three-millisecond change. One ordinary day contains 86,400 seconds, so the fraction is easy to reproduce:

0.003 ÷ 86,400 ≈ 0.000000035

That is about 35 parts per billion. It will not disturb a commute or alter sunrise by any noticeable amount. Scientifically, however, the direction and timing of the change matter more than its size.

Researchers have known for decades that exchanges of angular momentum between the core and mantle can produce decadal changes in rotation. The harder question was how those layers transfer the necessary twisting force, or torque.

The new study tested three candidates. Electromagnetic coupling can arise where moving, electrically conducting fluid in the outer core interacts with conducting material near the mantle’s base. Topographic coupling describes pressure acting around irregularities at the core–mantle boundary. Gravitational coupling occurs when the inner core’s orientation shifts out of alignment with mass variations in the mantle, allowing gravity to pull the system back.

The telling result was not merely that one modeled force was large enough. Models of electromagnetic and topographic torque could also reach the required magnitude, but their multidecadal changes were broadly opposite in phase to the observed pattern. Changing their assumed strength did not repair that timing mismatch. The modeled gravitational torque matched both amplitude and phase much more readily.

In the resulting picture, the gravitational interaction leads the dance. Electromagnetic or topographic coupling pushes back.

Evidence assembled from indirect views

Nobody watched the inner core turn. Its estimated motion comes from earthquake waves that followed nearly identical paths through Earth at different times. A 2023 seismic analysis found that paths showing earlier changes had changed little during the preceding decade. Comparison with records reaching back to 1964 supported an inferred turning-back of the inner core as part of an approximately seven-decade variation.

The latest researchers combined such seismic estimates with models of outer-core flow derived from changes in Earth’s magnetic field. They sampled different plausible inner-core histories and hundreds of possible flow realizations, then asked which combinations reproduced the measured day-length curve. An independent account of the work likewise reports that gravity emerged as the largest modeled influence, with the other mechanisms opposing it.

That chain of evidence deserves careful labels. The changing length of day is measured. Seismic waves and magnetic-field changes are also observed. Inner-core rotation and fluid-core flows are reconstructions from those signals. The proposed balance of torques is a model-based explanation that depends on the accuracy of both reconstructions.

The same distinction applies to the study’s deeper suggestions. Its best-fitting models are consistent with a deformable inner core, a weak mineral phase near the mantle’s bottom and, if electromagnetic resistance is important, a thin, electrically conducting, iron-enriched layer there. Those are constraints and tentative supports—not samples, images or direct detections. The authors report that changing the underlying core-flow model can shift best-fit torque parameters by as much as 30 percent.

Even so, the central achievement is wonderfully concrete. Earth’s slightly uneven turning becomes an instrument: a planet-sized dial whose movements help discriminate among competing accounts of an otherwise unreachable interior. A few milliseconds do not change an ordinary day. They change what that day can reveal.