The needle drags across the drum. In a timber-framed station near Yellowknife, a seismometer captures a pressure wave that left the South Sandwich Islands minutes earlier, plunged through nearly 12,000 kilometres of rock and molten iron, and grazed the surface of a sphere no human eye will ever see. That trace, almost identical to one captured a decade before, tipped John Vidale and his team to a discovery: the planet's innermost heart is not just turning. It is also warping.
The finding, published in February 2025 in Nature Geoscience, lands in a debate that has rattled through seismology for two decades [1]. "More is changing than just the inner core position; the soft outermost inner core probably deforms," Vidale wrote in the accompanying research briefing, and that single line is what the paper is now remembered for [2]. The new analysis comes down on a third answer to the long-running argument. Both things are happening, at once.
A two-decade record written in waves
Behind the paper sits a dataset almost old-fashioned in its patience. Vidale's team analysed 121 repeating earthquakes at 42 locations near the South Sandwich Islands between 1991 and 2024 [1]. Waves called PKIKP, the ones that punch straight through the planet and skim the inner core's outer skin, arrived at two medium-aperture arrays: ILAR, near Eielson in interior Alaska, and YKA, near Yellowknife [3]. Two steady receivers in the north give the team a control built into the planet's own geometry. Their earlier 2024 Nature paper had already shown the inner core's rotation is not the smooth drift older textbooks describe, and the 2025 paper extends that finding in a way that turns the discussion on its head [3].
The smoking gun came from the Yellowknife path between 2004 and 2008. After accounting for known rotation, the waves should have looked like the older ones. They did not. Even when the core had rotated back to the same orientation relative to the mantle, the waveforms had changed [1]. Fairbanks followed the expected pattern, but Yellowknife did not, and the Yellowknife path grazes the inner core's surface [4]. To isolate the cause, the team compared each core-crossing waveform against a reference phase that misses the inner core entirely. When the reference is steady and the deep path keeps drifting, the cause has to be down there, in the few hundred kilometres of iron that form the inner core's outermost shell. Twenty years of identical sources and steady receivers can separate signal from noise in the deepest, quietest part of the planet.
A deformable outer skin on the inner core
The inner core, a ball of iron about 2,400 kilometres across and 5,000 kilometres below our feet, is not uniform [5]. Pressures at the centre would crush any laboratory, yet the outermost layer, a thin skin just hundreds of kilometres thick, behaves differently. Co-author Guanning Pang of Cornell has put it plainly: maybe as soft as jelly [5].
That jelly sits right against the liquid outer core, the churning ocean of molten iron that drives the geodynamo, the dynamo that generates Earth's magnetic field. The outer core is well known to be turbulent, the kind of fluid motion you would see in a boiling pot. What no one had observed before is turbulence strong enough to leave a measurable fingerprint on the inner core's outer skin on a timescale of years, rather than geological eons [6].
The mechanism is mechanical. The inner core should remain in equilibrium relative to the mantle under gravitational and electromagnetic coupling, Pang explained. When disturbance occurs, the inner core offsets and oscillates. If the shallow inner core viscosity is low, it can deform [7]. The result is up to a kilometre or two of boundary displacement where the two layers rub together, and around a hundred metres of edge topography in places [8][9]. Vidale is blunt about the limits. "We're not 100 per cent sure we're interpreting these changes correctly" [8]. This matters because it gives geophysicists a handle on the viscosity of the inner core's outer shell, a number they have been guessing at for half a century, feeding directly into models of the geodynamo above.
Day length, magnetic jerks, and the deep Earth
Hidden in the data is a more public consequence. As the inner core wobbles and reshapes, it nudges the rest of the planet around with it. By conservation of angular momentum, when the iron sphere shifts, Earth shifts with it. Day length varies on a roughly six-year cycle, and one explanation is gravitational coupling between the inner core and lumpy features at the core-mantle boundary [3]. The new shape-change result sharpens that link, suggesting a deformation that would alter the length of a day by only milliseconds a year, but milliseconds that accumulate over eons [9].
A more intriguing target is the magnetic field itself. The geodynamo is generated in the turbulent outer core, and the inner core boundary is the dynamo's floor. If that floor is shifting shape on a human timescale, the patterns of the magnetic field above should respond. Vidale has pointed out that the magnetic field has had a series of sharp "jerks" in recent decades, visible in observatory records around the world. Whether those jerks have anything to do with the inner core boundary is, for now, an open question the new data may help to settle [8].
Xiaodong Song, a Peking University seismologist not involved in the study, published the influential 2023 paper on the inner core's multidecadal rotation that the Vidale team built on [10]. "After decades of research and debates, we are coming to an ever-clearer picture of the changing inner core," Song said [11]. Earlier studies had argued the changes were best explained by deformation, not rotation; the new picture incorporates both [11]. Deep-Earth seismology can now frame questions about the field as testable ones, with numbers attached.
The Australian angle, and what comes next
In Canberra, the news landed as part of an ongoing conversation. Prof. Hrvoje Tkalčić, Head of Geophysics at the Australian National University's Research School of Earth Sciences, called the new work an interesting concept worth exploring further [8]. On the school's project page, the inner core is a planet within a planet, a hot sphere of one hundred quintillion tons of iron and nickel about 5,150 kilometres beneath our feet, still waiting to be discovered. Modern global seismology, they write, is an inverted telescope for probing the Earth's deepest shell [12].
There are caveats. Vidale notes that, in all likelihood, this finding does not affect our daily lives one iota [8]. The 2025 paper does not say the rotation story was wrong, only that it is incomplete. The dataset is narrow, anchored in two arrays and one source region, and the contrast between paths is a clue for the next generation of studies. What the result does, regardless, is change the kind of question seismologists feel entitled to ask. For most of the modern era, the inner core was treated as a geological metronome on timescales of millions of years. The Vidale team is now arguing that the metronome is bending, the outer core tugging on it hard enough to write those bends into a seismic record. The deeper implication is methodological. If the inner core can be observed changing on a human timescale, it joins a short list of deep-Earth phenomena seismology can watch in something close to real time.