Ancient rocks in Western Australia show tectonic plates were moving 3.5 billion years ago

In the red dust of Western Australia's Pilbara region, some of the oldest well-preserved rocks on Earth jut from a landscape that looks like it belongs on Mars. These formations date to the Archean Eon, when the planet was young, microbial life was just establishing itself, and asteroids still pummeled the surface with regularity. Geologists have been visiting this terrain for decades, searching its ancient minerals for clues about what the early Earth was really like.

A team from Harvard University has now extracted one of the biggest answers yet from those rocks. By analyzing the magnetic signatures locked inside more than 900 cylindrical cores drilled from over 100 sites across a feature called the North Pole Dome, they found direct evidence that tectonic plates were moving 3.5 billion years ago. The study, published March 19 in Science, pushes back the earliest confirmed record of plate motion and rules out a longstanding hypothesis that the early Earth's outer shell was a single, unbroken, immovable slab.

Drilling into 3.5-billion-year-old stone

The work was led by Roger Fu, Professor of Earth and Planetary Sciences at Harvard, whose team has been conducting fieldwork in East Pilbara since 2017. Fu specializes in paleomagnetism, a technique that reads the magnetic orientation frozen into minerals at the time a rock formed. Because the orientation of magnetic grains records both the direction of Earth's magnetic pole and the latitude where the rock sat, paleomagnetic data functions like a geological GPS system, revealing where a chunk of crust was positioned billions of years ago.

The fieldwork itself was painstaking. Researchers used an electric drill fitted with a hollow, diamond-toothed bit, cooled by a hand-pump garden sprayer, to extract cylindrical cores from rock formations spanning roughly 30 million years of geological history. After each core came out, the team inserted a compass and goniometer into the hole to record the sample's precise orientation. Then came the lab work.

Back at Harvard, each core was sliced into thin sections, arranged on trays, and placed inside a magnetometer capable of detecting magnetic signals 100,000 times fainter than what a compass needle produces. The samples were heated in incremental steps up to 590 degrees Celsius, progressively stripping away magnetic layers acquired at different points in the rock's history. The process, repeated thousands of times, took approximately two years.

"We took a really big gamble," said Alec Brenner, lead author of the study and now a postdoctoral researcher at Yale. "Demagnetizing thousands of cores takes years. And boy, did it pay off."

From 53 degrees latitude to 77 in thirty million years

The magnetic data told a clear story. Over a span of about 30 million years just after the 3.5-billion-year mark, part of the East Pilbara formation drifted from roughly 53 degrees latitude to 77 degrees, a shift of tens of centimeters per year. The same block also rotated clockwise by more than 90 degrees. Then, within about 10 million years, the motion slowed dramatically and entered a period of relative stillness.

That rate of movement is comparable to modern plate tectonics. Today, the North American and Eurasian plates are separating at roughly 2.5 centimeters per year. The Pilbara rocks suggest that whatever was driving crustal motion 3.5 billion years ago could push plates at similar or even faster speeds.

To check whether this motion was a local phenomenon, the team compared their Pilbara results to an Archean site in South Africa, the Barberton Greenstone Belt. Previous paleomagnetic studies had shown that the Barberton rocks sat near the equator and barely moved during the same time interval. Two regions, thousands of kilometers apart, exhibiting different patterns of drift. That's precisely what you'd expect if Earth's outer shell was broken into separate plates moving independently, not behaving as a single rigid lid.

Stagnant lid, sluggish lid, or something else

The question of when plate tectonics began has generated decades of debate, with proposed start dates ranging from shortly after Earth's formation 4.5 billion years ago to as recently as one billion years ago. Part of the difficulty is distinguishing between different models of how Earth's outer shell might have behaved.

Geophysicists have proposed several possibilities. A stagnant lid would mean Earth's lithosphere was a single unbroken shell, like the crust of modern Venus. A sluggish lid would involve slow, limited plate movement. An episodic lid would mean periods of motion interspersed with stillness. And modern-style plate tectonics, often called an active lid, involves continuous movement of distinct plates along defined boundaries.

The new data from Pilbara decisively rules out the stagnant lid model for 3.5 billion years ago. The crust was moving, and it was segmented. But Fu's team is careful to note that their findings don't tell us which type of mobile lid was operating. The pattern of rapid drift followed by a quiet interval could be consistent with either sluggish or episodic behavior. Additional studies are underway to narrow the possibilities.

"We're seeing motion of tectonic plates, which requires that there were boundaries between those plates and that the lithosphere wasn't some big, unbroken shell across the globe," Brenner said.

The oldest magnetic flip on record

Buried in the same dataset was a second finding: the oldest known geomagnetic reversal. This is the phenomenon in which Earth's magnetic field occasionally flips polarity, a reversal that would cause a compass needle to point south instead of north. The last such reversal occurred about 780,000 years ago, and the geological record shows hundreds of them over the past several hundred million years.

Finding a reversal at the 3.5-billion-year mark extends the known history of this phenomenon enormously. Fu noted that the data suggest reversals occurred less frequently in the Archean than in more recent geological time. Because reversals are driven by the dynamo action of convecting molten iron in Earth's core, a different reversal frequency hints that the core's behavior may have been in a different regime than today's.

"It's not by itself conclusive, but it suggests that maybe the dynamo was in a slightly different regime than today," Fu said.

What a moving planet means for life

The implications extend well beyond geophysics. Plate tectonics drives the carbon cycle, creates mountain ranges, opens and closes ocean basins, and generates the volcanic activity that recycles nutrients. It shapes climate on geological timescales and, by extension, the environments in which life evolves. If plates were already moving 3.5 billion years ago, that means these planetary-scale processes were influencing the conditions for early life far earlier than some models predicted.

The Pilbara rocks themselves contain evidence of some of the earliest known life on Earth: stromatolites and microbialite formations deposited by single-celled organisms like cyanobacteria. Whether plate motion helped create the conditions for those organisms, by cycling nutrients, creating shallow marine environments, or driving chemical gradients, is a question that links this geological work to astrobiology.

There are caveats. Because the magnetic pole occasionally reverses, the team cannot determine whether the observed drift occurred in the northern or southern hemisphere. And paleomagnetic measurements from rocks this old always carry uncertainty: 3.5 billion years of burial, heating, and deformation can overprint or obscure original signals. The team's extensive demagnetization protocol was designed to address this, but no ancient rock study is immune to these challenges.

Still, the scale of the dataset, more than 900 cores from more than 100 sites, gives the findings a statistical weight that smaller paleomagnetic studies from the Archean have lacked. The gamble, as Brenner put it, paid off.