Earth’s Ancient Freeze: Limit Cycles That Pulsed Between Ice Worlds and Hothouses for 56 Million Years

Picture Earth as a glittering orb of ice, locked in perpetual winter for 56 million years. That’s the Sturtian glaciation, from roughly 717 to 660 million years ago. Standard models called it Snowball Earth—total freeze—or Slushball, with patchy open water. But neither fit the evidence. Glaciers lasted too long. Oxygen levels didn’t crash as predicted. Life hung on.

A fresh model changes everything. Published in the Proceedings of the National Academy of Sciences, it points to repeated swings: iceballs giving way to steamy hothouses, over and over. These ‘limit cycles’ explain the oddities. And they hinge on a massive volcanic outburst in the Canadian Arctic.

Silicate weathering acts as Earth’s thermostat. It pulls CO₂ from the air, cooling things down. During ice ages, that process stalls. Volcanoes keep pumping CO₂. Levels rise. Ice melts. But in the Sturtian, something kept resetting the clock.

Enter the Franklin Large Igneous Province. Emplaced around 717 million years ago, this basalt behemoth flooded the landscape. Its fresh rock weathered fast, sucking CO₂ and sparking the first freeze. “Enhanced weathering by LIPs has long been acknowledged as an important climate driver across geologic time. The Franklin LIP was emplaced at ∼717 Ma, essentially coincident (within 1 to 2 Myr) with the onset of the Sturtian, and could have provided a sufficiently large quantity of fresh basalt to draw down CO₂ and trigger a global glaciation,” the authors note in their PNAS paper, as reported by ScienceX.

Not all the basalt vanished in one go. Ice halted weathering. CO₂ built from volcanoes. Thaw came. Hot interglacials exposed the rest. Weathering resumed. Freeze again. Cycle. Repeat. Until the province eroded away. “If only a portion of the Franklin LIP was weathered away during the initial Snowball onset, the remaining volume of basalt would still be available for weathering upon deglaciation, reinitiating CO₂ drawdown during the interglacial hothouse climate until another Snowball was triggered and the cycle repeated. This cycling, back and forth between climate extremes, would continue until Franklin’s weathering power (i.e. unweathered basalt) was exhausted,” the study explains.

One cycle? About 4 million years—matches the later Marinoan glaciation. Stack 14 of them? You get 56 million. Oxygen holds steady, too. No total depletion. Life persists in brief thaws.

Model’s Power—and Limits

Researchers built a coupled box model. It tracks climate, carbon, oxygen. They tweaked volcanic outgassing, weathering rates, LIP scale. Thousands of runs. Limit cycles popped out when Franklin entered the picture.

Simplified, yes. No full ocean currents or biology. But it nails mismatches. “These mismatches between the predicted pCO₂ evolution and observed glacial duration, and between the predicted pO₂ evolution and observed isotopic and biological records, motivate alternative solutions to the Neoproterozoic glaciation problem,” the team writes.

Harvard’s Paulson School of Engineering echoed this in a recent release. Their take: Earth oscillated, not stagnated, reconciling Sturtian ice with physics, as detailed here.

And recent work builds on it. A March 2026 paper in Earth System Dynamics uses MPI-ESM1.2 to probe Snowball transitions from Last Glacial Maximum conditions. Clouds speed the freeze, cutting needed forcing. Key: transition hits at 0°C global mean—caps climate sensitivity at 6.2°C for doubled CO₂, since no Snowball in ice ages. “It is therefore possible to estimate an upper limit of climate sensitivity from the models with stable LGM states… we find it is implausible that Earth’s climate sensitivity exceeds 6.2 °C (3.9–8.4 °C, 5 %–95 % confidence),” per the study in ESD.

February’s Earth and Planetary Science Letters found dynamic climates even in Snowballs—tropical cycles like El Niño in rocks from Garvellach Islands. Open seas? Maybe. Chloe Griffin’s team reports evidence of active weather, challenging full shutdowns, covered by Science News.

Subglacial weathering prolonged some Snowballs, per a March Earth and Planetary Science Letters study from Tokyo’s ELSI. Ice didn’t stop all chemistry. CO₂ drawdown continued below sheets. “Our results demonstrate that subglacial weathering represents a previously unrecognized feedback mechanism that could account for the dramatically different durations of Neoproterozoic snowball Earth events,” lead author Shintaro Kadoya told Phys.org.

These threads weave together. Cycles. Feedbacks. No single freeze.

Why care now? Exoplanets. Earth-likes around dim stars risk Snowballs. Cyclic models predict habitable windows in the chaos—oxygen flickers, life endures. Modern climate? Echoes in tipping points. Albedo flips. Weathering feedbacks. But Earth’s sun is brighter now; another full Snowball unlikely.

Still. Cycles teach resilience. Earth didn’t stay frozen. It pulsed. Thawed. Lived.