sciencebriefs
13:00in productionCh. 1 · A frozen record read layer by layer/ 13:00 · ceiling 15 min
Earth & climate · Natural sciences

Ice core

Ice drilled from Antarctica now reaches back 2.7 million years, and the air bubbles trapped inside it are treated as a direct sample of ancient atmosphere, which this brief follows alongside the other proxies paleoclimatology relies on where ice cannot reach.

Ice cores are cylindrical samples drilled from polar ice sheets and glaciers, with the deepest reaching over three kilometres and containing ice as old as 800,000 years at sites such as EPICA Dome C, and isolated fragments retrieved from Antarctica's Allan Hills dated to roughly 2.7 million years. The brief covers how these cores are dated, chiefly by counting annual layers where possible and cross-checking against volcanic ash and known gas cycles where layers become too compressed to count, and what they reveal through trapped air bubbles, oxygen isotope ratios and dust content. It sets this within the wider field of paleoclimatology, which draws on tree rings, ocean sediment and coral records to extend climate reconstruction further back than ice can reach, and covers documented findings including repeated ice age cycles, the Medieval Warm Period and Little Ice Age, and unresolved questions such as the faint young sun paradox.

Chapters & takeaways6
  1. 0:08
    A frozen record read layer by layer

    Ice cores preserve annual snow layers that can be counted like tree rings, though this method breaks down at greater depths where the ice has been compressed and flattened by its own weight.

  2. 2:10
    Air bubbles as a direct atmospheric sample

    Air trapped inside ice as it forms preserves an actual sample of ancient atmosphere, letting researchers measure past carbon dioxide levels directly rather than infer them indirectly.

  3. 4:20
    Ice reaching back 2.7 million years

    While continuous cores such as EPICA Dome C extend to 800,000 years, isolated ice fragments recovered from Antarctica's Allan Hills in 2016 were dated to roughly 2.7 million years old.

  4. 6:30
    When the ice and the gas inside it disagree

    Gases can diffuse through the porous snow layer before it fully compacts into ice, meaning the ice at a given depth can be substantially older than the air trapped within it, an uncertainty that can exceed a thousand years at low-snowfall sites.

  5. 8:40
    Other proxies filling in where ice cannot reach

    Tree rings, ocean sediment cores and coral growth bands extend climate reconstruction beyond ice cores' reach, each proxy calibrated and limited in its own specific way.

  6. 10:50
    A record with real gaps still in it

    Questions such as the faint young sun paradox and the precise trigger for a rapid warming event fifty-five million years ago remain genuinely unresolved within the paleoclimate record.

Worth your time?

Yes. Study the whole thing.

4.5/ 5
What works
  • explains annual layer counting and its specific breakdown point at depth rather than treating dating as automatic
  • distinguishes ice age from gas age as a genuine, quantified source of uncertainty
  • gives specific, checkable ages for named ice core projects rather than a vague sense of how old the ice is
  • flags real unresolved questions like the faint young sun paradox rather than presenting the record as fully explained
What does not
  • resolve the faint young sun paradox or the cause of the Paleocene-Eocene Thermal Maximum
  • cover tree ring, sediment and coral proxy methods in the same depth as ice cores
  • explain the full mechanism behind Milankovitch cycles' influence on ice age timing
Study it if
  • anyone who wants to know how scientists actually measure ancient carbon dioxide levels
  • readers curious how ice core dating works and where it runs into trouble
  • people interested in the broader toolkit paleoclimatology draws on beyond ice
Skip it if
  • readers wanting a single definitive climate history timeline without the methodological detail
  • anyone looking for depth on any one proxy method beyond ice cores specifically
The written brief4 min read

A frozen record read layer by layer

Ice cores are cylindrical samples drilled straight down through ice sheets and high-altitude glaciers, and their basic logic is simple: snow falling each year buries the previous year’s layer, so depth corresponds to age, with the deepest documented cores reaching over three kilometres down and containing ice hundreds of thousands of years old. Dating relies first on counting these annual layers directly, since winter and summer snow compact differently and remain visibly distinct in central Greenland cores. That method has a hard limit, though: the material is explicit that flowing ice gradually thins and distorts these layers with depth, and at low-snowfall sites such as Vostok in Antarctica, direct layer counting becomes impossible beyond roughly 55,000 years, forcing researchers onto other dating methods, including matching known volcanic ash layers and correlating gas cycles with orbital patterns.

Air bubbles as a direct atmospheric sample

What makes ice cores distinctive among climate records is that they preserve an actual physical sample of ancient air, trapped in bubbles as falling snow compacts into ice. This lets researchers measure past atmospheric composition directly rather than inferring it from indirect evidence, and the material cites an early, striking result from this method: carbon dioxide concentration during the last glacial maximum measured about thirty percent lower than just before the industrial age began. Oxygen and hydrogen isotope ratios in the ice itself provide a separate temperature record, since the balance between heavier and lighter isotopes in snow varies with the temperature at which it originally formed, giving researchers two independent lines of evidence, gas composition and isotope ratio, drawn from the same physical sample.

Ice reaching back 2.7 million years

Several major drilling projects define how far back this record currently extends. Vostok, drilled by Soviet and later Russian teams from 1970 onward, reached ice around 420,000 years old at a depth of 3,310 metres. EPICA Dome C, a European project, reached bedrock at 3,260 metres in 2004, extending usable climate data to 800,000 years, the longest continuous ice core record cited. Beyond continuous cores, the material describes a genuinely striking outlier: fragments of ice recovered from Antarctica’s Allan Hills in 2016 were dated to approximately 2.7 million years old, by far the oldest ice yet dated from any core, though this ice does not form a continuous, year-by-year record in the way the deeper continuous cores do.

When the ice and the gas inside it disagree

A specific and important source of uncertainty concerns the difference between the age of the ice itself and the age of the air trapped inside it. Because gas can diffuse through the porous, unconsolidated snow layer, called firn, before that snow fully compacts into solid ice, the air actually trapped at a given depth is typically younger than the ice surrounding it, sometimes by a meaningful margin. At low-accumulation sites such as Vostok, the material states this uncertainty in the gap between ice age and gas age can exceed a thousand years, a limitation that matters directly for how precisely researchers can line up an atmospheric measurement with a specific point in climate history, and one the material treats as a genuine, quantified methodological challenge rather than glossing over it.

Other proxies filling in where ice cannot reach

Ice cores are only one tool within the broader field of paleoclimatology, which the material defines as reconstructing climate from before direct instrumental measurement existed, a record spanning barely a century and a half on its own. Tree rings offer year-by-year resolution for more recent centuries and, combined with radiocarbon dating of older wood, can extend several thousand years back. Ocean sediment cores preserve fossilised plankton and chemical signatures, including magnesium-to-calcium ratios in shells, that reconstruct past ocean temperatures over far longer timescales, while coral growth bands function similarly to tree rings for tracking sea surface conditions. Each of these proxies is calibrated and limited in its own specific way, and the material is clear that no single method covers the full span of climate history on its own.

A record with real gaps still in it

The material closes by naming genuine open questions within this broader record rather than presenting paleoclimate science as fully resolved. The faint young sun paradox, the puzzle of why early Earth remained relatively warm despite the sun emitting roughly thirty percent less radiation than today, is described as an unresolved tension pointing toward greenhouse gases or other mechanisms not yet fully pinned down. The precise trigger for a rapid warming event fifty-five million years ago, the Paleocene-Eocene Thermal Maximum, is likewise described as debated, with methane clathrate collapse proposed but unconfirmed. This honesty about specific, named gaps, rather than a tidy uninterrupted narrative, is what makes the material worth the time for a reader who wants the actual state of the evidence rather than a settled textbook summary.

Same field · Earth & climate4 of 47
Up next in Science

IceCube Neutrino Observatory

2010 · 12:59

Neutrinos don’t lie — and IceCube just made them point back at their makers.

12:59