Ice Cores

Polar ice is a stack of individual winters. Snow falls, is buried by the next season's snow, compresses into firn and then into solid ice, and does not melt. Each year's layer is still there, in order, with air from that year sealed inside it as bubbles.

That makes an ice sheet an archive with two properties that are rare in combination: it can be counted year by year like tree rings, and it physically contains samples of the ancient atmosphere. The deepest cores reach back roughly 800,000 years continuously, and discontinuous "blue ice" samples from Antarctica have been dated beyond a million.

Reading a Core#

An ice core is a cylinder, typically about 10 cm across, drilled in sections and reassembled into a continuous depth record. Several independent properties vary with the seasons, and layers are identified by looking for all of them to agree.

Visual stratigraphy#

Summer snow has larger crystals and different bubble structure than winter snow. In the upper part of a core, layers are visible to the eye against a light table.

Stable isotope ratios#

The ratio of oxygen-18 to oxygen-16, and of deuterium to hydrogen, in precipitation depends on the temperature at which the water condensed. Summer and winter snow carry different signatures, producing a sawtooth that can be counted. The same measurement, averaged over longer spans, is the basis for reconstructing past temperature.

Dust and chemistry#

Continental dust, sea salt, and sulfate arrive seasonally. Electrical conductivity and laser dust measurements record annual peaks that can be counted independently of the isotopes.

Melt layers#

In coastal and Greenland cores, occasional summer melt produces a distinct clear band. These are useful markers but are absent in the coldest interior sites.

The counting is done by multiple parameters at once, and the published chronologies report where the parameters disagree and by how much. The Greenland GICC05 timescale, built by counting annual layers in several cores, carries an explicit uncertainty that grows with depth — about 2% of the age at 60,000 years, which is stated openly rather than glossed over.

The Independent Fixed Points#

Layer counting on its own would be vulnerable to accumulated error. What makes the chronologies robust is that they contain dated events from outside the ice.

Volcanic eruptions. Large eruptions deposit a sulfate spike worldwide. Historically dated eruptions — Tambora in 1815, Laki in 1783, Krakatau in 1883, Vesuvius in 79 CE — appear as sulfate peaks at exactly the counted depth. In several cases the volcanic glass in the layer has been chemically matched to a specific volcano.

Radionuclide events. Cosmic ray bombardment produces beryllium-10 in the atmosphere, which is deposited in ice. Sharp global spikes in 774–775 CE and 993–994 CE, first identified in tree rings, appear in Greenland and Antarctic ice at the corresponding counted years. This is a single-year marker shared between two completely unrelated archives.

The bomb spike. Atmospheric nuclear testing left tritium and beta-radioactivity peaks in 1954 and 1963 that show up at the expected depth in every core.

Orbital tuning. Below the countable region, chronologies are constrained by matching climate cycles in the ice to Milankovitch orbital cycles, which are calculable from celestial mechanics with no reference to the ice at all.

The Depth Records#

• NGRIP and GISP2 (Greenland): annual layers counted to roughly 60,000 years, with the core reaching about 123,000 years.
• EPICA Dome C (Antarctica): 3,270 metres of core reaching roughly 800,000 years, covering eight full glacial cycles.
• Allan Hills blue ice (Antarctica): discontinuous ice dated by argon isotopes to beyond 2 million years, not a continuous record but a direct sample of atmosphere from that period.

Dome C matters for a specific reason: the record contains eight glacial-interglacial cycles, each with the same characteristic shape and each spaced at the interval Milankovitch theory predicts from orbital mechanics. That periodicity is not something a counting error produces.

The Trapped Air#

The bubbles are the part that is difficult to reinterpret. They are not a proxy for ancient atmosphere; they are ancient atmosphere. Extracting and measuring them gives carbon dioxide and methane concentrations directly.

Over 800,000 years, atmospheric CO₂ oscillates between roughly 180 ppm during glacial maxima and 280–300 ppm during interglacials, tracking the temperature record derived independently from isotopes. It does not exceed about 300 ppm anywhere in that record. The present value is above 420 ppm, and the isotopic signature of that additional carbon identifies it as fossil in origin.

Common objections#

"The Lost Squadron planes were buried under 75 metres of ice in 50 years"

The standard young-earth objection, used by Answers in Genesis and the Institute for Creation Research. Six P-38 aircraft force-landed on the Greenland ice sheet in 1942 and were recovered in 1992 under roughly 75 metres of ice. If 75 metres accumulate in 50 years, a 3,000-metre core represents only about two thousand years.

The measurement is accurate. The inference applies it to the wrong place.

The aircraft landed near the coastal margin of southeastern Greenland, where snowfall is heavy — several metres of accumulation per year — and the ice is actively flowing and thickening. Deep cores are deliberately sited at ice divides in the dry interior, where accumulation is a few centimetres per year and horizontal flow is minimal. That is why those sites are chosen. Dome C receives about 2.5 cm of ice-equivalent accumulation annually.

The two locations differ by roughly two orders of magnitude in accumulation rate, and the difference is measured directly at both.

"Layers could form more than once a year"

Individual storms can deposit snow distinguishable from surrounding layers, so a single parameter could in principle be miscounted.

Counting requires several independent parameters to agree — visual stratigraphy, oxygen and hydrogen isotope ratios, dust, and chemical conductivity, which respond to different seasonal processes.

The check is the dated events. A systematic multiple-layers-per-year error would place Tambora at the wrong depth, and it does not. The historically dated eruptions and the 775 CE beryllium-10 spike land at their counted years.

"Deep layers are too thin to count, so the deep ages are assumed"

Correct, and the published chronologies say so. Layers thin under compression with depth, and below a certain point annual resolution is lost.

This is why deep chronologies are not extended by counting alone. Below the countable range they rely on flow modelling, gas-phase correlation between cores, and orbital tuning, and the reported uncertainty grows accordingly — GICC05 states about 2% of the age at 60,000 years.

The counted portion alone already reaches roughly 60,000 years in Greenland, which is an order of magnitude beyond a 6,000-year chronology before any modelling is involved.

"A post-flood ice age deposited the ice rapidly"

Michael Oard of the Creation Research Society has developed the most detailed version of this: a single ice age following the flood, driven by warm oceans and volcanic aerosols, depositing the ice sheets within a few hundred years.

The model has to account for what the cores contain. Dome C records eight complete glacial-interglacial cycles, each with the same characteristic shape, spaced at the intervals predicted by orbital mechanics — the Milankovitch periods of roughly 100,000, 41,000, and 23,000 years. A single rapid event does not produce eight cycles at astronomical spacing.

It also has to explain the trapped air. Carbon dioxide oscillates between about 180 ppm and 300 ppm in phase with the isotopic temperature record across all eight cycles, and the volcanic and radionuclide markers appear at their historically documented years.

"The two poles are dated separately, so agreement proves nothing"

Greenland and Antarctic chronologies are built independently by counting.

They are then matched using globally mixed gases, particularly methane, which equilibrates between hemispheres within a couple of years. The synchronization is a check on both records rather than an input to either, and it uses a physical property of the atmosphere rather than an assumption about age.