Robert Mulvaney, a glaciologist with the British Antarctic Survey, laid it out in 2004. The cornerstone of ice-core climate reconstruction is the ability to measure two things: past atmospheric greenhouse gas concentrations and past temperature. The gas part is direct. Tiny bubbles of ancient air sit trapped in deep ice. Extract them, run them through a mass spectrometer, you have the composition.
Temperature is different. It is not measured. It is inferred. The inference comes from the isotopic composition of the water molecules released when you melt the core. Water is H2O. But the atoms come in flavours. Oxygen-16 dominates at 99.76 percent. Oxygen-18, with two extra neutrons, makes up the rest. Hydrogen is 99.985 percent ordinary hydrogen, one proton, no neutrons. Deuterium carries a neutron.
All stable. No radioactive decay.
Mass spectrometers measure the ratios of these isotopes in ice-core samples and compare them to a standard called SMOW — Standard Mean Ocean Water. The ice is always depleted in the heavier isotopes. The depletion is expressed as δ18O or δD. Both values tell the same story: less 18O and less deuterium during cold periods.
The technical reality is a fractionation cascade. It takes more energy to evaporate heavy-isotope water from the ocean surface. As that moist air travels poleward and cools, the heavy molecules rain out first. What reaches the ice sheets is progressively lighter. The colder the condensation temperature, the greater the depletion. That is the signal written in the ice.
As ever, the detail that matters is the calibration. The isotope ratio is not a thermometer. It is a relationship established through statistical work and cross-checked against other proxies. The cryosphere has yielded cores stretching back hundreds of thousands of years. The combination of ice, sediment, tree rings, and speleothems builds reconstructions that far exceed the instrumental record.
The devil remains in the calibration.
Sources
- COP31 – State of Climate Update 2026 press conference statement
- How are past temperatures determined from an ice core? | Scientific American
