How Old Is the Earth? Getting to 4.54 Billion Years

Before any physical method existed, the estimates came from scripture, most famously Archbishop Ussher's calculation in 1650 that creation occurred in 4004 BC, arrived at by summing genealogies and which was a careful piece of scholarship within its own assumptions. Naturalists then began arguing from physical processes. Buffon in 1774 heated iron spheres and measured their cooling, extrapolating to an Earth of about 75,000 years, which he suspected was too low and dared not say. Geologists from the late eighteenth century, above all Hutton and then Lyell, argued that the processes shaping the Earth operate slowly and steadily and that the observed thickness of sedimentary rock therefore required an immensity of time with, in Hutton's phrase, no vestige of a beginning. Estimates from sedimentation rates and from ocean salinity produced figures in the tens or hundreds of millions.

William Thomson, later Lord Kelvin, brought thermodynamics to the question in the 1860s and calculated how long a molten Earth would take to cool to its present temperature gradient. His answer was between twenty and four hundred million years, later narrowed to around twenty to forty, and he pressed it hard against geologists and against Darwin, for whom a short Earth was fatal since natural selection needs enormous time. Darwin was distressed by it and could not answer it. Kelvin's mathematics was correct and his premise was not, since he assumed no internal heat source, and radioactivity, discovered in 1896, generates heat continuously within the Earth, which invalidates the whole calculation. The episode is a standard example of a rigorous argument from a false assumption, and of the fact that a confident quantitative answer can be worse than an admitted uncertainty.

The method that settled it depends on a fact about radioactive decay: the rate is constant, unaffected by temperature, pressure or chemistry, and characterised by a half-life, the time for half a quantity to decay. Measuring the ratio of a parent isotope to its daughter product in a sample therefore gives the time since the system closed:

• •Uranium-238 decays to lead-206 with a half-life of 4.47 billion years, and uranium-235 to lead-207 with a half-life of 704 million years, which is a useful pair because the two run at different rates and cross-check each other
• •Potassium-40 to argon-40, at 1.25 billion years, widely used on volcanic rock
• •Rubidium to strontium and samarium to neodymium for very old material
• •Carbon-14, at 5,730 years, which is useless for geology and essential for archaeology, and which dates organic material up to around fifty thousand years
• •The method requires that the sample was a closed system, losing and gaining nothing, which is checked by dating the same rock with several independent decay chains and seeing whether they agree
• •Zircon crystals are the favoured mineral for the oldest work, because they incorporate uranium and reject lead when they form, so any lead present must be from decay, and they survive extreme heating

Claire Patterson, working on his doctorate at Chicago and then at Caltech, set out in the late 1940s to date the Earth using lead isotopes in meteorites, on the reasoning that meteorites are leftover material from the formation of the solar system. The obstacle was that his samples kept being contaminated with lead from the environment, at levels that swamped the signal, and he eventually established that industrial lead was everywhere: in the air, the dust, the water, the glassware and the reagents. He built what amounted to the first ultra-clean laboratory to get around it, and published his figure of 4.55 billion years in 1956, which has been refined and not overturned. The contamination finding then redirected his career entirely. He measured lead in ocean sediments and in Greenland ice cores, showed that concentrations had risen enormously since the introduction of leaded petrol in the 1920s, and spent two decades in public conflict with the lead industry, which funded opposing research and attempted to have him removed from advisory panels. His evidence contributed directly to the removal of lead from petrol and paint, which is among the largest public health measures of the century.

The figure is supported from several independent directions, which is why it is quoted with confidence. Meteorite dating using multiple decay systems agrees. Lunar samples returned by the Apollo missions give ages up to about 4.5 billion years, unaffected by the erosion and plate tectonics that reset terrestrial rocks. The oldest surviving Earth rocks are around four billion years old, and individual zircon crystals from Western Australia have been dated to 4.4 billion, which means liquid water and a solid crust existed remarkably early. Models of stellar and planetary formation, the age of the oldest stars in the galaxy, and the expansion age of the universe are all consistent with it. The remaining uncertainty is not about the order of magnitude but about the precise moment being dated, since planetary accretion took tens of millions of years and the question of when the Earth counts as having formed is partly a matter of definition.

The Earth is 4.54 billion years old, a figure obtained from meteorites rather than from terrestrial rock because the surface has been recycled. Early estimates came from scripture, cooling rates and sedimentation, and Kelvin's thermodynamic calculation of twenty to forty million years was rigorous and rested on the false assumption that the Earth has no internal heat source, which radioactivity supplied. Radiometric dating uses constant decay rates cross-checked between independent isotope systems, and Claire Patterson obtained the modern figure in 1956 after building the first ultra-clean laboratory.