Radiometric Dating
Isochrons, concordia plots, and cross-method agreement address the standard objections to radiometric dating.
11 min readUpdated
Radiometric dating measures elapsed time from the ratio of a radioactive parent isotope to the daughter isotope it decays into. This page covers how the methods work, how they check themselves, what the results are, and the specific objections raised against them.
The physics
Unstable isotopes decay at a rate determined only by the identity of the nucleus.
| Parent | Daughter | Half-life | Source |
|---|---|---|---|
| Uranium-238 | Lead-206 | 4.47 billion years | Jaffey et al., 1971 |
| Uranium-235 | Lead-207 | 704 million years | Jaffey et al., 1971 |
| Potassium-40 | Argon-40 | 1.25 billion years | Renne et al., 2010 |
| Rubidium-87 | Strontium-87 | 48.8 billion years | Rotenberg et al., 2012 |
| Samarium-147 | Neodymium-143 | 106 billion years | Kossert et al., 2009 |
| Carbon-14 | Nitrogen-14 | 5,730 years | Godwin, 1962 |
Given the current parent-to-daughter ratio and the decay rate, elapsed time follows. Everything else is controlling the assumptions that feed into it.
Decay rates have been tested against temperature, pressure, chemical state, and magnetic and electric fields. For the alpha and beta decays used in geochronology, no laboratory manipulation has produced a change beyond a fraction of a percent (Emery, 1972), and the theory explains why: these decays are governed by nuclear forces essentially unaffected by conditions outside the nucleus.
One class of exception exists. Electron capture decay can shift very slightly under extreme ionization, because it depends on the electron cloud (Ohtsuki et al., 2004). The exception is understood and quantified rather than unexplained.
How the assumptions are tested
Three assumptions are required: known initial conditions, constant decay rate, and a closed system. Each has a method for checking it.
Isochron dating removes the initial-conditions assumption
Rather than measuring one sample, several minerals from the same rock are measured, each with different original chemistry (Nicolaysen, 1961). Daughter isotope is plotted against parent isotope, both normalized to a stable isotope of the same element as the daughter.
If the rock remained closed and began with a uniform initial ratio, the points fall on a straight line. The slope gives the age. The y-intercept gives the initial daughter concentration, which means it is measured rather than assumed.
If samples were contaminated, the system leaked, or initial ratios were not uniform, the points scatter and no line forms. A failed isochron reports its own failure rather than returning a confident wrong answer.
Concordia detects lead loss
Zircon crystals incorporate uranium during formation and reject lead almost entirely, so lead inside a zircon accumulated after the crystal formed. Zircons also contain two independent uranium decay chains running at different rates: U-238 to Pb-206, and U-235 to Pb-207.
Plotting one against the other, closed systems fall on a curve called concordia (Wetherill, 1956). A crystal that has lost lead falls off the curve along a straight line, and where that line intersects concordia recovers both the original crystallization age and the timing of the disturbance.
This is two clocks in one crystal, with a built-in diagnostic for the specific failure mode most often raised as an objection.
Independent methods on the same rock
Uranium-lead, potassium-argon, rubidium-strontium, and samarium-neodymium involve different elements, different chemistry, different mobilities, and half-lives spanning two orders of magnitude. They share no assumptions about initial conditions or element mobility.
On well-behaved samples they routinely agree within one or two percent (Min et al., 2000). No single systematic error accounts for agreement across methods with unrelated assumptions.
Results
Clair Patterson (1956) determined the age of the Earth from lead isotope ratios in the Canyon Diablo meteorite combined with oceanic sediment representing average terrestrial lead, obtaining 4.55 billion years ± 70 million.
The measurement required lead contamination control far beyond what existed at the time, which forced Patterson to build the first ultra-clean laboratory. In the process he documented how thoroughly industrial lead had contaminated the environment, and spent much of his later career opposing leaded gasoline.
The current value is 4.54 billion years, within Patterson's original error bars. Individual meteorites dated by multiple independent chronometers give 4.567 billion years, and calcium-aluminium inclusions in the Allende meteorite, thought to be the oldest solids in the solar system, give the same figure.
Validation against known ages
| Event | Documented date | Radiometric result |
|---|---|---|
| Vesuvius eruption | 79 CE | Argon-argon: 1,925 ± 94 years before measurement, against 1,918 actual (Renne et al., 1997) |
| Hekla eruption | 1947 | Correct within error |
| Hawaiian historic flows | 1800–1959 | Dalrymple (1969) tested 26 historic flows: most returned ages indistinguishable from zero, but the 1800–1801 Hualalai flow gave an anomalous excess-argon age of roughly 1 million years, traced to xenoliths carrying inherited argon |
| Taupo eruption (New Zealand) | 232 ± 10 CE, from Chinese and Roman records of a reddened sky | Radiocarbon wiggle-matching against a New Zealand kauri tree-ring sequence: 232 ± 10 CE (Hogg et al., 2012) |
| Italian historic lava flows (Etna, Vesuvius, and others) | Documented eruption dates, several hundred years to a few decades old | Unspiked Cassignol–Gillot K-Ar technique reproduces the historical ages to within a few percent (Gillot & Cornette, 1986) |
| Egyptian dynastic chronology | Reign dates from historical king-lists, roughly 2700–1000 BCE | Radiocarbon dates on short-lived plant material match the documented dynastic sequence (Bronk Ramsey et al., 2010) |
Common objections
"Mount St. Helens rock dated to hundreds of thousands of years"
Steve Austin of the Institute for Creation Research submitted dacite from the 1980 lava dome for potassium-argon dating and reported ages from 340,000 to 2.8 million years, publishing the result in 1996 as evidence the method fails.
The measurements are real. The problem is the method's stated range.
Potassium-argon dating of a ten-year-old sample asks the instrument to measure an argon signal near its detection floor. At that level the reading is dominated by argon inherited from older crystals carried up in the melt — xenocrysts — which the dacite contained. The laboratory Austin used specifies a minimum age for reliable results, and the samples were well below it.
Austin also did not separate the mineral phases, which is standard practice precisely because different minerals retain argon differently.
Submitting a sample outside a method's stated range and reporting the result as a refutation does not test the method. The same technique returns correct ages for Vesuvius and Hekla, where the samples are within range.
"A Hawaiian lava flow from 1801 dated to over a million years"
Real, and reported by Dalrymple (1969) himself, not by a young-Earth source — the paper is a study of the failure mode, not an attempt to hide it. Whole-rock samples from the 1800–1801 Hualalai flow gave a potassium-argon age of roughly 1 million years, and the result is cited by the Institute for Creation Research as evidence the method is unreliable.
The flow carries abundant ultramafic xenoliths — fragments of older mantle rock torn loose and carried up by the magma, each already holding argon accumulated over millions of years before the 1801 eruption. A whole-rock measurement averages the young groundmass together with that inherited argon.
Removing the xenoliths and dating the groundmass alone gives an age indistinguishable from zero, matching the historical date. The anomaly is diagnosed, not explained away: it is the same inherited-argon failure mode as the Mount St. Helens dacite above, and it is why standard practice separates mineral phases rather than dating whole rock.
"Decay rates were faster in the past"
The central proposal of the RATE project (Radioisotopes and the Age of the Earth), an eight-year effort by the Institute for Creation Research and the Creation Research Society, published in 2005 by Vardiman, Snelling, and Chaffin. It argues that decay was accelerated by a factor of roughly a billion during the creation week and the Flood.
The proposal has a difficulty its own authors identified and published. Compressing 4.5 billion years of decay into a few thousand releases the same total energy over a far shorter period. The RATE team calculated the result and found it sufficient to raise Earth's temperature far beyond the melting point of its crust — enough to vaporize the oceans and the planet's surface.
Their report acknowledges this as an unsolved problem and proposes that God removed the heat by an unspecified mechanism. That is a coherent theological position and it is not a physical solution.
Accelerated decay also conflicts with astronomical observation. Supernova SN 1987A, at about 168,000 light years, showed a light curve powered by the decay of cobalt-56 with a half-life matching laboratory measurement. The light left before any proposed acceleration and shows the same decay constants.
"Helium in zircons shows the Earth is 6,000 years old"
Also from the RATE project. D. Russell Humphreys measured helium retained in zircons from the Fenton Hill borehole in New Mexico. Helium is produced by uranium decay and diffuses out of crystals over time; Humphreys argued the amount remaining implies about 6,000 years rather than 1.5 billion.
This is the most technically substantial young-Earth argument on this page, and it makes a real measurement rather than misapplying someone else's.
The critiques focus on the diffusion parameters. Kevin Henke and others have argued that the samples' thermal history was not adequately constrained, that the diffusion experiments were conducted under vacuum rather than under the pressure conditions the zircons experienced, and that the assumed temperature history of the borehole does not match the geological record.
The argument also has to answer the heat problem above, since it accepts that the decay occurred — it just compresses the timescale.
"Carbon-14 has been found in coal and diamonds, which should be too old to contain any"
Reported by the RATE project and repeated by Answers in Genesis. Carbon-14 has a half-life of 5,730 years, so nothing older than about 100,000 years should retain a detectable amount.
The measurements are generally around 0.1 to 0.5 percent modern carbon — at or near the detection floor of accelerator mass spectrometry.
Three known sources produce readings at that level. Instrument background: AMS machines have a floor around 0.05 pMC from ion source memory and sample handling. Contamination: coal and diamond samples pick up modern carbon during collection and preparation. In-situ production: uranium and thorium in surrounding rock emit neutrons that convert nitrogen-14 to carbon-14 within the sample, which is why coal seams near uranium deposits read higher.
The distinguishing test is that the reported values track sample handling and local radioactivity rather than geological age, which is what contamination and in-situ production predict and what a genuinely young age does not.
"Isochrons can be produced by mixing rather than by age"
Correct, and this is a real phenomenon documented in the geological literature — Andrew Snelling of Answers in Genesis has written on it at length. Two magma sources mixing can produce a linear array on an isochron plot that has no age meaning.
Geochronologists identify these. A mixing line produces a straight line on a plot of daughter ratio against the reciprocal of the normalizing isotope, which a true isochron does not. The test is standard and is applied before publication.
Mixing also does not explain agreement between methods. A mixing line in rubidium-strontium has no reason to coincide with the uranium-lead age of the same rock, yet the methods agree on well-behaved samples.
What the methods cannot do
Stating the limits matters, because misapplication produces most claimed failures.
- Sedimentary rock cannot be dated directly. Grains in sandstone record when those grains crystallized in an earlier source rock. Sedimentary sequences are dated by bracketing them between datable volcanic layers.
- Most fossils cannot be dated directly. They are dated from enclosing strata.
- Every method has a range. Carbon-14 is useless beyond about 55,000 years. Potassium-argon is unreliable below tens of thousands. Results outside the window are noise.
These limits are stated by geochronologists in the methods sections of their own papers. They are not the objections the young-Earth arguments above make.
What the evidence shows
The three standard objections — unknown initial conditions, variable decay rates, and contamination — are the three problems the methods were built to address, and each has a specific technique for detecting it: isochrons for initial conditions, concordia for lead loss, and cross-method comparison for systematic error.
The methods return correct ages for eruptions with documented dates. Independent methods with unrelated assumptions agree on the same samples. The age of the Earth has been stable at approximately 4.54 billion years since 1956.
The strongest contrary argument, accelerated decay, requires an amount of heat its own proponents calculate would melt the planet.
Independent confirmation from methods that do not involve radioactivity at all is in Dendrochronology and Ice Cores.