Radioactive Decay & Counting Statistics

T½ = ln 2/λ · σ ≈ √N · count, subtract, date

N = N₀e^(−λt) · σ ≈ √N · count, subtract, date

No one can say when one nucleus will decay.
A billion of them keep perfect time.

Every nucleus of an isotope has the same small chance of decaying each second, however long it has already lasted, so no one can predict which goes next. Yet half of any large sample is gone after one half-life, a quarter after two, as steadily as a clock. A counter catches only some of the decays, and it clicks even with no source at all. Count twice and the numbers differ, by about the square root of the count, which tells you how long you must count before a rate is believable. The same arithmetic dates a campfire from its charcoal, and says when a sample is too old to date.

Live: 400 nuclei of protactinium-234m (half-life 1.159 min, shown sped up). Each turns gray when it decays, at random; the counter clicks, and the survivors’ curve (amber) follows the smooth law beneath it (dashed). Every loop is a new sample.

Random nuclei, a predictable sampleNo one can say when one nucleus will decay, but half of a large sample goes in each half-life: N = N₀e^(−λt), with T½ = ln 2/λ.
What a counter countsA counter sees a fraction of the decays, R = εA, on top of a background that must be measured and subtracted.
Counts scatterA count C is uncertain by about √C, so precision costs time: four times the counting halves the relative uncertainty.
Measure and dateFit a half-life from counts on a log plot, and date a sample from the carbon-14 left, with an honest uncertainty, or say it is too old to date.
Stay in touch