Isotopic provenance

How can a laboratory tell where a sample was grown or mined with no record of it?

A drum of honey arrives with a label naming a valley in New Zealand, a certificate from a broker, and a paper trail that anyone could have written. Suppose we throw all the paper away. A laboratory can still say, with real confidence, that this honey did not come from where the label claims — and can sometimes say roughly where it did.

That should bother us. Documentation is the only record anybody deliberately made. If we discard it, what is left to read? The answer is that the sample itself was keeping a record the whole time, in a script nobody chose and nobody can rewrite. The interesting question is not whether such a record exists but why it should be readable at all.

Start with what we can measure. We could measure which elements are present and in what amounts — but concentrations are exactly what processing changes. Filter the honey, blend two batches, boil off water, and the elemental profile shifts. A useful record has to be something that survives being handled.

So ask a sharper question: is there any property of a sample that transfers through the whole chain — rock to soil to water to plant to product — essentially unaltered? Consider the ratio of two isotopes of the same element. Isotopes differ only in neutron count; chemically they behave almost identically, so no ordinary process separates them. That "almost" is going to matter, but hold it for a moment.

Now, where would a difference between places come from, if isotopes barely separate? Two quite distinct sources, and it pays to keep them apart.

The first is radioactive decay running for geological time. Rubidium-87 decays to strontium-87 with a half-life on the order of fifty billion years. A granite that is both ancient and rubidium-rich has therefore accumulated an unusually high ratio of strontium-87 to strontium-86; a young basalt has not. The bedrock under a field carries a number set by its age and chemistry, that number passes into soil water, into the plants, and into the tissues of anything that eats them. It is a signature of place because geology is a map.

The second is mass. Evaporation and condensation do slightly favour the lighter molecule, so water vapour leaving the ocean is depleted in oxygen-18, and each subsequent rainfall depletes what remains. Rainwater therefore gets isotopically lighter with latitude, with altitude, and with distance inland — a smooth, well-mapped continental pattern.

Here is the step worth pausing on. Why does strontium sail through the food chain unchanged while oxygen is reshuffled at every puddle? Because the size of that mass effect depends on the relative mass difference. Strontium-87 against strontium-86 differ by about one percent; oxygen-18 against oxygen-16 by twelve. So strontium is inherited faithfully and oxygen is continually reset.

That gives us two kinds of tracer doing two different jobs, and confusing them is the classic error. The heavy radiogenic ratios record origin. The light ones record process and climate — which is still evidence, just about a different question. Carbon is the clean example: plants using the C3 pathway sit near minus twenty-seven parts per thousand, C4 plants such as maize and sugarcane near minus twelve. That does not locate a field, but it does reveal cane sugar stirred into honey, and it is how anti-doping laboratories separate pharmaceutical testosterone — synthesised from C3 plant sterols — from the hormone a body made itself.

One more thing to notice: no single ratio names a place. Each measurement only draws a contour across a reference map — an isoscape — built from samples of known origin, and a claim is only as good as that map. Two or three independent systems intersecting is what narrows a region, and even then the honest output is exclusion: this sample is inconsistent with these origins.

Two refinements keep this from being over-sold.

The first is that tissues are archives with different clock speeds. Tooth enamel mineralises in childhood and is never remodelled, so it reports where someone spent their first years. Bone turns over across roughly a decade and reports a later average. Hair grows about a centimetre a month, so a single strand can be sectioned into a timeline of movement. Choosing the wrong tissue answers a question you did not ask.

The second is that the geological signal can be drowned. What a plant takes up is the bioavailable strontium in soil water, which is not simply the bedrock value: windblown dust, fertiliser, irrigation water and sea spray all contribute. Modern seawater sits at about 0.7092 and is remarkably uniform worldwide, so a coastal site can be pushed toward that value regardless of what lies beneath it. This is why practitioners build local baselines from plants and small animals rather than reading a geological map and calling it done.

• How isoscapes are actually built, and what happens to confidence where sampling is sparse.
• Lead isotopes in archaeology and in ore tracing, where the source rocks are few and sharply distinct.
• Whether a determined counterfeiter could blend feedstocks to land on a target signature, and what that would cost.

Every term the collection defines is gathered in the glossary.