The Gold Standard of Archaeology | Reservoir Effects, Dietary Mixing and Human Bone Chronologies

1. Begin with what bone collagen actually dates

Back to contents

A radiocarbon measurement on collagen estimates the age of carbon in the sampled tissue. It does not directly stamp the day of death, burial or monument construction. The relationship is usually useful because collagen turns over during life, but it still reflects dietary protein integrated over years.

Write the target event separately from the dated material: death, deposition, grave construction or use phase. Then explain the association that connects the collagen measurement to that event.

2. Recognise the reservoir mechanism

Back to contents

Terrestrial plants exchange carbon with the atmosphere, while marine and some freshwater systems can contain older dissolved carbon. Organisms feeding in those systems may therefore appear radiocarbon-old relative to contemporaneous terrestrial material. A person who consumes them can inherit part of that offset.

The size of the effect is not universal. It depends on the reservoir, food web, place, period and the fraction of dietary carbon routed into collagen. “Marine diet” is a warning label, not a correction value.

3. Join sample identity, diet and local baseline

Back to contents

Build one record that connects sample ID, skeletal element, collagen quality, context, stable isotopes, faunal baseline and candidate reservoir correction. Stable carbon, nitrogen and sometimes sulfur isotopes inform dietary sources, but they do not by themselves produce a calendar correction.

Prefer local, contemporaneous evidence. Modern fish or a published average may not represent the archaeological water body or food web.

4. Use short-lived terrestrial material where possible

Back to contents

If the question can be answered with secure, short-lived terrestrial material from the same event, that sample often avoids dietary offset altogether. Seeds, annual plant remains or articulated terrestrial herbivore bone may provide a stronger chronological anchor than human collagen.

The phrase “same context” still needs scrutiny. Residual charcoal, curated objects or intrusive material can be older or younger than the burial. Association quality and carbon-source suitability are separate tests.

5. Treat dietary mixing as a model, not a fact

Back to contents

Mixed diets require assumptions about source end-members, trophic enrichment, routing and reservoir age. Several mixtures can sometimes fit the same isotope values. Carry plausible alternatives into the chronological model instead of selecting one unreported correction.

A rough teaching approximation is offset ≈ dietary reservoir fraction × local reservoir offset. It is useful for intuition only. Real calibration must represent uncertainty in the fraction, offset, calibration curves and laboratory result.

6. Worked example: the burial that looks too early

Back to contents

A human rib returns a conventional radiocarbon age that appears about 250 years older than charred grain sealed in the grave fill. Collagen quality is acceptable. Stable isotopes support a mixed terrestrial and freshwater diet, while fish bone from the local river indicates a substantial but variable freshwater offset.

The analyst compares three models: fully terrestrial, a modest aquatic contribution and a larger contribution. The first conflicts with the secure short-lived grain; the two mixed models place death later but with different ranges. The result is not “subtract 250 years.” It is a set of calibrated scenarios tied to explicit dietary and local-reservoir evidence.

The report dates the burial primarily from the secure terrestrial sample and uses the human bone as a sensitivity case. The apparent disagreement becomes evidence about diet and model dependence rather than an excuse to choose the preferred number.

7. Use perfect pairs carefully

Back to contents

A perfect pair combines contemporaneous short-lived materials from different reservoirs in a secure closed context, such as human and articulated terrestrial herbivore bone deposited together. Their radiocarbon-age difference can test for a dietary offset.

The pair must truly be contemporaneous and independently suitable. A displaced animal bone, long-lived wood or loose deposit cannot carry the same diagnostic weight.

8. Changed-context transfer check: a coastal cemetery

Back to contents

Move the method to a coastal cemetery where marine food is expected. A generic global marine correction is still insufficient because local ΔR varies. The team samples terrestrial fauna, marine fauna and short-lived plant remains from secure contexts, then models individual diets with isotope evidence.

The principle transfers: define the target event, identify the carbon pathways and propagate local uncertainty. The numerical correction changes; the evidence architecture does not.

9. Know the limits of correction

Back to contents

Stable isotope end-members can overlap, freshwater effects can vary sharply within one catchment and diet changes over a lifetime. Bone elements integrate different periods, preservation may differ and a Bayesian chronology cannot repair a misidentified sample or an unjustified reservoir model.

Do not assume all high nitrogen values imply one aquatic fraction, apply a universal freshwater offset or quote a corrected range without the uncorrected result and assumptions. Precision produced by a narrow prior is not new evidence.

10. Audit the chronology as a chain of decisions

Back to contents

Trace each reported range back through sample identity, pretreatment, quality indicators, conventional age, isotope values, dietary model, local reservoir evidence, calibration curves and chronological constraints. Archive the code or settings used for mixtures and sensitivity runs.

Re-run the chronology under plausible alternative diet fractions and reservoir distributions. If the historical conclusion changes, report that dependency prominently.

11. Frequently asked questions

Back to contents

Does a reservoir effect make human bone unusable?

No. It makes source and model assessment essential; secure terrestrial material may still be preferred for the event date.

Can stable isotopes provide the correction directly?

Not alone. They constrain dietary models that also require end-members, routing assumptions and local reservoir evidence.

Will Bayesian modelling solve the discrepancy?

Only if the model represents the real carbon pathways and associations. Strong mathematics cannot rescue a wrong sample model.

12. The gold standard dates evidence, not wishes

Back to contents

A defensible human-bone chronology begins with the carbon pathway. It uses local baselines, treats dietary mixing as uncertain and tests whether conclusions survive plausible reservoir models. Where a secure terrestrial sample can answer the question, it may be the cleaner anchor.

Before accepting a date, ask what tissue was measured, what diet supplied its carbon and which evidence supports the correction. If any link is hidden, the calendar precision is premature.