The Gold Standard of Archaeology | Loss-on-Ignition, Organic Matter and Carbonate Interpretation

1. Start with the archaeological question

Back to contents

Decide whether loss-on-ignition is being used to compare occupation deposits, identify organic-rich features, characterise buried soils, screen sediments for further analysis or support a palaeoenvironmental sequence. An LOI percentage is not an activity label. Define the context, expected contrast and competing formation processes before selecting temperatures or sample spacing.

2. Understand the sequence of measurements

Back to contents

A common sequence dries the sample to constant mass, ignites it at a lower temperature to estimate organic matter, then heats it more strongly to estimate carbonate from carbon-dioxide loss. Exact temperatures and durations vary among protocols. Report each weighing basis and formula. Values produced at different settings are not automatically comparable, even when both are called LOI.

3. Control mass, geometry and furnace conditions

Back to contents

Small samples are sensitive to balance error; large or deep samples may heat unevenly. Use clean pre-weighed crucibles, consistent sample mass and grind or homogenise only when compatible with the research question. Record furnace ramp, hold time, cooling procedure and crucible position. Cool in a desiccator before weighing because hot or rehydrating samples can shift mass. Run duplicates and reference material across furnace loads.

4. Separate estimates from chemical identity

Back to contents

Mass lost at the organic-matter step is not pure carbon, and carbonate loss is not a direct mineral identification. Bound water may be released, organic material may remain, and carbonate can begin decomposing below the chosen upper step depending on mineralogy and grain size. Use terms such as LOI-estimated organic matter and LOI-estimated carbonate. If mineral identity or carbon concentration matters, confirm with methods such as elemental analysis, calcimetry, spectroscopy or diffraction.

5. Build comparisons within one matrix and protocol

Back to contents

LOI is most defensible for relative patterns among comparable sediments processed together. Parent material, clay content, salinity and modern contamination can change response. Include local natural deposits, archaeological backgrounds and known reference layers where available. Randomise or balance samples across furnace runs so a site phase is not confounded with a particular batch.

6. Worked example: a dark floor deposit beside a hearth

Back to contents

A dark layer yields 14% mass loss at the organic step, while nearby construction sediment yields 4%. The hearth ash yields 9% at the organic step and a much larger high-temperature loss. Duplicate floor samples agree within 0.4 percentage points, and micromorphology shows fine organic residues mixed with trampled mineral sediment. The team concludes that the floor is relatively enriched in combustible material; it does not convert 14% directly into carbon or call the deposit domestic waste. The ash demonstrates why colour and one LOI step are insufficient.

7. Read stratigraphic patterns before isolated numbers

Back to contents

Plot LOI with depth, context boundaries, grain size and other proxies. A gradual rise may reflect soil development or changing sediment source; a sharp peak may mark a lens, contamination or a tiny heterogeneous inclusion. High-resolution sampling can expose short events but also increases sensitivity to local fragments. Preserve enough material for repeat or independent analysis because ignition destroys the tested portion.

8. Changed-context transfer check: a shell-rich midden

Back to contents

A midden sample loses substantial mass at the high-temperature step. The transferable principle is to interpret mass loss through matrix and independent evidence. Here shell carbonate dominates, but burning, dissolution and mixed sediment may alter the result. A standard carbonate calculation derived for homogeneous sediment cannot establish shell abundance, species or depositional history. Sieve records, shell counts and mineralogical checks are needed.

9. Know the common misconceptions

Back to contents

A standard furnace setting does not make results universal. LOI is not synonymous with organic carbon, and multiplying by a fixed conversion factor can conceal variation in organic composition. A high-temperature loss is not uniquely calcium carbonate. More decimal places do not improve accuracy when furnace conditions or sample heterogeneity dominate. Agreement between duplicates tests repeatability of that aliquot, not representativeness of the archaeological context.

10. Audit the chain from context to balance

Back to contents

Archive context and depth, sampling rationale, storage, pretreatment, particle-size decision, wet and dry masses, crucible IDs, balance precision, furnace program, run layout, duplicates, blanks and every intermediate weighing. Calculate percentages from retained raw masses with scripted formulas. Test sensitivity to outliers, batch effects and alternative thresholds, and state where sample consumption limits verification.

11. Frequently asked questions

Back to contents

What temperature should be used? Follow a validated protocol suited to the matrix and report it exactly; there is no context-free universal setting. Can LOI compare different sites? Only cautiously when preparation, temperatures, duration and sediment properties are compatible. Is the test quantitative? It produces quantitative mass loss, but composition estimates remain method- and matrix-dependent. Can the sample be reused? Usually not for analyses affected by heating, so subsampling must be planned.

12. The gold standard reports what burned, not what was imagined

Back to contents

Strong loss-on-ignition work combines controlled weighing with a formation question, comparable matrices and independent checks. It calls the outputs estimates, shows the thermal protocol and retains the raw mass chain. The gold standard uses a simple method without giving it a simple meaning: archaeological interpretation begins after the balance reading, not before it.