The Gold Standard of Archaeology | Phytolith Redundancy, Redeposition and Plant-Use Attribution

1. Begin with the particle, not the plant name

Back to contents

Phytoliths form when silica precipitates in or around plant cells. Their shapes can reflect cell type, tissue and plant group. The first observation is therefore a morphotype described under a stated nomenclature, not a crop species written directly into the catalogue.

The International Code for Phytolith Nomenclature provides a shared descriptive language. Consistent names improve comparison, but nomenclature does not remove biological ambiguity. Taxonomic attribution needs reference collections and an argument at assemblage level.

2. Redundancy and multiplicity pull in opposite directions

Back to contents

Redundancy occurs when different taxa produce similar morphotypes. Multiplicity occurs when one taxon produces several morphotypes. A common shape may therefore be uninformative alone, while a pattern of shapes, proportions and anatomical connections can be more diagnostic.

Do not count every particle as an independent species indicator. State which forms are non-diagnostic, which combinations narrow attribution, and whether the relevant reference material covers the taxa and plant parts plausible at the site.

3. Sampling defines the assemblage you can see

Back to contents

A hearth lens, floor, vessel fill and background sediment answer different questions. Record the sampled unit, volume or mass, coordinates, stratigraphic relation, extraction procedure, counted area and number of identifiable particles. Include blanks and duplicate counts where appropriate.

Bulk extraction can quantify and rotate particles for identification, but it removes microscopic context. Thin sections preserve distribution and articulation but can hide features because particles are cut at random angles. The method should follow the inference.

4. Reference collections need local and anatomical coverage

Back to contents

A reference plant should be identified, sampled from known organs and processed transparently. Leaves, stems, inflorescences and husks can produce different spectra. Soil or modern-surface assemblages help describe background inputs and dispersal.

Report whether the comparison is with a single specimen, several individuals or a published database. A close match to one modern specimen supports resemblance; it does not establish uniqueness across all plants that might have contributed.

5. Redeposition changes the meaning of presence

Back to contents

Phytoliths can enter sediment through plant decay, ash, dung, construction material, erosion or imported soil. Trampling, bioturbation and later cultivation can fragment and move them. A particle in a floor may therefore come from the floor material, activity on the floor or older sediment incorporated into it.

Look for articulated phytoliths, clusters, orientation, preservation, weathering and association with ash or other microremains. Thin-section context can distinguish some primary deposits from dispersed or reworked inputs, although rarely with absolute certainty.

6. Worked example: a floor with many grass forms

Back to contents

A room floor yields 240 identifiable phytoliths. Ninety are common elongate forms, 70 are rondels, 30 are dendritic forms, 10 are possible sedge forms and 40 are other or uncertain. Bulk samples suggest cool-season grasses, but elongates and rondels are redundant across many taxa.

A thin section shows a small articulated dendritic cluster within a compact floor lens, while most particles are isolated and fragmented. The bounded claim is that grass-derived material, including inflorescence tissue, contributed to the floor and that much of the assemblage was dispersed or reworked. A species-level crop claim would need a suitable reference comparison and stronger diagnostic evidence.

7. Combine proportions, context and independent proxies

Back to contents

Assemblage proportions can distinguish contexts only when recovery and counting are comparable. Compare floor, hearth, vessel and background samples, and report uncertainty from unidentified or poorly preserved particles.

Starch, pollen, macroremains, chemistry, micromorphology and use-wear can support or challenge a plant-use interpretation. Agreement among proxies is useful when their formation pathways differ; repeated measurements of the same pathway do not create independent confirmation.

8. Changed-context transfer check

Back to contents

Move the floor example to phytoliths recovered from dental calculus. Morphological ambiguity and reference coverage still matter. The depositional route changes: particles may derive from eaten plants, dust, processing or handling, while archaeological sediment contamination must be assessed.

A room-level activity claim cannot be transferred unchanged to an individual diet claim. Reframe the question, document laboratory controls and use evidence appropriate to calculus formation.

9. Limits and common misconceptions

Back to contents

High phytolith concentration does not automatically mean intensive crop use. Plants differ in silica production, preservation differs among morphotypes, and extraction can alter counts. Absence may reflect low production, dissolution, inadequate sampling or identification limits.

Do not treat a morphotype name as a species, convert an assemblage directly into biomass, or call isolated particles in mixed sediment an in-situ activity surface. State the taxonomic and behavioural level that the evidence can support.

10. Audit the chain from slide to activity

Back to contents

Archive sample provenience, preparation steps, microscope settings, count rules, image vouchers, analyst decisions, reference comparisons and raw counts. Record uncertain and unidentifiable forms instead of excluding them from the denominator without explanation.

Recheck a blind subset and compare inter-analyst classifications. Then trace every behavioural statement back through taxon, tissue, deposition and context. A missing link should narrow the claim, not be filled by a familiar story.

11. Frequently asked questions

Back to contents

Can phytoliths identify a crop species?

Sometimes a diagnostic assemblage, articulated tissue or morphometric comparison can narrow attribution, but many forms are redundant. State the tested alternatives.

Are thin sections always better than bulk samples?

No. Thin sections preserve microcontext; bulk samples support rotation, identification and quantitative comparison. Their strengths are complementary.

Does burning destroy all phytolith evidence?

No. Heating can alter colour, fluorescence, preservation and dissolution risk. Interpret those changes with context and suitable experiments.

12. The gold standard is a bounded attribution chain

Back to contents

Strong phytolith analysis does not leap from a familiar shape to a familiar plant-use story. It shows how morphology, assemblage, reference material, taphonomy and context constrain one another.

Before naming a plant or activity, ask: could another taxon make this form, could the same taxon make other forms, and could the particles have arrived another way? The surviving claim should be the one that remains after all three questions.