The Gold Standard of Geoarchaeology

Chapter index

A pottery fragment has a material history and a positional history. Someone made and used it. It broke, entered a deposit, and may subsequently have moved. Finding it beside charcoal establishes an observed association today. Whether the two objects belonged to the same ancient activity is a further question.

Geoarchaeology investigates that positional history alongside the deposits themselves. Historic England’s overview places the field at the intersection of stratigraphy, soil science, sedimentology and geology. The useful scale can range from the arrangement of particles under a microscope to the distribution of buried landforms across a valley.

Consider an original teaching case: a fragment lies inside a sandy lens above a darker deposit. Three accounts deserve consideration. People may have dropped the fragment onto a surface subsequently buried by sand. Water may have carried it into the sand. Someone may have dumped older material while raising the ground. Each account predicts different relationships to surrounding particles, boundaries and other finds.

The first task is therefore to describe the relationship that needs explaining. “People lived here” is too broad to direct a sediment investigation. “Did this artefact-bearing sand accumulate where people used the objects, or was it transported?” identifies a question with observable consequences.

The Archaeology root addresses the wider reconstruction of past human life. Geoarchaeology contributes the Earth-science account of the setting and the record. It cannot identify a person’s intention from sediment alone, but it can show whether the context supports the proposed account of their activity.

Contents · Next chapter

Four useful words describe different things. Material is what something consists of: mineral grains, organic matter, shell, ash or construction debris. A deposit is an accumulation with a history of emplacement. Soil develops through processes acting on material over time. An archaeological context is a recorded unit or feature used to organise observations and relationships.

These categories can overlap without being interchangeable. A deposited sand may later develop soil features. A pit fill may contain fragments of an earlier soil. Two contexts distinguished during excavation may share a material source but represent different events. The context number identifies a record; it does not prove a formation mechanism.

Use an invented pair of samples to expose the distinction. Both contain 70% sand in a laboratory fractionation. Sample A comes from a laminated natural deposit; Sample B comes from a compacted construction surface containing crushed building material. Equal sand percentages establish one compositional similarity. They do not establish equal age, function or origin.

This is why a description should preserve both contents and arrangement. Record the matrix, inclusions, boundaries and internal structure before assigning a process name. An interpretation such as “deliberate fill” should remain distinguishable from the observations supporting it, including mixed inclusions and the geometry of its emplacement.

The word “natural” also needs care. A deposit can move through physical processes while its sediment supply has been changed by human land use. Conversely, a human-built surface can later be altered by roots and water. Explaining the sequence is more informative than forcing every sample into a permanent human-or-natural category.

Contents · Previous chapter · Next chapter

Relative order answers which event came before another. It does not assign a calendar date. In a straightforward depositional sequence, the lower deposit formed before the deposit that covers it. A later cut can cross both, allowing younger material to occur deeper than older material elsewhere in the section.

Imagine four recorded units. Silt A is covered by sand B. Cut C truncates A and B. Fill D occupies C. The supported sequence is A before B, B before C, and C before D. A charcoal fragment within D is part of the fill’s contents, but its burning could predate the filling event. The fragment and the enclosing deposit do not acquire identical histories merely by sharing a sample bag.

Now add a separate pit E whose edge never intersects C. If both pits cut B, both are later than B. Their mutual order remains unknown. Drawing one below the other in a convenient diagram does not create evidence that one is older. A good sequence can contain unresolved branches.

The Stratigraphy guide develops the distinction between physical relationships and time. For geoarchaeology, that distinction prevents a common mistake: using depth as a universal age scale instead of reading the actual contacts.

Check the event named in the sentence

Compare “the pit is older than its fill” with “every object in the fill is younger than the pit”. The first follows from cutting before filling. The second does not: old objects can enter a younger fill. Repair the inference by naming whether the claim concerns manufacture, burning, deposition, cutting or later disturbance.

Contents · Previous chapter · Next chapter

A section preserves material that survived. It need not preserve every event or every interval of time. Erosion can remove deposits, while an exposed surface can remain without receiving a substantial new layer. A thin boundary may therefore represent a significant break in the surviving record.

The following original arithmetic separates thickness from duration. Suppose 0.6 m of sediment survives between two independently constrained boundary events separated by 1,200 years. Dividing thickness by elapsed time gives 0.0005 m per year, or 0.5 mm per year. This is an average net preserved accumulation over that interval under the stated boundary assumptions.

It is not evidence that 0.5 mm arrived every year. The same preserved thickness could result from brief deposition followed by long stability. It could also be the remainder after a larger accumulation was partly eroded. Equal arithmetic averages can hide very different histories.

If evidence instead indicates that deposition occupied only 200 years within the interval, the corresponding average during that depositional period would be 0.6/200 = 0.003 m per year, or 3 mm per year. The calculation changes because the denominator now represents a different process interval. Neither rate alone resolves how deposition varied within its interval.

When reading a report, ask whether a rate concerns gross input, net accumulation or preserved thickness. Check whether compaction or erosion could affect the numerator. A precise division can be mathematically correct and still answer the wrong historical question. The repair is to define the quantity before interpreting its value.

Contents · Previous chapter · Next chapter

Sediment interpretation improves when source, movement and final setting are considered together. A grain-size distribution reflects the material available as well as the processes selecting and transporting it. A site at a slope foot and a site beside a channel may receive overlapping materials through different routes.

Historic England’s geoarchaeology guidance discusses slope, river, wind, wetland and coastal processes alongside human contributions. These are useful process families, not colour codes that identify every deposit at a glance.

For an original comparison, picture a sandy patch in three settings. On a windy open surface, selective wind transport is a candidate. In a channel margin, water transport is a candidate. Inside a building platform, imported construction material is a candidate. The word “sand” fits all three; the wider geometry and associations help distinguish their histories.

Follow the evidence outward. Does the deposit thicken toward a plausible source? Is it continuous with a channel feature? Does it contain materials consistent with deliberate import? Do contacts and structures support the proposed emplacement? These questions link the small sample to a landscape or activity rather than making the sample carry the whole interpretation.

The Sedimentology root provides a fuller account of sediment transport and deposition. Here the archaeological question adds a further requirement: explain how that movement affected the association between the sediment and the evidence of human activity. A transported artefact can remain informative, but its location answers different questions from an undisturbed use surface.

Contents · Previous chapter · Next chapter

Soil formation can change material after deposition. The University of Nebraska–Lincoln’s soil-forming processes lesson distinguishes additions, losses, transformations and translocations. These categories help separate material entering or leaving a profile from material changing or moving within it.

Apply that vocabulary to an invented buried surface. Wind adds fine material. Water removes a dissolved component. A mineral changes through weathering. Fine material moves from one part of the profile into another. These are four different claims, and each requires an observation appropriate to the process. A darker zone alone cannot prove all four.

The archaeological consequence is that a visible horizon may express later alteration rather than a separate episode of occupation. Equally, fragments of an older horizon can be redeposited. A soil description and a depositional sequence should inform each other without being collapsed into one list of coloured bands.

Ask what survived of the original arrangement. If roots cross a boundary, the boundary may still be traceable while some small contents have moved. If a surface was truncated, an absence of an upper horizon may reflect removal rather than failure to develop. The useful interpretation identifies the scale and consequence of alteration.

A practical learning exercise is to annotate a published profile twice: first for visible deposits and contacts, then for features potentially associated with soil development. Keep uncertain features marked as uncertain. The aim is to notice that deposition and subsequent modification can overlap spatially while occupying different places in the history.

Contents · Previous chapter · Next chapter

People excavate, import, mix, burn, compact, discard and maintain material. Geoarchaeology makes those actions physically explicit. A raised surface requires material to arrive; a cleaned surface requires material to leave; a maintained floor may combine repeated additions with wear and removal.

Consider an original room-scale case. A compact layer contains angular fragments of an earlier wall, lies above an uneven substrate and has a level upper surface. Deliberate levelling is a plausible explanation. It becomes stronger if the arrangement fits the building sequence and differs from adjacent naturally accumulated material. The observations still need comparison with alternatives, including collapse followed by later compaction.

The same care applies to waste. A concentration of refuse does not necessarily mark the place where the objects were used. A dumped mixture can record disposal choices, material sources and the availability of a hollow. It may preserve little of the original spatial arrangement of the activities that generated the waste.

Think in verbs before assigning a room label. “Material was repeatedly added and compacted” is a different level of inference from “this was a workshop”. The second needs evidence of the relevant activity, not merely a surface capable of supporting it.

This approach also avoids treating human action as an interruption of an otherwise natural history. Human modifications can redirect water or change sediment supply, after which physical processes continue. A useful reconstruction follows those interactions through time and identifies which part of the sequence each observation supports.

Contents · Previous chapter · Next chapter

A loose sample and an intact sample preserve different kinds of evidence. A bulk sample may support a compositional measurement. An oriented block can preserve relationships among particles, voids and boundaries. Neither is universally better; the appropriate choice follows from the question.

In an invented sampling decision, the question is whether a fine band lies above or within a trampled surface. Combining both into one bag may produce an accurate chemical average while losing the relationship needed to answer the question. A spatially documented approach must preserve the boundary and the relevant contrasts.

The existing micromorphology and microcontext guide explains why orientation and contextual records matter. The broader principle is that preparation can remove information. Once a relationship has been mixed away, greater instrument precision cannot reconstruct it automatically.

Sample support means the physical area, volume or interval represented by a measurement. Two results may be expressed in the same units while representing different supports: a thin band, an entire fill, or a composite from several locations. Their comparability depends on those choices.

Before interpreting a number, complete a sentence: “This measurement represents this material, from this location and interval, prepared in this way.” If that sentence cannot be completed, preserve the uncertainty. Real archaeological sampling should be designed with the relevant specialists and permissions; classroom reasoning can use documented records without disturbing archaeological material.

Contents · Previous chapter · Next chapter

Particle-size analysis describes a distribution under a stated method. It can reveal similarities, contrasts and mixtures. Turning that distribution into an environmental explanation requires context, including source materials, preparation and evidence of subsequent change.

The particle-size analysis specialist explains why sieve measurements, settling methods and optical measurements should not be treated as interchangeable. A class boundary and measurement basis belong with the reported percentage.

An original mass-balance check

Suppose a laboratory exercise begins with 120 g of dry material under a defined protocol. It records 72 g in the chosen coarse fraction, 36 g in the finer fraction and 12 g as unrecovered material. The recovered mass is 108 g. Coarse material is 72/108 = 66.7% of recovered mass, but 72/120 = 60% of starting mass.

Both percentages can be correctly calculated. They describe different denominators. Reporting 66.7% without the recovery information would conceal that one-tenth of the starting mass is unaccounted for. The missing material might not have the same size distribution as the recovered material.

The strongest conclusion is therefore conditional: two-thirds of the recovered mass belongs to the defined coarse fraction. The original sample’s complete distribution has not been established by that result alone. Check recovery and replicate preparation before making a comparison that assumes no selective loss.

Nothing in this calculation identifies a flood, a wind regime or a flow velocity. Those are process interpretations. The arithmetic establishes a measurement property; the next task is to connect it with the deposit’s fabric and setting.

Contents · Previous chapter · Next chapter

A mixture model can make an interpretation testable by stating exactly what is being combined. It can also expose how much depends on assumed source properties. Start with a deliberately simple, original two-source exercise.

Source A contains 80% coarse material by dry mass, and Source B contains 20%, using the same preparation and class boundary. A deposit contains 50%. If f is the fraction of the deposit’s mass supplied by A, and mixing is conservative with no selective gain or loss, then 0.50 = 0.80f + 0.20(1 − f). Rearranging gives 0.30 = 0.60f, so f = 0.50.

Under this model, half the mass came from each source. That is a model result, not a direct observation of past transport. The assumptions include two relevant sources, representative source measurements and no later process changing the measured fractions.

Now change only Source A to 70% coarse material. The equation becomes 0.50 = 0.70f + 0.20(1 − f), giving f = 0.60. The same deposit measurement now implies 60% from A. Uncertainty about the end members changes the reconstruction even if the deposit measurement is precise.

Check whether the model can explain the observation

If the deposit were 90% coarse, neither mixture of the original 80% and 20% sources could produce it. A fitted fraction greater than one would not describe a physically valid mixture. It would signal that the source set, preparation, measurement or conservative-mixing assumption needs revision. An impossible answer is useful evidence about the model, not a number to hide.

Contents · Previous chapter · Next chapter

Micromorphology examines the internal fabric of soil or sediment. The information is not simply a magnified version of a bulk percentage: neighbouring particles, coatings, voids and fine layers can preserve relationships that a mixed sample loses. Interpretation depends on the documented location and orientation of the material examined.

Consider a fictional thin section with a fine band above a coarser aggregate-rich zone. The band is continuous across one part of the section but interrupted by a channel-like feature elsewhere. One question concerns the original layering; another concerns the later interruption. A description should retain both, rather than calling the entire sample uniformly intact or uniformly disturbed.

Representativeness remains a separate issue. A carefully prepared section may provide strong evidence about its own location while leaving the wider room or landscape unresolved. Selecting only the most visually striking micrograph can produce an attractive but unbalanced account of a heterogeneous deposit.

A useful record connects the large-scale section drawing, sample position, block orientation and microscopic image. It also distinguishes what was visible before preparation from what became visible during analysis. That chain allows another researcher to understand why the microfeature bears on the archaeological question.

For learners, the appropriate first skill is comparison. Explain what additional relationship the microscopic view reveals, and what it cannot establish beyond its sampled area. The Teaching Geoarchaeology guide offers a teaching route through sediments, field records and site formation.

Contents · Previous chapter · Next chapter

Evidence related to fire raises several questions: what material burned, where it burned, how residues reached the sampled location and what happened afterwards. These questions should not be collapsed into the single label “hearth”.

Historic England’s research overview describes microscopic study of plant ashes. The immediate lesson for interpretation is to distinguish identified material from its depositional setting. A residue can be genuine evidence of combustion while having been moved after burning.

In an original comparison, Deposit A contains a concentrated residue associated with a local surface and a coherent sequence of surrounding changes. Deposit B contains similar residue dispersed among heterogeneous dumped material. The similarity in residue does not establish equivalent use of the two locations. A sequence compatible with burning in place needs testing against dumping, reworking and subsequent alteration.

Conversely, failure to find a conspicuous ash layer is not a simple demonstration that fire was absent. The investigation must consider what evidence the formation and preservation conditions would allow to survive. This does not make every interpretation equally likely; it defines what a negative observation can actually test.

Write conclusions at the supported level. “Combustion residue occurs in this fill” is narrower than “people cooked at this precise spot”. To justify the second statement, connect residue identification with spatial arrangement, contextual relationships and relevant activity evidence. The difference is a scientific question about the record, not merely a choice of cautious wording.

Contents · Previous chapter · Next chapter

Chemical enrichment is a comparison. A high numerical concentration means little archaeologically until the measurement, local substrate, soil history and comparison group are known. A background from a different material may give a misleading reference even when measured accurately.

Weihrauch and Söder’s Altenburg hillfort study examines this problem in heterogeneous soils. It develops an approach based on locally less-influenced areas while recognising that such areas need not be completely untouched. The study illustrates why the background itself is an inference requiring justification.

An original background calculation

Suppose three matched background samples yield 180, 200 and 220 mg/kg under one stated extraction procedure. Their mean is 200 mg/kg. A target sample measures 500 mg/kg, producing a target-to-background ratio of 2.5. If the target instead belongs to a substrate whose appropriate background is 400 mg/kg, the ratio becomes 1.25.

The target measurement has not changed. The interpretation of its contrast has changed because the comparison became more appropriate. Neither ratio identifies a specific room function, and neither is a count of people or years of occupation. Additional contextual evidence must support the historical claim.

The soil-phosphate and background-gradients specialist develops sampling and analytical comparisons. Use it when the main uncertainty is how a mapped concentration relates to activity. At root level, preserve the distinction between measuring an analyte, demonstrating enrichment and explaining its cause.

Contents · Previous chapter · Next chapter

A date becomes archaeologically useful through its relationship to an event. The material’s formation, its arrival in a deposit and a person’s activity may be separated in time. Dating methods do not remove the need to establish that relationship.

Optically stimulated luminescence provides a useful example. The USGS Luminescence Dating Laboratory explains the connection between stored radiation dose, resetting and dose rate. For a simplified burial-age model, age equals equivalent dose divided by the relevant average dose rate.

In an invented calculation, an equivalent dose of 18 Gy divided by 3 Gy per thousand years gives 6 thousand years. The units matter: Gy divided by Gy per thousand years leaves thousand years. This is a demonstration of the equation, not an age determination for a real sample.

For an illustrative sensitivity calculation, assign independent one-standard-deviation uncertainties of 1 Gy and 0.3 Gy per thousand years. First-order propagation gives a relative uncertainty of √[(1/18)² + (0.3/3)²], approximately 0.114. Multiplying by 6 gives about 0.69 thousand years. Independence and the approximation are explicit assumptions.

That numerical uncertainty does not account for every possible systematic problem. Incomplete resetting or an inappropriate dose-rate history can compromise the simple interpretation. Nor does a burial estimate automatically date the manufacture of an artefact mixed into the sediment.

Mahan and colleagues’ guide to reporting luminescence results emphasises supporting dose, dose-rate, age-model and stratigraphic information. A rounded age without that account is an incomplete basis for reconstructing the site.

Contents · Previous chapter · Next chapter

A deposit model connects observations from separate locations to represent the distribution of buried units. It is a reconstruction between sampled points, not a photograph of everything underground. The interpolation assumptions and gaps belong with the model.

Carey and colleagues’ research on deposit modelling discusses integrating sediment sequences, geotechnical data and archaeological records. That integration matters because records made for different purposes can have different resolutions and descriptive conventions.

Depth is not elevation

In an original example, Borehole A begins at elevation 12.0 m and meets a boundary 2.0 m below ground. Its boundary elevation is 10.0 m. Borehole B begins at 10.5 m and meets a similar-looking boundary 1.0 m below ground. Its boundary elevation is 9.5 m.

Although B’s boundary is shallower below its ground surface, it is lower in absolute elevation. Comparing raw depths alone would misrepresent the geometry. The calculation assumes both ground elevations use the same datum and that depths are measured consistently.

Similarity still does not prove the boundaries are equivalent. Correlation needs supporting sediment characteristics, sequence relationships and chronological evidence where relevant. A line drawn between the points is one hypothesis to test, especially if a channel, erosion surface or fault could intervene.

Preserve the observation locations on the final model. A smooth surface without visible data support can make unsampled space appear certain. A useful model identifies where another observation would discriminate between competing geometries rather than merely adding detail where the history is already reasonably constrained.

Contents · Previous chapter · Next chapter

Sampling should address the feature or contrast that matters. A regular grid can provide systematic coverage, but its spacing may miss small or irregular features. Targeted sampling can investigate a specific model, while also risking neglect of unexpected areas. The design should state what each component is intended to establish.

Imagine an original valley survey with four boreholes aligned along a modern access route. All encounter a fine deposit. It is reasonable to report that the deposit occurs at those locations. It is not yet reasonable to claim that its geometry across the entire valley is known. The accessible line may not intersect the most informative boundaries.

One additional borehole placed across the competing predictions of two landscape models can be more informative than several placed beside existing points. That is a reasoning principle, not a universal numerical sampling prescription. Actual placement must also account for access, permissions, preservation and technical constraints.

Historic England’s deposit-modelling case study, published on 5 February 2024, describes how landscape understanding helped target investigation on the Bexhill to Hastings link road. The value was the connection between the buried setting and an appropriate evaluation strategy, rather than treating every part of the landscape as equivalent.

For a learning project, compare two proposed sampling plans against the same question. State what each would observe, what it could miss and what a negative result would mean. “No feature encountered” is a result about the sampled locations and method; widening it into “no feature exists” requires a defensible detection argument.

Contents · Previous chapter · Next chapter

When an interpretation is weak, the next useful step depends on the kind of weakness. More precise measurements help some problems; a better comparison or preserved spatial relationship helps others. Begin by identifying the unsupported link.

Consider two apparently conflicting measurements. Before deciding that one laboratory is wrong, compare sample locations, preparation, measurement basis and the quantity each result estimates. A bulk average and a thin-band measurement can differ because they answer different questions.

Also check whether supposedly independent evidence is circular. If a deposit was labelled a flood solely from grain size, using that label to validate the grain-size interpretation adds no new support. A separate observation of geometry or structures may provide a more meaningful test.

A good repair ends with a revised claim. If a room’s function remains uncertain, it may still be possible to establish repeated material addition or a contrast in sediment composition. Retaining a narrower supported conclusion is useful progress. It preserves what has been learned while making the next research question precise.

Contents · Previous chapter · Next chapter

Contents · Previous chapter · Next chapter

Is geoarchaeology the same as geology?

It draws on geological and other Earth-science methods to answer archaeological questions. The distinction is partly the purpose of the investigation: understanding deposits, landscapes and preservation in relation to human history. A geological description can be essential without, by itself, explaining a past human activity.

Can one method identify how a layer formed?

Sometimes a method provides a strong constraint, but many features have more than one possible cause. The key is whether the proposed history explains the combined observations and survives relevant alternatives. Repeating one measurement does not necessarily address uncertainty about its interpretation.

Does a deeper find have to be older?

No. Later cuts, fills and disturbances can place younger material below older material elsewhere. Older material can also be redeposited in younger contexts. Read contacts and material histories before turning depth into an age claim.

Can the present river position explain an ancient settlement?

It can help frame questions, but present topography is not automatically the ancient landscape. A reconstruction should establish the relevant former landforms and timing. The relationship between settlement and water must be demonstrated for the period being discussed.

Is an intact deposit useful for every analysis?

No. Spatial integrity, chemical preservation and suitability for a particular dating method are different properties. A deposit may retain one kind of information while losing another. State what “preserved” means for the specific question.

What can a beginner learn without excavating?

Read published sections and borehole logs, calculate elevations, distinguish observation from interpretation and compare proposed sample supports. Annotated public research records can support demanding exercises without collecting material or disturbing a site. Practical field and laboratory work require appropriate training and supervision.

Contents · Previous chapter · Next chapter

Start with one published deposit description. Rewrite its conclusion as a sequence of claims: material identified, arrangement recorded, process proposed, event dated and human interpretation made. Then identify the evidence attached to each step. This exercise reveals where a clear description becomes a more demanding historical inference.

For physical processes, continue through Sedimentology and Stratigraphy. For the changing condition and arrangement of biological remains, Taphonomy provides a connected route. These subjects overlap because the same deposit can preserve several kinds of evidence with different histories.

For practical analytical questions, choose the specialist that matches the unresolved relationship: particle-size distributions, micromorphology and sample orientation, or soil-phosphate backgrounds. Teaching and self-study can continue through the Geoarchaeology teaching guide.

Professional routes connect archaeological investigation, soil and sediment analysis, dating, landscape modelling and heritage conservation. Useful preparation includes quantitative reasoning, careful description, spatial records and the ability to explain uncertainty. Course and role requirements vary; inspect the actual work and current programme information before choosing a route.

Sources and how to use them

The linked Historic England overview and 2015 geoarchaeology guidance establish the field’s scope. The Nebraska soil-science lesson explains process vocabulary. The Altenburg research provides a specific example of the background problem in soil phosphorus interpretation. The USGS laboratory explanation and Mahan and colleagues’ reporting guide support the luminescence discussion. Carey and colleagues’ deposit-modelling paper and Historic England’s 2024 case study connect spatial reconstruction with investigation design. These sources were checked on 23 October 2026; the worked numerical cases and transfer exercise are original educational examples.

Use the How X Works hub to continue studying mechanisms across subjects. The capability to carry forward is concrete: explain what a measurement represents, how the material reached that position, which event is being reconstructed and what evidence could change the conclusion.

Contents · Previous chapter · Continue to How X Works