A Roman bowl excavated in Britain can look almost identical to one found in southern France. The glossy red finish, moulded ornament and overall shape might all match. Those similarities cannot establish where either bowl was made. A workshop could imitate another’s products convincingly, while differences in firing might leave two vessels from the same production centre with very different colours.
The clay body contains evidence of geological origin and manufacturing technique. Minerals in its clay reflect the geology of its source; the grains mixed into it can reveal how the potter prepared the material; and firing leaves further traces. By comparing these characteristics with excavated workshops, clay deposits and pottery from other sites, archaeologists can sometimes narrow a vessel’s origin to a region or, with unusually good comparative evidence, to workshops situated quite close together.
Archaeologists call this ceramic fingerprinting, but the term should not suggest a unique code for every ancient kiln. No universal register of such signatures exists. Workshops next door to each other may have used clay from the same deposit, and one potter might have blended material from several sources. A persuasive attribution draws on the excavation record, examination of the fabric under a microscope, chemical measurements and a securely identified group of locally made pots.
What archaeologists mean by a ceramic fingerprint
The term *fabric* refers to the texture and composition of a fired ceramic body: its clay-rich matrix, mineral particles, any material deliberately added and the spaces left as pores. Quartz grains or fragments of crushed pottery may be easy to spot in the broken edge of a coarse cooking vessel, while the ingredients of a fine bowl can be difficult to distinguish without magnification. Such differences need not indicate different places of manufacture. Potters selected their materials to control drying, workability, strength, appearance and performance over heat.
Two similar words distinguish the place where an object was found from the place it was made. *Provenience* generally means an artefact’s archaeological findspot and context. In ceramic research, *provenance* usually means the manufacturing origin, although museum historians also use the word differently. Thus, a bowl recovered at a Roman fort in Britain has a British provenience even if its clay points to southern Gaul. Tracing that distance is central to the history of exchange, as Patrick Sean Quinn’s ceramic petrography handbook explains.
No single mineral provides a dependable fingerprint. Quartz is widespread, calcium-rich deposits occur across many landscapes, and even distinctive volcanic particles can come from more than one source. Researchers examine the combined pattern of grain types, relative chemical concentrations, clay texture and signs of manufacture. Analysts must then ask whether that pattern resembles a secure reference group more closely than it resembles other plausible sources. Without the right comparison material, a seemingly precise match may prove quite weak.
From raw clay to fired vessel
Clay minerals are mixed with particles that do not behave plastically, including quartz, feldspar, limestone fragments and sometimes tiny fossils. Common clay minerals include kaolinite, illite and smectite. Before shaping a vessel, a potter might settle the clay in water to remove coarse material, mix clay from several deposits or add temper to change its working properties. Sand, shells and crushed old pottery known as *grog* could all serve as temper. Under the microscope, however, a grain of sand or other inclusion does not announce whether someone added it or whether it was present in the original deposit.
Drying and firing change this mixture. Free water escapes, clay minerals lose hydroxyl groups bound within their structure, and heating can destroy some mineral phases while producing others. Carbonates may decompose under sufficiently strong heat, iron compounds respond to the available oxygen, and high temperatures may leave portions of the body glassy. The particular changes depend on the clay’s original composition and grain size, as well as the firing atmosphere, heating rate and duration. There is no single set temperature at which every clay behaves in the same way.
Conditions can vary even across one vessel. In experimental open firings measured by Marino Maggetti and colleagues, temperature differences reached 390°C within a single firing, and the mineral indicators did not always match the temperatures recorded directly by instruments. Dark interiors, lighter margins and red exteriors may reflect changes in heat and oxygen exposure. Their colours cannot, on their own, identify a workshop or give an exact firing temperature.
Reading the broken edge: fabric and thin-section petrography
The investigation often starts at the excavation table, before any expensive instrument is used. On a clean, existing break, archaeologists examine core and edge colours, the size and sorting of particles, pores, surface finish and traces of forming. A hand lens can separate coarse, sandy fabrics from finer, calcium-rich ones. This initial sorting lets a team arrange thousands of sherds into provisional groups and decide which examples need more detailed analysis.
For thin-section petrography, a small fragment is prepared as a slice generally about 30 micrometres thick. Under a polarising microscope, transmitted light makes mineral grains and textures visible that would otherwise escape notice. A specialist can recognise quartz, feldspar, mica, calcite, volcanic rock fragments or ceramic grog. Grain relationships offer further evidence: a mixture of unusually coarse and fine particles might indicate added temper, while elongated pores and aligned grains can suggest how the potter shaped and worked the clay.
Identifying the mineral is only the beginning. Limestone containing microscopic fossils may point toward one geological setting, whereas volcanic grains suggest other possibilities. The analyst still has to determine whether such material occurred naturally in the clay, was deliberately mixed in, or arrived after burial. The UCL Institute of Archaeology’s ceramic petrography service uses this combination of geological identification and technological interpretation to investigate prehistoric, Roman and later pottery.
Different-looking wares can turn out to share much of the same fabric. Two vessels with contrasting painted decoration might derive from related clay sources, while bowls decorated with the same motif might have been made from quite different materials. The second finding could indicate imitation or a design shared among workshops. A thin section cannot identify the owner of a pottery business or name an individual craftsman, but it can show the ingredients and manufacturing choices behind the finished piece.
X-ray fluorescence: measuring elements in pottery
X-ray fluorescence, abbreviated XRF, measures elements from the characteristic X-rays emitted when a specimen is excited by the instrument. In a laboratory, analysts can measure prepared powders or fused samples; portable devices allow measurements in a museum or store without removing a large portion of the object. This is useful when sampling restrictions are strict. The high-precision XRF work by Adan-Bayewitz, Asaro and Giauque showed that the method could distinguish ceramics from two neighbouring production centres in Roman Galilee.
An XRF reading samples only part of an object. How far it penetrates depends on the element and ceramic matrix, and a measurement made through weathering, glaze or slip may describe the surface rather than the clay beneath. Grain size, curvature, roughness, instrument calibration and settings all affect the figures. Fine red-slipped pottery is a particular concern because the carefully prepared coating can differ chemically from the body. Robert Tykot’s assessment of portable XRF discusses the resulting limits on precision, accuracy and surface interpretation.
Good comparisons therefore begin with appropriate standards, recorded calibration and measurements at several suitable points. Conservation has to come first: an uncleaned surface may carry residues that would be destroyed by preparation. For questions about the underlying ceramic body, a small exposed break or carefully prepared specimen can give a more reliable answer than the outside of a thickly slipped bowl. Portable XRF is valuable for screening collections, but colourful clusters on an uncalibrated scatter plot are not evidence that each group came from its own kiln.
Neutron activation and mass spectrometry
Instrumental neutron activation analysis (NAA or INAA) has long been used to investigate pottery origins. A prepared specimen is irradiated with neutrons in a research reactor. Some nuclei capture those neutrons and become radioactive isotopes; the gamma rays they release during decay can be measured to calculate concentrations of many elements. Some of the measured trace elements differ among geological sources. The University of Missouri Research Reactor’s Archaeometry Laboratory has built substantial comparative datasets through this form of geochemical analysis.
NAA normally uses material taken from the ceramic body, which helps avoid some of the difficulties posed by weathered surfaces and applied slips. It does not make sampling problems disappear. A fragment rich in unusually large mineral grains, for example, may produce different measurements depending on the part selected. Analysts also need to know which elements may have changed during firing or burial and whether the comparison groups were measured under consistent laboratory conditions.
Inductively coupled plasma methods can measure a ceramic’s elemental composition in another way. ICP-MS introduces material into a hot plasma, ionises it and uses a mass spectrometer to detect even very low concentrations of many elements. ICP-OES analyses the light emitted by excited atoms and ions instead. Laser-ablation ICP-MS can investigate extremely small selected areas, though these may represent a coating, an individual inclusion or a patch rather than the entire vessel. A 2002 interlaboratory comparison of NAA, XRF, ICP-OES and ICP-MS demonstrated why calibration between laboratories matters when their results are combined.
Precise elemental measurements still have to be compared with material of known origin before researchers can assign a village or pottery yard. Researchers compare sets of elemental concentrations with known groups, consider uncertainty and look for meaningful patterns. Principal component analysis and other multivariate techniques can help, but groups on a plot may arise from different sources of clay, recipes, firing conditions or changes after burial. Statistics are most effective when they help assess an archaeological question already grounded in evidence.
X-ray diffraction, microscopy and the firing process
X-ray diffraction (XRD) examines how a specimen’s crystalline minerals scatter X-rays. Chemical analysis may reveal calcium in a sherd, for instance, but diffraction can help establish whether it remains in calcite or occurs in another mineral phase. Since minerals respond differently to heat and burial, such distinctions matter. Residual carbonates, surviving clay minerals and phases formed at higher temperatures can limit the range of plausible firing conditions, although they rarely reveal one exact temperature.
Scanning electron microscopy (SEM) magnifies the fired body enough to reveal details of its pores, the fusing of fine clay particles and the boundary between body and coating. Combined with energy-dispersive X-ray spectroscopy (EDS), it can also measure elements in selected areas. This helps explain, for example, how a fine iron-rich slip on glossy Roman red ware was prepared and fired. Quinn’s 2022 study of petrography, geochemistry and SEM treats these investigations as complementary lines of evidence.
Each technique provides a different kind of evidence. Thin sections can reveal volcanic inclusions, XRF can compare elemental proportions, XRD can distinguish crystalline minerals, and SEM can expose glassy structures produced by firing. Pots made from broadly similar ingredients might acquire different microstructures in different kilns; similarly red surfaces can conceal quite unrelated clay recipes. Understanding how the observations relate gives more dependable answers than relying on one supposedly decisive machine.
Kilns, wasters and the evidence left by workshops
A kiln excavated in place gives researchers a known production location against which to compare loose sherds. Nearby they may find work surfaces, clay-settling pits, water installations, fuel waste and dumps of pots ruined during firing. These rejects, called *wasters*, may be warped, cracked, collapsed or fused into one another. When their association with a kiln is secure, they demonstrate that at least some pottery of the same kind was made at that spot.
The equipment inside and around a kiln can reveal how production worked. Supports and spacers kept vessels apart, perforated floors directed the heat, and in some traditions protective containers called *saggars* sheltered the pottery’s surfaces. Their forms show how wares may have been arranged during firing. Other potters relied on open firing rather than permanent kilns, and designs varied substantially. Ash or glaze on surviving equipment provides clues, though reuse and contamination must be considered before drawing conclusions.
Marks impressed on amphorae and table bowls provide evidence of a different kind. Depending on the period and object, a stamp might refer to a potter, proprietor, workshop relationship or administrative practice. Stamps and moulds could also pass from one producer to another, so a shared name does not prove a shared clay source. Confidence increases when stamps, excavated kiln products and the fabric of an unmarked vessel point to the same production setting.
La Graufesenque: Roman pottery made on an extraordinary scale
La Graufesenque lay where the Tarn and Dourbie rivers meet near modern Millau in southern France. Known in antiquity as Condatomagus, it became a major manufacturing centre for Roman red-gloss tableware in the first and second centuries CE. Archaeologists have uncovered workshops, kilns, storage facilities and other parts of the production area. The city of Millau’s museum and archaeological site account describes this landscape and how widely its products travelled.
The vessels belong to the category archaeologists call *terra sigillata*, fine Roman tableware often coated with a red slip. Some pieces were plain; others carried figures and patterns made in moulds. Their gloss came from specially prepared slip and controlled firing, rather than a modern-style glaze. Southern Gallic potters initially worked within traditions associated with Italian centres such as Arezzo, but their own wares later reached customers across much of the western empire. The regional museum collection’s account describes the La Graufesenque forms, manufacturing methods and distribution.
Inscribed pottery fragments from La Graufesenque give an uncommon view of workshop practice. They list potters, vessel types and quantities, although many entries are abbreviated or damaged and cannot be read like modern factory paperwork. Anthony King’s study of the graffiti explores what these lists reveal about the organisation of labour and production, including references that may concern work performed by enslaved people.
The securely excavated pottery at La Graufesenque provides a reference for identifying other wares. Exported bowls whose shapes, decoration or stamps suggest southern Gaul can be compared petrographically and chemically with that production material. A close fit strengthens the case for an origin in the tradition represented at La Graufesenque. It cannot identify an individual potter, firing or named kiln by itself. Neighbouring producers could use the same clays, and manufacturing recipes could change over the industry’s long history.
Dressel 20 amphorae: when clay chemistry meets written evidence
Dressel 20 amphorae were containers made for transport, particularly olive oil from Baetica, the Roman province occupying much of modern Andalusia. Their rounded shape is familiar from finds across the western empire, including forts and Rome’s vast dump of broken transport jars, Monte Testaccio. The Amphorae ex Hispania research catalogue documents the evidence for their contents and the geographic spread of the type.
Each Dressel 20 amphora can carry several kinds of information. Its shape establishes the vessel’s typological family; stamps on the handles can illuminate production organisation without necessarily identifying an individual craftsman. Painted labels known as *tituli picti* preserve details of handling and quantities. A study of the painted delta-position controls shows how the placement and repetition of numbers can also reflect administrative rules. Those labels belonged to particular commercial procedures and should not automatically be interpreted as the name of an estate.
The clay itself adds evidence about manufacture. Researchers compare Baetican amphora fabrics with pottery from workshops along the Guadalquivir valley. When a distant find agrees in shape, stamp, date and composition, attribution to that region becomes much more convincing. Pinpointing one workshop is harder. The jars came from numerous production sites, and businesses in the same area could use closely related clay deposits.
Questions of ownership require separate evidence. A 2025 analysis of Dressel 20 stamps and Roman contracts investigates different arrangements for leasing, labour and production rather than assuming every business worked in the same way. A match in clay composition may locate where the amphora was made, but it cannot tell us by itself who owned the olive trees, filled the jar or received the profits when it reached Rome.
How a reliable reference group is built
An unknown sherd needs known material for comparison before a laboratory result can have geographical meaning. Researchers establish reference groups using kiln wasters, plausible clay deposits and securely contextualised vessels from places of manufacture. They sample enough examples to include ordinary variation within a production area, instead of choosing a handful of pots whose appearance happens to match the unknown specimen.
Researchers then test whether the reference samples form a coherent group. A single apparent cluster can conceal several clay mixtures or successive periods of production. Unusual results deserve investigation rather than automatic removal: minerals vary across a deposit, and added temper is not always evenly distributed. A study of modern cooking-pot production at Pereruela in Spain showed how much chemical variation can exist within one production community. In some cases even samples from one workshop or different parts of the same pot might be assigned misleadingly to separate origins.
With the reference group established, analysts compare the unknown pot against it and other plausible sources. The chemical resemblance must be considered alongside its thin-section fabric, date and distribution. If the only available reference comes from one workshop, a match establishes compatibility with that material; it cannot rule out an unexamined workshop using similar clay. Reports should distinguish between an attribution to a broad region, a production centre, a fabric group and an origin that remains provisional.
Burial can change the fingerprint
Fired pottery lasts well in the ground, but burial can still change its chemistry. Soil water may dissolve or introduce minerals, salts can crystallise in pores, carbonates can build up and elements may migrate. The original firing may already have left different zones within one vessel. A measurement made centuries later may therefore reflect both the potter’s ingredients and changes since its manufacture.
Evidence from a Late Bronze Age kiln at Kommos on Crete makes the problem clear. Researchers expected pottery produced at one known location to form a convenient chemical reference group. Instead, Buxeda i Garrigós, Kilikoglou and Day found that different pastes, firing histories and changes during burial complicated the pattern. A kiln context can confirm a manufacturing place, but it does not mean every sample taken there will have the same composition.
The coating can complicate measurements too. Slips may be made from different or more finely prepared clay than the vessel body; glazes can introduce lead, alkalis and other ingredients that dominate a surface reading. An open-access study of Hellenistic tableware by Charlton and Miše considers what happens when instruments measure different parts of the same object. Before sampling, researchers must decide whether they are tracing the body clay, coating, pigment or some other manufacturing choice.
Burial changes and contamination have to be assessed before assigning an origin. Good reports state where the specimens came from, how their surfaces were treated and measured, what normalisation was used and whether burial changes are suspected. They also test possible alternative sources. Several independent observations supporting a cautious regional attribution tell us more than a confident kiln name resting on one unexplored chemical result.
Other ways to trace pottery production
The approach works well beyond Roman ceramics. Archaeologists have used comparable methods to study prehistoric exchange, Hellenistic tableware, medieval production and industries outside Europe. Cooking pots, roof tiles, bricks, kiln furnishings and metallurgical crucibles can all be investigated, though their ingredients and uses raise different questions. A crucible might retain material from heated metal, whereas coarse temper may have been deliberately added to a cooking vessel. The quartz-rich, glazed material called Egyptian faience must also be distinguished from ordinary clay-bodied pottery when chemical results are interpreted.
At Montelabate near Perugia in central Italy, excavators identified four amphora kilns. Analysts investigated the products with portable XRF, reference samples and comparisons among elements, then set them against amphorae found at other sites. The published study’s institutional record explains the calibration and matrix problems involved in reliable non-destructive readings. Because the source kiln is known, its results can be tested against travelling amphorae rather than assuming vessels of similar shape were made in one place.
Several related techniques address the date of a kiln rather than the source of its clay. Archaeomagnetic dating compares magnetism retained in material heated in place with regional reference records. Where the fired structure has not been moved, this may help date its last operation; a study of two kilns at Kato Achaia in Greece provides an example. Thermoluminescence can under suitable conditions estimate time since heating. Neither method independently establishes the production origin of an isolated sherd, so the distinction between dating and provenance remains essential.
What the evidence can tell us about ancient economies
Once a fabric has a defensible origin, its distribution can help explain ancient economic activity. Dated vessels at a military settlement, for instance, may show changes in supply links or access to particular markets. When a new ceramic group appears suddenly, trade, changing taste, troop movements and local copying are all possible explanations. Coin finds, inscriptions, excavation layers and the relative frequency of pottery types help historians decide which explanation fits.
The number of producers cannot be read directly from the number of chemical clusters. A large workshop may blend several clays to make its material workable or compensate for changing supplies. A small producer might instead use the same local deposit for centuries. Chemically consistent pottery does not by itself establish a centralised factory; variability does not necessarily mean separate workshops. The Pereruela example demonstrates why treating the two as equivalent is risky.
Sometimes pottery once thought to be imported turns out to have been made locally. Such a result changes the question from how finished vessels travelled to how skills, techniques or fashions spread. Potters could migrate and produce familiar wares using new clay sources, or local craftspeople could copy a design encountered through trade. Fabric analysis helps historians separate the movement of objects from the movement of ideas and people.
Looking at pottery without damaging it
Existing breaks in pottery and good museum displays let visitors recognise some of the same features. Visitors can examine grain size and shape, pores, differences between slip and body, and marks left by forming. Reference collections show how fabrics compare. None of these clues provides a secure manufacturing address on its own: a dark core does not necessarily indicate a low firing temperature, nor does a bright red surface prove an Italian or Gallic origin.
Objects in excavations and collections must be handled under appropriate permission and conservation rules. Scratching a sherd with a steel point, placing acid on it or making a fresh break can destroy surface traces and residues valuable to future investigators. Older hardness and spot-test practices are unsuitable for visitors and unauthorised collection work. An object loses much of its research value if its archaeological context and records are neglected, however detailed its mineral fabric appears.
A reliable attribution usually depends on several kinds of evidence. Kiln wasters establish an actual manufacturing location; thin-section petrography identifies ingredients and working methods; chemical measurements test similarities; inscriptions, chronology and findspots show how pottery moved. Together they may identify a production centre, or they may support no more than a region. That limitation belongs in the historical conclusion: archaeologists can only locate a pot’s origins as precisely as the evidence permits.
Discussion
Comments