A distribution map of Neolithic obsidian in the Near East is one of the most persuasive images in archaeology. Dots radiate from volcanic sources in central and eastern Anatolia and thin steadily with distance, and the eye reads movement, contact, and exchange. Yet the map records only where fragments were eventually discarded and recovered. It says nothing directly about how they travelled, who carried them, how many hands they passed through, or what anyone believed they were worth.

Exchange studies therefore rest on a chain of inferences, and each link has its own weaknesses. An object must first be assigned to a source. Its movement must then be attributed to a mechanism, whether reciprocal gift, profit-seeking trade, or centralised redistribution. Finally, its significance to those who moved it must be reconstructed. An error at any link carries forward to the next. Confident conclusions about ancient networks often conceal uncertainty that has built up at every stage.

R.G. Collingwood insisted that nothing is evidence until it answers a question, and that the historian's first task is to ask the right one. Exchange studies show the principle clearly, because the same material pattern can answer several incompatible questions equally well. What follows examines three links in turn: provenance determination, mechanism inference, and value reconstruction. In each case, the aim is to state what current methods can establish and exactly where their authority ends.

Provenance research depends on what Weigand, Harbottle, and Sayre formalised in 1977 as the provenance postulate. The chemical or isotopic differences between sources must be greater than the variation within any single source. Where the postulate holds, sourcing can be remarkably secure. Obsidian is the classic case, because individual volcanic flows tend to be chemically uniform. In the 1960s, the trace-element work of Renfrew, Dixon, and Cann distinguished Anatolian sources with a confidence that flint, chert, and most clays have never allowed.

Ceramic petrography examines thin slices of pottery under the microscope, identifies the mineral inclusions, and matches them to regional geology. David Peacock and others shaped its modern form. Its strength is spatial: a distinctive volcanic or metamorphic signature can rule out large areas at once. Its limitations are built into the method. Similar geologies recur across wide regions, potters routinely mixed clays and added temper, and raw clay itself could travel. A petrographic match identifies where the materials came from, which is not necessarily where the vessel was made.

Lead isotope analysis faces subtler difficulties. It has been applied extensively to Late Bronze Age copper, including the oxhide ingots recovered from the Uluburun shipwreck. The isotopic ranges of different ore deposits can overlap. Recycling and alloying produce mixed signatures that match no single source. Most fundamentally, any comparison is only as complete as the reference database: a mine that has never been sampled, or was worked out long ago, is simply invisible. Lead isotope results are therefore most reliable as exclusions, stating where metal cannot have come from, rather than as positive identifications.

Trace-element methods, from neutron activation to laser ablation mass spectrometry, run into a further problem with metals. Smelting is a chemical transformation, and elements divide unevenly between the metal and the slag. An artefact's composition is therefore a distorted record of its ore. The long-standing 'tin problem' of the Bronze Age shows both the difficulty and the potential. Recent tin isotope studies have linked Uluburun ingots to deposits in Anatolia and Central Asia, but such results depend on which deposits have been characterised so far.

Even a secure sourcing result establishes only two points: an origin and a findspot. The route between them, the number of intermediate transactions, and the time that elapsed all remain unknown. This is not a failure of the methods but a limit on what they can logically show. Provenance is a necessary condition for reconstructing exchange, not a sufficient one. Much interpretive overreach in the literature begins when a sourcing result is quietly treated as though it described a journey.

Takeaway

A provenance result gives two dots on a map, not the line between them. The line is always an inference, and it should be argued for rather than simply drawn.

In the 1970s Colin Renfrew proposed that exchange mechanisms might be read from the shape of distribution curves. In down-the-line exchange, goods pass between neighbouring communities, which should produce a smooth exponential decline in frequency with distance from the source. Directional trade toward central places should generate secondary peaks. Redistributive systems should concentrate material at administrative centres. The model was elegant, testable, and enormously influential, because it seemed to offer archaeologists a quantitative route from pattern to process.

Ian Hodder's simulations soon complicated the picture. Quite different processes, including essentially random movements of objects across a landscape, could generate fall-off curves that were statistically indistinguishable from those predicted for specific exchange modes. This is a textbook case of equifinality: several causal pathways converge on the same material outcome. Uneven preservation, uneven excavation, and varying rates of discard distort the curves further. The observed distribution is already several steps removed from any ancient transaction.

The problem deepens with Karl Polanyi's distinction between reciprocity, redistribution, and market exchange. These categories differ chiefly in their social relations: obligation, authority, and price. The Trobriand kula, documented by Malinowski, involved shell armshells and necklaces circulating in opposite directions around a ring of islands, bound up in elaborate obligations and reputations. Excavated centuries later, the kula would leave a scatter of shell ornaments. The social logic that defined the institution would leave nothing at all.

The rare cases with surviving documents work as controls, and they are sobering. The Old Assyrian archives from Kültepe, ancient Kanesh, comprise over twenty thousand tablets from early in the second millennium BCE. They record private family firms shipping tin and textiles from Assur to Anatolia, financed through partnerships and credit. Yet the main commodities are almost invisible archaeologically: the tin was alloyed into bronze and the textiles perished. Without the tablets, this commercial system would be poorly understood and quite possibly misclassified as palatial or reciprocal exchange.

The more promising response has been to infer mechanism from associations rather than from distributions alone. Standardised weight systems, sealing practices, storage buildings, harbour facilities, and the settings in which goods were consumed each narrow the range of plausible mechanisms. Bronze Age balance weights found across Europe and the Mediterranean, for instance, imply that communities had some shared way of measuring value. Yet even indicators that agree with one another narrow the possibilities without settling them. Most ancient economies were almost certainly mixed, combining several modes at once.

Takeaway

Identical distributions can arise from radically different social worlds. The mechanism of exchange lives in relationships, and relationships rarely survive in the ground.

Value reconstruction faces the deepest interpretive hazard: projecting modern assumptions about scarcity and price onto ancient objects. The cemetery at Varna, on the Black Sea coast of Bulgaria, contains some of the earliest worked gold anywhere, dated to the mid-fifth millennium BCE. Much of it lies in a small number of graves, including symbolic burials without skeletons. To modern eyes the value of gold seems obvious. Methodologically, however, its value at Varna must be demonstrated from context, not assumed from the material.

The debate over Athenian painted pottery shows how strongly modern valuations can distort interpretation. Michael Vickers and David Gill argued that these vessels, now prized by museums and collectors, were inexpensive goods imitating silver and gold tableware. Price graffiti scratched beneath some vase feet suggest modest sums. Their thesis remains contested, but it exposed how much of the scholarly ranking of Greek crafts reflected the antiquities market and the tastes of eighteenth- and nineteenth-century connoisseurs rather than ancient evidence.

Igor Kopytoff's concept of the cultural biography of things, developed alongside Arjun Appadurai's work on the social life of commodities, offers a more disciplined framework. Value is not a fixed property. It shifts as objects move through different phases of their lives: commodity, gift, heirloom, offering. Archaeologists can sometimes trace these phases through wear from use, repair, long-term keeping, and the manner of final deposit. Neolithic Spondylus shell ornaments in central Europe, for instance, were repeatedly repaired, which suggests a significance their raw material alone does not explain.

Several indirect indicators of value are now standard. One is skeuomorphism, where cheaper materials imitate prestigious ones. Others are the distance and effort involved in acquiring an object, its restriction to particular contexts, and its deliberate removal from circulation through hoarding or burial. Each indicator is ambiguous. Imitation reveals a ranking but not the size of the gaps within it. Distance says little about value when goods moved down-the-line. Placement in a grave might signal high esteem, or simply the disposal of items whose social life ended with their owner.

The most defensible position recognises that value was relational and plural. The same object could sit within different systems of value for the producer, the intermediary, and the final recipient, and those systems need not have been comparable. Collingwood's ideal of re-enacting past thought is most strained here, because valuation is precisely the mental act that leaves the least direct trace. Triangulating across independent lines of evidence remains the only method that does not quietly substitute our judgements for theirs.

Takeaway

What an object was worth belongs less to the object than to the relationships around it. Assuming value from material alone means reading our own economy into theirs.

Taken as a whole, current knowledge of ancient exchange is unevenly reliable. Provenance determination is strongest for chemically uniform materials such as obsidian. It is weaker for metals and ceramics, where transformation, mixing, and incomplete reference data intervene. Value can be reconstructed within well-documented contexts, but those reconstructions rarely transfer beyond them. Mechanism inference remains the most fragile link, because it requires reading social relations from material patterns that cannot decide between them.

Several directions promise improvement. Larger, openly shared reference databases for ore and clay sources would reduce false identifications. Organic residue analysis and ancient protein analysis are beginning to recover perishable goods, such as foodstuffs and dyes, that dominated many exchange systems. Network models that show uncertainty explicitly, rather than drawing confident lines between sites, would make the gaps in inference visible.

Above all, documented cases like Kanesh should be used systematically as controls, testing whether archaeological methods alone would recover what the texts reveal. The honest answer, so far, is: only partially. That partial success is not a reason for despair. It is a precise measure of how far our inferences can responsibly travel.