The AI Journal

A Cycladic marble figure, three or four thousand years older than the Parthenon, can reach auction with a certificate that satisfies every test the trade and the law apply to it: no live claim of theft, no breach of an export rule written after the piece left the ground, a chain of ownership documented back past any date that matters. By that test the object is clean. By the only test that matters to someone studying the Early Bronze Age Aegean, it is close to worthless. Nobody can say what grave it came from, what it lay beside, which island, which generation, which hand. These are not two verdicts in tension. They are answers to two different questions, and the antiquities trade has spent half a century treating them as one.

Gill and Chippindale’s survey of the published corpus of Cycladic figures found that the overwhelming majority carry no record of where they were found at all, and traced the shortfall to collecting itself: rising prices from the 1960s onward rewarded speed out of the ground and into a dealer’s hands, and paid nothing for the one fact that only exists at the moment of excavation and can never afterward be restored (Gill & Chippindale, 1993). A buried figure’s findspot, its grave goods, its position relative to other burials, are information, not decoration; once a figure is prised free without that record, the information is gone regardless of whether removing it broke any statute of its day. Most of these figures left their islands before the laws that would now call that removal illegal existed. The objects are legally unimpeachable and scientifically dead, and the two facts have the same cause.

Elia’s account of looting makes the underlying point explicit: archaeological context is a nonrenewable resource, consumed once and not thereafter replaceable by better technology, deeper pockets, or a change in the law (Elia, 1997). A shipwreck can be re-surveyed; a demolished building can sometimes be reconstructed from records; a grave dug out by lamplight and sold piece by piece cannot be re-excavated once it no longer exists. This is why the trade’s own preferred metric — is the paperwork in order — measures something real but orthogonal to the thing archaeology actually loses. Paperwork can be fixed retroactively, by a change in convention, an expired statute of limitations, or a sympathetic court. A destroyed findspot cannot be fixed by any of those.

The 1970 UNESCO Convention gave the trade a different kind of bright line, and museums eventually adopted it as their own: an object is treated as acceptable if its export from its country of origin is documented, or can plausibly be assumed, to predate November 1970. The Association of Art Museum Directors made that date the operative test for member museums in its 2008 guidelines, requiring published provenance back to it or an explanation for the gap (Association of Art Museum Directors, 2008). The date answers a real question: after 1970, a museum that acquires an undocumented object is choosing to reward a market that is actively destroying sites now, in a way that acquiring something that left the ground in, say, 1920 is not. But the 1970 line was never a claim about whether an object’s context survives. An object dug up by torchlight in 1920 and sold without a word written down about where it came from passes the 1970 test as cleanly as a figure excavated last month with full stratigraphic recording and donated the same year. The date sorts objects by how recently the market’s appetite was fed. It does not sort them by how much of their archaeological content the appetite consumed.

The Euphronios krater shows how little daylight sits between the categories the trade treats as opposites. Looters opened an Etruscan tomb near Cerveteri in December 1971 and sold what they found to a dealer, who sold it on to Robert Hecht, who sold it to the Metropolitan Museum in 1972 for a reported one million dollars, supported by a provenance story — a Lebanese collector, a family holding — invented to survive exactly the kind of scrutiny the 1970 date was meant to apply. The story worked for over three decades. What it could never do was restore the tomb, whose contents and arrangement were destroyed in the digging regardless of whether the paperwork later proved sound or fraudulent. The krater was eventually shown to be illegally exported and the Metropolitan agreed to return it to Italy in 2006 (The Art Newspaper, 2006), which settled a question about title. It did nothing for the question about context, which had already been answered, permanently, in December 1971. Brodie and Renfrew, reviewing the record of museum acquisition in the decades after 1970, found American museum associations repeatedly pushing to loosen the very test meant to discourage this kind of purchase, with little organised objection from archaeologists who should have been distinguishing the two questions loudest (Brodie & Renfrew, 2005).

The strongest objection to separating them this cleanly is that the legal test was never trying to answer the scientific one, so faulting it for a bad proxy mistakes its purpose. Title law exists to settle who may possess an object and to give a source country or a dispossessed owner a route to recover it; deterrence, restitution and sovereignty are the questions it was built to answer, and it answers them reasonably well without needing to say anything about a findspot. On this view the complaint here is a category error dressed up as a discovery.

The objection is right about what the legal test is for, and it still misses what has gone wrong in practice. Nobody disputes that title law answers a real and separate question. The failure is that museums routinely let satisfying it stand in for answering the other one: an acquisition cleared under the 1970 rule is described, implicitly or explicitly, as responsibly handled, and that language borrows the authority of a scientific judgement the clearance never made. Renfrew made the same point from the opposite direction, arguing that the trade's entire notion of a clean object measures legitimacy of possession while the archaeological loss it leaves behind goes unpriced and unmentioned (Renfrew, 2000). A registry built to record legal and ethical acquisition risk has never claimed to record research value, and could not, since an object can clear it with a December-1971 findspot as empty as a forger’s signature. A museum that wants to say an object is scientifically useful has to make that case separately, on separate evidence, and most do not bother, because the legal clearance already supplies the reassuring word.

A figure with no recorded grave and a krater with a fabricated one failed in the same place, at the instant someone chose not to write down where either had been. Nothing a court, a convention, or a repatriation agreement does afterward reaches that instant. The certificate on the auction lot promises the object will not be taken back off the wall. It was never designed to promise, and cannot deliver, the one thing the wall cannot replace.