What is a Certificate of Authenticity, anyhow, and where did they come from?

The past, the present, and the hopeful future of the most misunderstood piece of paper in the art world.

by Alejandro E. Trimper

Author's note. Let me put all the cards on the table before we start, because regular readers know that is where I like to put all my cards. There are people with law degrees, PhDs, laboratories, and decades inside artist foundations. I am not one of them. I have, however, spent about twenty years in the gallery and art advisory business, longer if we count the years I spent listening to my mother conduct deals before I understood what I was hearing. I have held a lot of paper, watched a lot of deals close, and watched some go up in smoke with that eerie silence that tells you everyone in the room has just learned something expensive. Lived experience does not replace scholarship, law, or science, but it notices things a classroom or back office can miss.

This is lived experience informed by public records, not legal advice or a self-coronation. Except for the historical examples, public cases, source names, and business disclosures made openly here, the critique is structural: I am one art lover talking to you, hopefully another, about the history of this document, what it is being asked to do, why it often cannot do it, and what a more honest system might look like if the scary men behind the curtain ever allow the current one to change.

Why I'm writing this

A few weeks ago, I published What a COA Should Actually Tell You, a buyer's guide to the document in your hands before you write the check. The response was the same question phrased a dozen ways: okay, but where did this thing come from, why is it such a mess, and where is it going?

Fair questions. The buyer's guide tells you what to ask now. This companion piece explains how the COA came to exist, what the present system gets right and badly wrong, and where I hope it lands a decade from now. A law-review version would be easy and completely unlike me, unless your idea of a wonderful evening is footnotes before dessert. So imagine we're sitting comfortably in our new offices in Greenwich, CT, a glass of your preferred beverage in hand, a COA on the coffee table.

PART ONE. THE PAST

Before the certificate, there was the chain

For much of art history, there was no stand-alone COA because the market did not operate at today's scale or distance. A patron commissioned a painting, the master or workshop produced it, and the work moved through people both sides knew. Authenticity was answered by the human chain between the brush and the wall.

That chain has a name: provenance, the documented history of a work's ownership and movement. Surviving records might include a bill of sale, estate inventory, dealer's letter, customs entry, exhibition label, catalogue listing, or photograph. Together, they formed a story that could be checked. The story could still be wrong, incomplete, or improved by an enthusiastic dealer, but authenticity lived in a network of evidence, not in the title printed at the top of one page.

The expert enters the room

When the chain broke, the market brought in a connoisseur: a scholar or dealer with deep experience and a famous eye. Giovanni Morelli studied the details artists rendered habitually, ears, hands, fingernails, curls of hair, the bits a copyist might overlook. Conscious style can be copied. Habit is harder.

That tradition created enormous knowledge and concentrated enormous authority in human beings, who arrive with loyalties, blind spots, reputations, and bills to pay. Bernard Berenson, the leading attributor of Italian Renaissance painting in his era, had a secret financial arrangement with dealer Joseph Duveen. A 1912 contract reportedly paid Berenson 25 percent of the profit on pictures Duveen acquired on his advice. It was more complicated than the dinner-party word kickback, and Berenson was a serious scholar. The structural point remains: the opinion creating value was tied to the transaction.

Let that sit for a second.

The market has been trying ever since to solve the same problem: how do you use expertise without pretending the expert's incentives disappear? The answer is not to throw out experts or pretend a structure can eliminate every interest. It is to disclose incentives, explain the reasoning, identify the evidence, and make the conclusion capable of challenge. The industry is still working on that, and honestly, it always will be.

Then the piece of paper appears

The portable, signed COA familiar to buyers became commonplace as the modern market expanded during the twentieth century. Auction houses scaled up. Dealers shipped works across oceans. Buyers and sellers stopped knowing one another. Insurers, customs authorities, lenders, tax authorities, estate lawyers, and museum registrars all needed documentation that could travel with the object.

Then prices became large enough to make forgery a business model. Han van Meegeren's fake Vermeers remain the obvious warning flare, but hardly the only one. In 1943, through an art broker, Hermann Göring acquired Christ with the Adulteress for 1.65 million guilders. Once a painting could buy a house, human nature made the rest predictable: somebody was going to start making forgeries in a basement.

Artists signed certificates. Galleries issued them. Estates and foundations created review processes. Publishers developed print conventions. Catalogues raisonnés became authorities in their own right. In some markets, a genuine work with a thin file became harder to sell than a questionable work carrying a confident-looking certificate. That inversion, the paper outweighing the thing it describes, is the seed of almost every problem that follows.

What the COA got right, and why no two look alike

Credit where credit is due. The COA did work the old relationship-based market could not do at scale. In New York, when an art merchant furnishes a COA or similar writing to a buyer who is not an art merchant, the document creates an express warranty for the material facts stated as of the sale. In normal English: a seller-issued COA can give the buyer contractual recourse if the work is not what the seller represented. That matters, even though a warranty still depends on its terms, the law, and the seller's ability to answer for it.

Documentation also helps insurers, lenders, museums, and future buyers. It is central to Nazi-era provenance research and restitution. The hero is not one magic page. It is the file: records that can be found, compared, challenged, and understood.

There is no market-wide form for a COA. A 1985 studio note, a signed photograph, a publisher's chop, a one-page gallery certificate, a foundation document, and an independent report can all be valid in context. Era, medium, and market history matter. Papers are lost in moves, separated during reframing, flooded in basements, thrown away during estates, and abandoned when galleries close. Missing old paper is not proof of authenticity or forgery. It is evidence that time happened. The question is whether what survives makes sense for this artist, work, date, medium, route through the market, and issuer.

The COA developed like an urban sprawl with no architect or design, leaving modern buyers blindly navigating a chaotic landscape without a map.

PART TWO. THE PRESENT

The paper is carrying too much

COAs are such a holy mess because the market asks a one-page document to do the work of a relationship, warranty, provenance file, scholarly opinion, and forensic report. That is too much. A serious COA takes the reader through a process and requires substance, often more than one page; a weak one relies on its title and little else. That difference is the whole game.

When the seller and the judge occupy the same chair

Sellers often try to certify their own inventory. Sometimes the document is careful, specific, and backed by substantial data and research. Sometimes it is one sentence on letterhead with a signature large enough to distract from the glaring void of any real information. Both are called Certificates of Authenticity.

A seller-issued certificate is not meaningless. In New York, it may be legally significant as a warranty. It is also not independent proof. The person making the representation has a financial interest in the sale. That is not an accusation; it is a structural fact, and structural facts belong in the light, not under nice typography hidden in the shadows.

The same caution applies to estates and foundations. They may have unmatched archives, studio records, family knowledge, and scholarship, along with fiduciary, legacy, reputational, or market interests. That does not disqualify their work. It means authority should come with a visible methodology, reasoned conclusions, and full disclosure of who decided what and why.

Then the doors closed

For several major artists, the formal review channel collectors once relied on was slammed shut. The Warhol board was authorized to cease operations in 2011. The Basquiat committee disbanded in September 2012 after eighteen years and more than two thousand reviews. The Haring Foundation also dissolved its authentication committee in 2012. Their histories differ, and each had its own drama, but the warning to the market was clear: even an authenticator who wins can spend a fortune getting there.

The New York City Bar warned that litigation costs were driving scholars away and supported legislation intended to protect good-faith opinions. Versions of that proposal have been introduced in every two-year legislative session since 2013 to 2014; as of this writing, the current bill remains in committee. IFAR, after more than fifty years of service, has also brought its work to a close. A market already short on independent places to ask difficult questions now has one fewer.

Authentication did not disappear. It fragmented. Some responsibility moved to catalogue raisonné projects, which can contain errors, omissions, and disputed attributions; some to independent specialists; some to conservation science; some back to auction houses and dealers; and some into silence. Silence in this market is rarely neutral and never a good thing for the average Joe.

When authorities disagree

A catalogue raisonné is a scholarly record of an artist's accepted work. In practice, inclusion can function like a super-certificate, while non-inclusion can make a work commercially untouchable. Many catalogues are extraordinary works of scholarship, and digital projects have made some more open and revisable. Their methods, standards, conflicts, and revision policies still matter because a scholarly record can also determine a market.

A collector holding a COA for a work omitted from a respected catalogue raisonné, or a catalogue entry with no surviving certificate, has a disagreement between authorities, not a tidy answer. Do not count documents like votes. Ask what each conclusion rests on and which evidence can be tested.

Independent should be a description, not a mood

A meaningful independent opinion should disclose who retained and paid the expert, what was examined, whether it was seen in person, what records and comparables were used, what technical work was done, what limits remain, and what other financial relationships exist. The conclusion matters. The path matters more. A reasoned report can be challenged and revisited. A yes-or-no sentence asks for trust without showing the work.

In my own gallery practice, I have retained Richard Polsky Art Authentication at standard fees when outside review was warranted, especially for my Keith Haring subway-drawing collection. That is a disclosure, not an endorsement of every conclusion. Trimper Gallery and AET Group LLC have no other business, compensation, referral, or financial relationship with Richard Polsky Art Authentication. We have hired Richard to assess works from our own collection and have referred clients to him independently of the gallery. A piece arguing for disclosure should not make the reader hunt for it.

When science walks into the room

For much of the twentieth century, authentication was often framed as a connoisseurship problem: does the brushwork look right, does the composition fit, does the work have the right feeling? Those questions still matter. Another set has entered the room: what is this object made of, and could those materials have existed when and where the work claims to have been made?

The Knoedler scandal is the modern cautionary tale. The gallery closed in 2011 after roughly 165 years. Prosecutors said Glafira Rosales sold more than sixty fake modern works to two New York galleries, and buyers paid more than $80 million. The Rosales works arrived without documented provenance, and technical testing in several disputes identified materials or treatments inconsistent with the claimed histories. The evidence did not need to identify the hand holding the brush. It only had to show that the object and the story did not agree.

The lesson is not that the eye is useless and the lab infallible. That is a satisfying movie ending and a bad professional standard. Visual judgment, documentary research, and technical analysis catch different errors. Let them disagree before a sale, not in court after one.

PART THREE. THE FUTURE

Two ways of knowing, and the answer is probably both

The debate is often staged as the trained eye versus the machine. It is the wrong frame. The archive asks where the work has been. The eye asks whether it fits the artist's practice. The lab asks whether the object can physically be what the story claims. A computational model asks whether measurable features resemble a defined corpus. The future is scholar plus scientist, with the researcher, conservator, lawyer, artist, and buyer able to see what each actually did.

What science can prove, and what it cannot

Technology rarely proves that a named artist created a particular work. It is better at testing whether the object can be what the claim says it is. A pigment, binder, or support introduced after the claimed date can change the conversation.

Context still matters. A later pigment may come from restoration; a modern adhesive may belong to a lining; a sample may miss the relevant area. Instruments produce data, and trained people interpret it. Science is strongest when method, sampling, uncertainty, and limits are written down. A technical report should be capable of repetition and challenge. That is the point.

What artificial intelligence changes

Machine-learning attribution is no longer hypothetical. Published studies have trained models on documented paintings and drawings using image features, line behavior, texture, and other measurable patterns. Results vary by artist, medium, corpus, image quality, and test design, exactly the sentence that disappears when technology reaches the sales brochure.

The honest claim is not the computer knows. It is that, for a defined artist, medium, image type, and dataset, a model may offer a reproducible consistency check. Show the training and test sets, image requirements, threshold, uncertainty, and error rates. Without them, a percentage is decoration.

The source images being used are the vulnerability. Feed a model misattributed works, and it learns the mistake. Train it only on pristine museum images or on works from a single period, and it may punish a real work photographed poorly or created during a stylistic shift. AI should be another instrument with a visible methodology. Consistency with the documented source files is important, and a fundamental understanding that even AI has its limitations and can and will be wrong. There will be no AI coronation, at least not anytime soon.

What a digital registry can do, soberly

The word blockchain has been attached to enough art-market promises to require a small period of silence before anyone uses it again. I am not talking about turning every painting into an NFT or pretending technology can make weak provenance strong.

A bad fact on a blockchain is still a bad fact.

The useful idea is a versioned, tamper-evident record that preserves what the file said at each point. A decade ago, in a business plan for the Global Certificate of Authenticity Registry, I called this an object's virtual biography: images, dates, materials, ownership transfers, condition, exhibitions, reports, and sales history gathered around one work. That remains the durable idea. The archive cannot authenticate a bad starting record, but it can expose later changes and keep evidence from quietly losing a page between sales. A QR code, seal, or tag can link the object to the file. It cannot prove that they belong together.

Prevention begins with living artists. Before a work leaves the studio, the artist or authorized studio can record identifying images, date, dimensions, materials, inscriptions, edition information, process evidence, and the first transfer. Later owners can add exhibitions, conservation, sales, and condition changes without overwriting the origin record. A future fake would then have to counterfeit more than a signature. It would have to counterfeit evidence created before the fraud existed.

The coliseum for the works already here

The inherited market is harder. Countless works already circulate with dead witnesses, missing archives, competing certificates, and stories strengthened mainly by repetition. They cannot simply be uploaded and declared clean. When duplicate, triplicate, or otherwise conflicting works appear, bring them into the same arena. Let the claims fight, not the people. Compare provenance, construction, materials, imagery, signatures, labels, publication history, and technical data against one another and the documented corpus.

Some works will emerge well supported. Some will be misattributed, altered, assembled, or false. Some will remain unresolved. The point is a record of what survived, what failed, and why. A rejected or disputed work should not vanish, subject to appropriate legal and privacy controls, only to return with a new owner, a new certificate, and a conveniently shorter history. The ambition is not zero fraud. It is to make fraud harder to plant, contradictions easier to expose, and known fakes harder to recycle.

What I hope a COA looks like in 2036

I hope it looks like a complete digital file, not a magic sheet of paper. Keep a one-page cover, people like cover sheets, and there is no reason to fight every human instinct at once. Behind it: provenance with documented gaps; named scholarly analysis; technical examination when appropriate, with method and limits; a computational check where a reliable corpus permits one; strong images; version history; conflict disclosure; and a warranty with defined terms and remedies.

Review should be proportionate. Not every work needs hyperspectral imaging, X-ray fluorescence, a lawyer, three scholars, and a closing binder large enough to injure someone. Value, risk, artist, medium, and evidence should determine the file. State what was done, what was not, who did it, and why. A disclosed limitation is not a defect. An undisclosed one is where trouble starts.

I also hope the system becomes comfortable saying uncertain. The market treats uncertainty as a sales failure, so it gets hidden inside confident language. In scholarship, uncertainty is information. A file that says this gap remains unresolved or this result is consistent but not conclusive is treating the buyer like an adult. The same goes for conflicts: who paid the expert, who owns inventory, and whether the lab or registry profits from the sale. Those facts do not automatically invalidate an opinion. Hiding them weakens it.

Back to the document in your hands

The practical questions are still the questions in What a COA Should Actually Tell You. Who issued this? What did they examine? What evidence did they rely on? What did they not examine? What conflicts exist? What warranty survives the sale? Does the document show its work?

A Certificate of Authenticity is not proof because the title says certificate. It is a document about a process, and only as good as the evidence, people, and structure behind it. You can still buy art with confidence. Confidence should simply be earned rather than typeset. A seller should walk you through the file without making you feel rude for asking, and you should walk away when the answers are not there.

The past gave us a piece of paper that tried to do the work of a relationship. The present asks it to do the work of an investigation. The future, if we build it well, can preserve the investigation itself: a living archive for new works, an arena of evidence for old ones, and a certificate that finally has something real to point to.

The romance of collecting does not come from trust without evidence. It comes from trust earned through evidence. You still get to fall in love with the work. You simply get to know what you are bringing home.

Your future self will thank you for asking the harder questions now.

Alejandro Eduardo Trimper is the owner of Trimper Gallery, a collector, and someone who often thinks of himself as having a passionate love affair with Keith Haring's subway drawings. He writes on documentation, transparency, the return of passion to collecting, hospitality, place, and the impact technology is having on galleries.

Trimper Gallery does not authenticate artworks or offer opinions on authenticity. The Gallery has retained Richard Polsky Art Authentication at standard fees when independent review was warranted; no other business, compensation, referral, or financial relationship exists between Richard Polsky Art Authentication and Trimper Gallery or AET Group LLC.

Nothing in this essay is legal advice. Except for historical examples, public legal sources, public cases, source names, and the business disclosures made openly here, no factual accusation is made against any unnamed entity. The critique is structural. The hope is structural.

Unprompted. Essays on art, collecting, hospitality, place, and the life built around looking closely.

Search This Blog