
Art attack: spot the forgery
Image credit: Alamy
Museums, galleries, collectors and auction houses are turning to science and technology to help differentiate a genuine art work from a forgery.
From Rembrandt to Monet, Da Vinci to Picasso, high-value art works have always been a target for criminal forgers. However, now AI; ultra-high-resolution photography; electron microscopy; materials science, and multi-spectral cameras are helping to differentiate the copies and forgeries from the real, assisting art historians and valuers in deciding what should be in the museum basement and what should be proudly on show.
The art market, worth $64bn (£45.3bn) in 2019, has been described as the third most unregulated market in the world today, behind drugs and sex. With nine-figure sums being regularly achieved at auction for fine art, a great deal is at stake. Experts cannot agree on the scale of the forgery problem, but a 2014 report by the Swiss Fine Art Expert Institute estimated that over half of all the artwork in circulation is fake, and although some dispute this, others are not surprised. Given the scale of intentional deceit, experts are often loath to discuss the results of their investigations, especially those conducted for private collectors.
Traditionally, authentication of art works relied on expert connoisseurs who specialised in the output of specific artists, or with expert panels formed after an artist’s death to protect their legacy from fakery. Their learned but subjective opinions, combined with a work’s provenance – a paper trail akin to a used car’s service and ownership history – are increasingly supplemented by scientific analysis. The technical skill required to fake a Da Vinci or a Rembrandt is huge, requiring access to old canvasses, pigment recipes from the master’s studio and an inherent artistic talent at genius level. Sensibly, many forgers have turned their attentions to the works of 20th-century artists who used paints and canvases that are still readily obtainable and whose work is arguably easier to mimic.
One infamous recent exhibition of the paintings of the Italian artist Modigliani in 2017 at the Ducal Palace of Genoa closed early after it emerged that 20 out of 21 paintings on display were counterfeit. The artist is a favourite with forgers, who both copy known works and create new art in the style of Modigliani with his signature elongated faces. The confirmation of a genuine Modigliani is complicated by the artist’s short and chaotic life. He died at the age of 35, often giving away drawings in exchange for a drink. In 2018, one of his nudes fetched $157m (£111m) at Sotheby’s in New York. He remains one of the most copied artists of the 20th century.
Sky-high inflation of values at auction have attracted ever more sophisticated forgers. Bruised by their expensive encounters with fakers – Sotheby’s, for example, offers a five-year refund guarantee if a work sold at auction is later found to be inauthentic – the leading auction houses have bought in or developed in-house expertise to scientifically test the authenticity of art before it goes on sale, and bolster the opinions of their expert valuers.
Many galleries and collectors have to date been coy about allowing scientific examination of their collections, or at least talking about the results. Several scientists that E&T spoke to were unwilling to go on the record with the results of their investigations into the authenticity of work in important private and public collections, being bound by confidentiality agreements designed to avoid unnecessary embarrassment.
A nuclear giveaway
Art produced after the beginning of the nuclear age in 1945 can be definitively identified by the presence of the radioactive isotopes Caesium 137 and Strontium 90. These elements of the atomic age were first released into the Earth’s atmosphere following nuclear explosions and tests.
The 20th-century work of the Russian Avant-Garde attracted many forgers in the 1960s and beyond, as values began to increase. Elena Basner, a Russian art historian, had been spending a lot of time trying to weed out fakes from the collection at the Russian Museum in St Petersburg, where she worked at the time.
In a collaboration with physicists and chemists, the team patented a technique using mass spectroscopy to identify the post-atomic isotopes, positively identifying post-war imitations, which far outnumber the highly valued originals produced between 1900 and the 1950s. The isotopes appear in oils used to carry the pigments in oil paints, absorbed by the linseed plant as it grew.
It can also be used to date paper, wooden supports and paints to a resolution of just a few years between the late 1940s and the present day.
Unusually, a 2021 exhibition of the Russian Avant-Garde at the Museum Ludwig in Cologne takes the opposite approach, breaking the taboos around ‘inauthentic’ work. Counterfeits discovered in their own collection are displayed in this exhibition alongside the original paintings, or works that served as templates for the fakes. Fakes flourished in this genre of paintings because they were banned in the Soviet Union by Stalin who both disliked them and considered them subversive. Paintings were often smuggled out of the country in diplomatic pouches to protect the identities of the sellers, resulting in a broken paper chain of ownership or provenance. These gaps in the paperwork presented ripe conditions for forgers to exploit the chaos.
Petra Mandt, curator of the exhibition, says: “It is time to pull back from a reliance solely on expert connoisseurs based on style and provenance, and broaden our analysis to include technical examination by scientists, combining the two to judge authenticity.” Selected paintings from the museum’s collection that were suspect were compared to work that was without doubt genuine work by the artist in question.
Mandt describes some of the techniques deployed to expose the fakes. As a first step, visible light photography of the image is augmented by ultraviolet (UV) imaging. The fluorescence of some materials under UV light can give an indication of materials used. Infrared reflectography can reveal any under drawing as carbon-based marks like graphite are opaque to infrared. A layered X-ray can show each level of the painting and any subsequent reworkings including the brushwork, which is important to verify a particular artist’s style. Microscopic analysis can detect an artificial patina applied by forgers to ‘age’ the painting’s surface and can also show whether a signature was added at a later date.
Jilleen Nadolny is the director of the UK office at ArtDiscovery, a company based in London and New York that uses science to support the authentication and attribution of art works. The team there uses a multidisciplinary approach combining technical art history with scientific analysis of paintings and other objects. She is keen to stress that there is no magical quick-fix technique to determine fake from genuine works. “A lot of our cases are confidential due to the high-profile nature of the work, but our team has identified many forgeries over the last 10 years,” she explains. She will say on record that she has seen many works by the prolific German forger Wolfgang Beltracchi, who claims to have forged hundreds of paintings by more than 50 artists, as well as forgeries of unknown medieval painters to forgeries of art world superstars such as Jackson Pollock and Vincent van Gogh. There are few artists of any renown who have escaped the attentions of forgers.
One technique commonly deployed at ArtDiscovery is Raman spectroscopy, a non-invasive analytical technique which can be used to identify the pigments used in a painting. A laser is focused onto microsamples taken from an object and the light scattered back measured. Every pigment that is Raman active will produce a particular spectrum, which can be referenced against the company’s proprietary spectral library – one of the world’s largest.
Artists from the Russian Avant Garde painted from the later 19th century to the mid-20th century. Raman spectroscopy can identify pigments such as rutile form titanium white, which was not available in the Soviet Union until mid-century. Meanwhile, phthalocyanine blue was not used in Soviet-made oil paints until 1966. The presence of either of these pigments can help to definitively place the creation date of a painting long after its supposed date.
“Forgers are getting smarter, and we have to take that into account and use protocols that combine many different approaches,” Nadolny explains. “Fortunately for us a painting is a complex object made up of many different materials – the canvas or support, the paint and pigment and the binding media (the liquid – oil, glue, polymer, etc that ‘binds’ together the pigment to make the paint). It’s a lot for a forger to accomplish without making a mistake that we can identify as an anachronism.” Brush hairs made of synthetic fibres embedded in the paint, or the presence of canvases or supports containing manmade synthetics like polyester or nylon are giveaways that can help to reveal the actual date that a painting was made, in contrast to the date that it is claimed to have been made.
It is not only in the confirmation of fakeries that technology can be deployed. It can also confirm that a piece of art is either the work of a great artist, a work of a pupil in their studio or made after their lifetime as a copy.
Matthias Alfeld, assistant professor of X-rays in Art and Archaeology at the Delft University of Technology, was among the first scientists to apply non-destructive X-ray fluorescence technology to the analysis of the pigments in old master paintings, and developed the first mobile scanner for examining art.
X-ray Fluorescence (XRF) can map the distribution of metallic elements in the salts that make up the pigments in oil paints, and penetrate layers of paint, that might have been reworked or overpainted by later conservation efforts.
Revealing the presence of an anachronistic pigment can point to the work of a forger. “One of the paintings that we looked at in detail was Rembrandt’s Saul and David, which art historians could not be sure was painted by the master himself as it was painted over a number of years between 1651 and 1658, and is unsigned,” says Alfeld. The artist also used quite different pigments and styles in the two phases of work – XRF identified two different Cobalt-containing blue pigments, for example.
To compound the issue of authentication, someone had chopped the painting in half in the 19th century, perhaps to fit it into a smaller room, or to display the two key figures as separate paintings, an early version of buy one get one free. There were also areas of replacement patches and overpainting that had to be investigated before restoration of the painting now hanging in the Mauritshuis collection in The Hague could begin.
After years of work, the painting can be confidently attributed to Rembrandt himself, rather than an anonymous pupil in his workshop. “Our job is to deliver analyses to the art historians. We interpret the results together. The technology is fascinating but will never replace the role of the art connoisseur,” Alfeld adds.
At the top end of the market, where multi-million dollar sales are the norm, the art world cannot ignore the science. Twenty years ago the word of the curator or connoisseur was all that was needed to confirm the legitimacy of an art work. But today’s buyers – the likes of hedge fund managers investing in art – now routinely insist upon a thorough technical examination and certification of a piece before laying out the cash. As well as confirming fakes, scientific testing can also add value. For example, in 2015, ArtDiscovery was able to produce technical imaging (using X-rays) that helped turn a painting bought for $5,000 (£3,500), described as ‘in the style of John Constable’, into a $5m (£3.5m) work that was recognised as painted by the master himself. Bargain!
AI says no
Swiss company Art Recognition uses artificial intelligence to assess the authenticity of a work objectively by analysing an artist’s brush strokes and techniques in a high-resolution photographic reproduction of the painting.
In 2019, they were asked to apply their AI analysis to a van Gogh painting. Art Recognition have developed a deep convolutional neural network – a kind of machine learning adapted to work with images. This is first trained on images of undisputed works. To help the AI detect forgeries, the system is then fed with well-known fakes and forgeries.
The company was asked by a leading van Gogh expert to investigate a disputed self-portrait from the Norwegian National Museum. To help human experts interpret the AI’s results, the algorithm highlights hotspots on the painting in red which are most important in its decision making. Uncoloured areas have the smallest influence on the final decision. The AI assigned the self-portrait as a 97 per cent probable authentic van Gogh.
Weeks later, the official announcement by the van Gogh Museum in Amsterdam confirmed its authenticity after several years of study. The firm said AI has the advantage of being utterly objective and can deliver results quickly.
Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.