While digging through the news, I came across a piece on the physics of painting: titanium white, a fake Vermeer sold to Göring himself, and a short paragraph about the carbon-14 “bomb peak.” The journalistic summary fit into two sentences, but it contained a claim that makes you want to go to the primary source: you can fake an old canvas, but not old linseed oil. The binder in the case of the forger Robert Trotter, who had dated his painting 1866, turned out to date from the postwar era.
The summary was built around a single 2019 paper. Looking into the sources, I found that the “bomb peak” has a second life that the press barely writes about. Three years before the painting article, another group of researchers, another country, another discipline—and the same pages of PNAS—published an analysis of two hundred and thirty-one ivory samples seized from smugglers. Both paints and tusks are measured by the same method, for the same reason. The Cold War inadvertently built a global clock, and now that clock catches both brush-wielding forgers and ivory traffickers.
Radiocarbon dating was invented in 1949, and it works through decay. Every living organism contains carbon-14; when the organism dies, exchange with the atmosphere stops, and the isotope quietly dwindles with a half-life of around 5,700 years. By measuring what remains, scientists can give the age of archaeological finds to within tens or hundreds of years. For objects less than a century old, the method was historically useless: too little material decays over such a span, and the measurement drowns in its own uncertainty.
Nuclear testing turned the scheme upside down. An explosion is not only decay, but fusion: neutrons strike atmospheric nitrogen and produce the same carbon-14 as cosmic rays do, only in industrial quantities. Testing began in 1945, took off after 1950, and stopped in 1963, when the USSR, the United States, and Great Britain signed the Partial Test Ban Treaty. Over those years, the concentration of atmospheric carbon-14 nearly doubled. After 1963, the movement reversed—roughly four percent a year; half of the excess vanishes over eleven years.
In classical dating, the clock runs in dead matter. In the bomb peak, it runs in the atmosphere—and backward. Carbon dioxide carrying marked carbon is drawn in by photosynthesis, passes into leaves, flax, grain, teeth, and since 1955 every living thing on the planet has carried in its cells the mark of a specific calendar year. The flax fiber that forms a painting’s support locks in carbon in the year it grows. The linseed oil the painter uses to make paint—in the year the seeds are pressed. The protein in the binder—in the year a chicken lays the egg. This is not the date of death; it is the birth year of the material. You can buy an old support for a painting at a flea market. Old linseed oil in a tube—no.
Robert Trotter’s painting On the Coast at Scheveningen was designed to pass for the 19th century: the reverse bore the date 1866. The inscription was not lying. Dating a canvas sample yielded a window for the plant’s growth from the last quarter of the 17th century to the middle of the 20th, and the required 1866 sat comfortably within it. The pigment in the lower layer caused Trotter no trouble either: titanium white and barium sulfate only entered the market in the 20th century, and their presence establishes only a lower bound—“no earlier than.” A “no earlier than” boundary is a weak signal: the forgeries of Han van Meegeren and Wolfgang Beltracchi were built precisely around the lower bounds of when materials appeared.
The binder decided everything: a mixture of vegetable oil and protein, meaning egg or glue. A single paint stroke weighing 160 micrograms, after purification, became 58 micrograms of material, then 19 micrograms of carbon dioxide, which was turned into graphite and fed into a gas ion source mass spectrometer in Zurich. The answer was unambiguous: the binder’s carbon matched atmospheric concentrations after 1955. Since the peak has both a rising and a falling slope, the same isotope level fits two windows at once—1958–1961 or 1983–1989. In 1985, Trotter himself admitted that he had painted the work with modern oil paints.
He chose a naïve, primitivist style, with broad planes of unmixed paint—exactly the style in which this proof-of-concept method works best. His forgeries were hybrids: old supports, selected by age, plus new upper layers.
Lead white adds an almost surreal degree of precision. This pigment has been used since antiquity—Pliny described it in the 1st century, Cennino Cennini in the 15th—and the carbonate in it historically drew carbon straight from the air. Modern industrial lead coatings are depleted of the isotope: fossil raw materials lost all their carbon-14 long ago. So if the lead white in a “Rembrandt” either contains a bomb mark or looks twenty thousand years old, the experts have an awkward question on their hands. The method is silent on acrylics as well: paints made from synthetic polymers contain no carbon-14 at all, and the bomb peak does not apply to all postwar art at once.
The method also has a ceiling worth remembering: it can falsify, but it cannot certify. A reading of “after 1955” kills a painting inscribed 1866. A reading of “no earlier than 1650” does not establish that the experts are looking at a Rembrandt: the artist may have worked with old oil on an old support, and the same method cannot determine that. This ceiling is not a dead end, but a boundary of competence.
The second paper is about elephants. When border agents seize a large shipment of tusks, DNA analysis answers the question of “where”: the method developed by Samuel Wasser and colleagues assigns a sample’s origin to within roughly ±300 kilometers. The question of “when”—the time of the animal’s death—is answered by the bomb peak.
Thure Cerling and co-authors measured 231 samples from fourteen seizures of half a ton or more between 2002 and 2014. Ninety percent of those elephants had died less than three years before the seizure. One sample out of 231 had a lag of more than six years: two hundred and thirty-one months, nearly twenty years, and it came from Tridom—the border region of Cameroon, Gabon, and Congo. Behind this lies an answer to a specific exculpatory argument: that the market is being flooded with old ivory from government stockpiles amassed before the international trade ban. If that were true, the ages of seized ivory would be broadly distributed, from months to decades. Of the 231 tusks, only one even resembles that picture.
The regional pattern was more vivid than the overall conclusion. By genetic attribution of origin, the samples divide into East Africa, Tridom, West Africa, and Zambia; the earliest and largest seizure—6.5 tons in Singapore in 2002—came predominantly from Zambia, while a Malaysian container weighing around six tons in 2012 mixed ivory from three regions. Transit to the point of seizure was counted in categories: fast—up to a year, medium—from one to two years, slow—more than two. East African ivory moved quickly: roughly 35% of samples had a lag of less than a year, with a fast:medium:slow ratio of 35:41:24. Tridom moved more slowly—only 7% fast, with a ratio of 8:35:57.
Median lag rose from 7–13 months for seizures before August 2010 to 13–32 months after 2011; later shipments took longer to assemble, and the authors do not explain why, leaving the dynamic open.
The whole thing rests on calibration. The air over equatorial Africa matches neither the northern nor the southern “clean” zones, so instead of using a global curve, the researchers built their own—from elephant hair collected in known years. The linear relationship F14C = −0,0049928 × year + 11,09197, with a reliability of 0,988, works for the years 2001–2013 and cannot be applied before 1995. Elephant diets introduce distortions of their own: elephants eat bark and wood, the transfer of nonstructural carbon in perennial tissues stretches the mark across years, and proteins in keratin and collagen renew with a half-life of around 140 days. A tusk cut in the year of death carries carbon fixed several years earlier.
The history of an organ and the moment of death are different dates, and sometimes years lie between them.
It is worth pausing over the scale here. In 1949, a single date required gram quantities of pure carbon—meaning kilograms of sample. In 1980, accelerator mass spectrometry reduced the requirement a thousandfold. The Zurich study made do with twenty micrograms of carbon: one intact stroke from a painting. This is not a technical feat; it is a fundamental threshold of acceptability—the sample can be taken without visible damage to the painting.
The uncertainty, meanwhile, is measured in years. The expanded uncertainty of the tusk lag is ±1.55 years, or ±18.5 months at 95-percent confidence, and that is for measurements on a fairly steep part of the curve. Earlier studies—Vogel, Uno—dated tusks using general atmospheric curves, and their ages diverged systematically from known dates by two years in the same direction.
Cerling and colleagues did not merely surpass them in precision; they showed that the error was not in the instrument, but in the calibration: global zones do not describe the air of equatorial Africa. When two scientific schools produce different numbers for the same object, the source of the discrepancy is rarely the measurements—it is more often the comparison model they chose. Flat stretches of the curve widen the window even further: on a gentle slope, one isotope level corresponds to an entire period, like the same 1983–1989 in the Trotter case. One more caveat comes from Fedi and colleagues: in paper, the bomb peak is weakened, and chronological precision falls.
The story would be incomplete without an expiration date. After 1963, atmospheric carbon-14 declines by roughly four percent a year, and by the early 2030s the mark will fall below the sensitivity threshold of instruments. Everything born afterward will grow beneath a clean sky, without a mark. A few years after 2030, the most unusual dating tool in the history of science will run dry.
The method lives beyond forensics. It has been used to establish that human neurons do not renew over a lifetime; that around ten percent of fat cells are replaced every year; that the Greenland shark, at 392 ± 120 years old, is the oldest known vertebrate, because the mark can be read in the lens of its eye. It is also used to check annual rings in fish otoliths to within two years: the rise of the peak in 1956–1960 was steep, and a steep slope gives precise dates. It is also used to measure the harvest year in bottles of wine.
The method has a moral countdown, too: for a single megaton-scale ground explosion, Andrei Sakharov estimated in 1958 roughly six thousand six hundred deaths and around ten thousand cases of cancer and genetic disease from carbon-14 alone. Atmospheric tests left around 545 megatons in total, and the expected harm from that carbon stretches eight thousand years into the future. The clocks used to catch ivory traffickers are an application of the very same event that signed that price tag.
The question behind the hook—can physics catch a fake Vermeer?—has a precise answer: yes, if the painting contains anything organic born after 1955. But the answer is followed by a sober no. No dating method proves authenticity. No instrument can distinguish an honest Rembrandt from a magnificent imitator if the imitator worked with old linseed oil on an old support. The bomb-peak clock counts down to the early 2030s—and then comes an age in which faking an old canvas and fooling physics will once again be merely a matter of technique.