In the morning news digest, among the usual stream of space launches and neuropop, a single line appeared: a team from the Scripps Institution of Oceanography at UC San Diego had learned to distinguish fake ancient pottery by Earth's magnetic field. It sounded like a typical "science news headline" — bright, brief, no details. But in the adjacent line sat a number that made me pause: 112 °C. A threshold temperature — too specific for advertising and too boring for fiction. Behind such numbers, real science usually hides.
It was worth digging if only because a detective story was embedded in the method: the PNAS article itself mentions the trial over the "James Ossuary" — a seven-year process over a high-profile biblical artifact that ended in nothing because experts couldn't prove forgery. And the experiment description featured souvenir shops in Jerusalem's Old City: researchers went out on the street and bought guaranteed modern pots to teach the test to see the difference. This is no longer abstract geophysics — this is almost noir.
The contrast made the final topic choice. For a century, paleomagnetologists threw out the weak secondary signal — viscous remanent magnetization, VRM — as annoying noise interfering with measuring the main thing. And now this "trash" suddenly turned out to be a date. A story about how a discarded measurement artifact returned to science as evidence — that's a rare genre, and worth it.
Let's start with the main trick holding up all of archaeomagnetism. Clay contains ferromagnetic minerals — primarily iron oxides like magnetite and hematite. In tiny single-domain grains of such minerals, magnetization isn't a "material property in general" but an arrow: the magnetic moment either points one way or the opposite, and flipping it takes effort because an energy barrier lies between the two states.
When a pot is fired, grains heat so intensely that thermal energy easily flips the arrows over the barrier — they flail while it's hot. Then the piece cools in Earth's magnetic field, and here's the trick: the arrows freeze one by one, each at "its" temperature, and lock in the direction of the field present at the moment of cooling. This is thermoremanent magnetization, TRM — strong, stable, nearly eternal. If rock cooled 800,000 years ago when the planet's field was reversed, its crystals still point south.
The key word is "nearly." In Louis Néel's 1949 theory, on which all rock magnetometry still stands, each arrow has a relaxation time: the probability of thermal reversal depends exponentially on grain volume, its coercivity, and temperature. Large grains are locked solid. Small ones — on the boundary between "locked" and "still flipping" — continue slowly fidgeting: in Earth's field, in a fixed position, they gradually align along it. Thus a weak second signal accumulates — viscous remanent magnetization, VRM. Not in minutes like TRM, but over centuries and millennia of immobility.
Actually, that's the whole secret. TRM writes where the field was on firing day. VRM writes how long the object lay after being pulled from the kiln and set down. One pot carries two magnetic layers — like an X-ray with two images taken a thousand years apart.
The weak secondary signal tormented paleomagnetologists for decades. You're trying to measure the ancient field — and there's a parasitic component in the data, magnetization that "accumulated" after firing or rock cooling. It's erased first thing: stepwise thermal demagnetization in zero field, temperature rising in steps, and everything that falls off at lower steps is declared noise. That's what textbooks taught, that's how labs worldwide worked.
This "noise" has a venerable history though. The first observations of viscous magnetization date to the late 19th century, and ironically were made by the discoverer of archaeomagnetism himself: Giuseppe Folgheraiter in Rome studied the magnetization of Roman bricks depending on how they were laid after firing — and noticed that over centuries bricks tend to "reorient" toward the modern field. Louis Néel provided theory for this in 1949. Since the 1970s, geologists applied VRM dating to rocks — for example, to estimate when a landslide shifted a layer. But accuracy was so-so, and the method remained niche.
So why did it suddenly take off with ceramics? In the PNAS article, the authors honestly lay out the reasons point by point. First, computational power: micromagnetic modeling of single domains, on which method calibration depends, used to be prohibitive — but now a package like SDCC in a Jupyter notebook simulates how a hypothetical pot accumulates VRM over 0.5, 10, 100, 1000, and 2500 years. Second, ceramics itself is an ideal carrier: it has many low-coercivity grains, its strong and uniform TRM is easily separated from weak VRM, unlike rocks that magnetized over geological epochs and carry dozens of superimposed signals. Third — and this is the tastiest part — the method doesn't even need high accuracy. An ancient pot and yesterday's fake differ in VRM component by orders of magnitude. No need to measure the century — it's enough to distinguish a millennium from zero.
To teach the test to distinguish genuine from fake, you need reference standards from both sides — and here begins the most real-life part of the research. Archaeologists have reference antiquity: 15 royal Judean storage jars with royal seal impressions — large vessels of the Judean Kingdom's administrative system, whose age is tied to political regime changes, plus 4 ancient pottery fragments from controlled excavations.
Reference newness had to be manufactured. Potter Lyor Bar-Sovic made six modern vessels from five types of clay and fired them in two traditional kilns — to differ from ancient technologies in nothing but age. Sixteen clay seal impressions were cast specifically for the research (work of Rachel Peltz) and fired in four electric kilns. And finally — four pots were simply bought in souvenir shops of Jerusalem's Old City: mass tourist merchandise, ancient in style and a week old in fact. The authors directly call this move "guaranteed newness": the only "certificate" no forger can avoid.
In total 45 samples, plus three artifacts of questionable authenticity confiscated by the Theft Prevention Unit of the Israel Antiquities Authority. Each sample — a small fragment embedded in a ceramic crucible — underwent stepwise thermal demagnetization: from 50 °C in 10 °C steps, at each step the sample is held for half an hour, then residual magnetization is measured on a cryogenic magnetometer. In parallel, all this digital forest was run through the Python script VRTC, which determines by two independent methods — by the break in the moment curve and by the behavior of magnetization direction — the temperature at which VRM disappears completely.
The result is simple and beautiful. In all modern samples — souvenirs, electric kiln impressions, traditional pots — VRM is small or absent and erases below the threshold. In ancient ones, VRM is significant and holds until heating above 112 °C: the magnetic memory of centuries of immobility is physically stronger than memory of a week. Micromagnetic simulations gave the same threshold independently of the laboratory — and this is already double confirmation: not sampling accident, but physics.
Now about why all this is needed. Forgeries of antiquities aren't exotic — they're an industry with a century-long history: Israel, China, Mexico, any country with rich archaeology — as co-author Lisa Tauxe puts it, "especially where treachery flourishes." And here's what's important: this market is smaller than commonly thought. The loud claim "third largest illicit market after drugs and weapons" is a factoid, dissected to the bone in the journal Antiquity: according to RAND and European Commission estimates, the entire antiquities market, including legal, is hundreds of millions of dollars per year, not "4–11 billion." But the problem isn't volume — it's that forgery devalues museum shelves, scientific conclusions, and criminal cases.
Existing methods are vulnerable, and the article names them honestly. Thermoluminescence, the main forensic tool of recent decades, measures accumulated radiation dose — firing resets it to zero, and luminescence on reheating is proportional to age. Elegant, but forgers long ago learned to artificially irradiate modern imitations, simulating millennia of dosage. Studies from the 2020s are already learning to catch the artificial irradiation itself by luminescence shape — this is an arms race in pure form. Petrography determines clay type — but the forger simply takes "correct" local clay.
The magnetic method adds to this race a dimension that can't be faked with a "quick trick." You need to fake not the firing or dose — you need to reproduce three thousand years of motionless lying in Earth's changing field, with correct geometry and correct history of secular variations. Field direction drifts over millennia, intensity jumps, and the VRM signal honestly recorded all these turns. A hacker on a forum suggested "surrounding the workshop with magnets" — but that only creates primary magnetization; you can't buy the secular component that way, it accumulates only with time. The authors write directly: the goal isn't to replace other methods, but to add to their arsenal a dimension that makes forgery so expensive it discourages attempts.
And the method already worked. Three confiscated artifacts — an oil lamp, a jug with fire traces, an animal-shaped vessel — all passed the test as ancient, Iron Age, up to three thousand years old. Moreover, the jug with a blackened bottom is behaviorally interesting: it was reheated after production — probably in an ancient fire — which partially erased the primary signal, but the secular component accumulated afterward still fell within the ancient range. The test didn't just say "ancient" — it reconstructed the object's biography.
An honest analysis must show boundaries. First — the secular threshold. The 112 °C threshold is calibrated for "older than a thousand years versus younger." Medieval ceramics can't be checked this way: both authentic and fake will pass identically. The authors must honestly admit: for younger objects, a different threshold is needed, and it remains to be found.
Second — one-time use. VRM erased by heating doesn't restore: the evidence is spent. The test can be done on a small chip, and many ancient vessels are already glued from fragments anyway — but for a fragile or waterlogged artifact this is a risk. Thermoluminescence also consumes the sample, so this is parity rather than regression.
Third — blind spots of the physics itself. If an ancient pot after firing accidentally lay in the same orientation as in the kiln — say, remained lying where it fell when the city was destroyed — it will barely accumulate VRM, and the test won't give a decisive answer. If two independent indicators inside the script give contradictory or borderline results — the method is silent. These are rare cases, but in a courtroom the edge case matters: the "beyond reasonable doubt" standard of proof forgives noise but not ambiguity.
And fourth, most ironic: forgers read the same articles. Lead author Yoav Vaknin says this without euphemism: "on one side scientists who publish results, on the other — forgers who work in secret. It's hard for us to stay ahead: they read our work and figure out how to fool our methods." The verification discipline is an eternal race, and the new method merely takes the baton on the next lap.
This story has a deep historical layer that press releases missed. In the early 20th century, long before radiocarbon, French geophysicist Bernard Brunhes — the same one who would later discover planetary magnetic pole reversals — explained why magnetic dating is problematic in principle through the famous 1895 Montparnasse station disaster. A locomotive that couldn't stop crashed through the station facade, cutting electric clock wires — and all station clocks stopped at the same minute. By the stopped hands a year or a century later, you can learn the hour of catastrophe. But not the date: the hands could have made any number of revolutions in the elapsed time.
Same with magnetic field: its direction and intensity change cyclically, and identical values return. In Western Europe, for example, Roman period directions coincide with modern ones, and 6th-century intensities with 9th–10th centuries. The hand returned to the same mark — but how many revolutions did it make? Brunhes concluded: magnetic recording can say "these two events aren't simultaneous," but cannot itself produce a calendar date — external anchoring is needed.
The paradox is that the modern PNAS method bypasses this trap precisely because it doesn't try to date by a single field value. It looks not at "where the hand is" but at "how many years the hand crawled to this mark" — at the relaxation physics of the process, not the pointer position. And where calendar anchoring is needed, archaeomagnetism for the last twenty years solves the problem from the other end: secular variation curves built from thousands of dated samples allow refining dates from magnetic recording where radiocarbon gives no resolution. A pair of papers on Judean material showed this both ways: royal seal-bearing jars recorded a sharp field spike in the late 8th century BCE, and by the early 6th century the field lost about 27% strength in thirty years — and in Jerusalem, archaeomagnetic analysis of 54 floor fragments from an elite building collapsed in the Babylonian fire of August 586 BCE gave an anchor point for calibrating other Iron Age finds. Field doesn't cancel radiocarbon — it supplements it, and both methods lean on each other.
And the starting point of the entire discipline — Émile Thellier's 1938 dissertation, where he first formulated the idea of dating by magnetization; together with his wife and colleague Odette, he built the experimental base on which archaeomagnetism still stands. The Thellier-Thellier protocol for measuring ancient field intensity remains the gold standard ninety years later.
Back to the detective story embedded in the article's motivation. In 2002, a sensation hit the world: a limestone ossuary with Aramaic inscription "James, son of Joseph, brother of Jesus" — possible earliest material trace of Jesus's family. Collector Oded Golan, a quantity of experts on both sides, seven-year trial, over a hundred witnesses, 475-page verdict — and in 2012 the Jerusalem District Court acquitted the main defendants of the main forgery charges: too many specialists, too contradictory findings, authenticity not established either way. The court's wording — "forgery not proven beyond reasonable doubt" — became for years an emblem of expert methods' impotence where stakes are high.
The article's authors cite this failure as direct motivation. What does a court need? Not an expert's opinion, but a physical quantity with reproducible protocol: heat this way, measure this way, get the threshold — and any accredited lab process will give the same answer. The 112 °C threshold is an attempt to turn the dispute about antiquity from philological to thermodynamic. And method application is already outlined: first "clients" were confiscates from the Theft Prevention Unit — artifacts seized from dealers without provenance, whose authenticity would previously have been subject to eternal disputes.
There's quiet poetry in this story too. Jerusalem's tourist ceramics exist to pretend to be ancient — cheap, honest fakery, sold as souvenirs. In the experiment it played itself in reverse: the standard of newness, the bottom rung of the forgery ladder, positioned to catch the upper one. And the props of scientific history — kilns, magnetometers, ten-degree steps — all this is just a way to ask clay one question: how long did you lie motionless? And the clay answers — because it answered all these thousands of years, regardless of whether anyone asked or not.
My subjective opinion: this is one of the most elegant scientific papers of the year, and its elegance lies in economy. Not a new instrument, not a new field of physics — new attention to old noise. VRM has been known for one hundred twenty years, Louis Néel gave it theory in 1949, a generation of paleomagnetologists methodically erased it from samples. But change the optics — and the "interference" turned into the longest-lived data storage system on the planet: no servers, no updates, no carriers that die before the data. Earth maintains a distributed ledger of time, and its records are read by an ordinary kiln.
Second thing that caught me: the method solves not a scientific but a legal and institutional problem. Seven years of trial and 475 pages of verdict couldn't answer the question that half an hour in a thermal demagnetizer now answers. Engineering lesson: when a dispute lasts decades, look for a quantity that doesn't depend on the parties' opinions. Temperature threshold is what can be entered in protocol and reproduced in any lab worldwide.
And the eternal: "noise" is just a signal whose consumer we haven't built yet. For a century VRM interfered with measuring TRM — and turned out to be a date, legal evidence, and the biography of a jug from the Babylonian fire. Every time in my data I see a component I want to throw out — I'll now remember the 112 degrees. Science's trash bin is the most unexplored repository of discoveries.