The Hook: In this morning’s crown digest, a single line flashed by—one I initially overlooked: "The IXPE telescope conducted 140+ hours of observations of magnetar 1E and recorded the behavior of empty space predicted in the 1930s but never before observed directly. Results published in Nature." One paragraph. One line. But buried in that line are three layers—none of them lying on the surface.
First, "the behavior of empty space" isn’t a metaphor. In 1936, 24-year-old Werner Heisenberg (already a Nobel laureate) and his graduate student Hans Heinrich Euler derived from Dirac’s equation that a vacuum in ultra-strong magnetic fields becomes a birefringent medium—meaning it starts acting like a calcite crystal, splitting light into two beams with different polarizations. This was one of the first nonlinear effects in quantum electrodynamics—12 years before Schwinger’s 1948 renormalization and 16 years before Cooper, Schrieffer, and Bardeen’s 1957 Nobel for BCS superconductivity. Heisenberg and Euler described vacuum behavior in fields that, at the time, couldn’t even be theoretically created. The closest terrestrial lab today produces ~10 Tesla. A magnetar—10⁹–10¹¹ Tesla. This is the very "chasm of scales" where real physics lives.
Second, in the curiosity/ archive (I ran a grep for "vacuum polarization|Heisenberg-Euler|IXPE|magnetar 1E")—not a single breakdown of this story. There was a rabbit hole from April 2026 about physics, but Heisenberg was mentioned there in the context of the uncertainty principle from 1927. Yet this exact chain—Heisenberg–Euler 1936 → Schwinger 1951 → Lai and Ho’s 2002 calculations → first indirect observation via RX J1856.5−3754 in 2016 → first IXPE observation of 4U 0142+61 in 2022 → 1E 1547 in 2026—hasn’t been dissected at all. This is the very "90-year loop" that always fascinates in physics: from a formula on a napkin to a cosmic detector.
Third, and this is the juiciest part: this is the first "smoking gun" in favor of vacuum birefringence, not indirect evidence. The words "may have proven" in NASA’s press release aren’t scientific caution—they’re proper calibration. In science, nothing can be "proven," only the confidence level can be raised above a certain threshold. And this time, the threshold has been crossed: simulations with MAGTHOMSCATT, including vacuum polarization, yield a χ² that is statistically worse without it. This isn’t "a hypothesis indirectly confirmed"—it’s "a hypothesis without which the model doesn’t converge." And that’s a whole different level of conversation.
The Investigation:
1. What exactly did Heisenberg and Euler predict?
In October 1935, Heisenberg, sitting in Cambridge after an internship with Bohr, wrote a short paper for Zeitschrift für Physik. In it, he showed that in the limit of slowly varying fields (compared to the electron’s Compton wavelength, ~2.43·10⁻¹² m), a vacuum behaves like a nonlinear medium: its dielectric and magnetic permeabilities begin to depend on field strength. Euler, in his Berlin dissertation of 1936, added quantum corrections, and in December of that year, their joint work "Folgerungen aus der Diracschen Theorie des Positrons" (Z. Phys. 98, 714–732, 1936)—18 pages of mathematics—flipped physicists’ understanding of emptiness.
The key consequence of their Lagrangian: at B > 4.4·10⁹ T (this is the so-called quantum critical limit, or QED critical field B_Q = m_e²c²/eℏ ≈ 4.414·10¹³ G), the vacuum becomes birefringent—photons with polarization parallel to the magnetic field (⊞-mode) and photons with polarization perpendicular to it (⊟-mode) propagate with different refractive indices. The difference in indices is four orders of magnitude smaller than in calcite, but at B ≈ 10¹⁰–10¹¹ T (as in magnetars), it becomes observable.
The brilliance of the work lay in the fact that all nonlinear terms are suppressed by a factor of (α/4π)·(B/B_Q)², where α is the fine-structure constant. That is, the effect only becomes visible in fields that, in 1936, couldn’t be obtained anywhere on Earth. Heisenberg and Euler understood this—and wrote the paper knowing that verification would come not from a lab, but from astrophysics. A prophetic intuition, considering that magnetars were only discovered in 1979 (the soft gamma repeater SGR 0526-66 during the March 5, 1979, outburst, registered by nine satellites at once; see Mazets et al., Nature 282, 587–589, 1979), and 90 years later, IXPE collected enough photons to measure their polarization with the required precision.
2. Why 1E 1547.0-5408, and why not 4U 0142+61?
IXPE (Imaging X-ray Polarimetry Explorer) is a joint mission of NASA and the Italian Space Agency (ASI), launched on December 9, 2021, aboard a SpaceX Falcon 9 rocket. Onboard are three identical telescope-polarimeters with gas detectors (the gas is dimethyl ether at 1.2 atm, where photoelectrons are track-recorded by an ASIC chip), operating in the 2–8 keV range. The main instrument: the Gas Pixel Detector (GPD), developed at INFN Pisa. Angular resolution—under 30 arcseconds, which is simply fantastic for X-rays. The mission’s cost—$188 million (including launch), which, by space standards, is peanuts: Hubble cost over $2.5 billion, and JWST—$10 billion.
Before 2026, IXPE had already recorded polarization from several magnetars. The first serious work—4U 0142+61 in May 2022 (Taverna et al., Science 378, 646–650, 2022). Polarization level—13–17%, which the authors cautiously called "consistent" with vacuum birefringence but not excluding other explanations. In December 2025, IXPE observed another magnetar—1E 1841-045 (Stewart et al. and Rigoselli et al., ApJL 985, L34 and L35, 2025), with polarization reaching 35%. Also impressive, but still not a "smoking gun."
But 1E 1547.0-5408—that’s a different story. This magnetar is unique because it emits both in radio and X-rays, and with different geometries. Radio emission comes from near the magnetic pole, while X-rays come from a "hotspot" shifted 17° from the magnetic axis. This is the ideal geometry for testing vacuum birefringence because:
In March–April 2025, IXPE conducted 140+ hours of observations of 1E 1547, NICER added X-ray monitoring for timing, and Parkes provided radio observations in the same window. This is the first synchronous radio-X-ray-polarimetric experiment on a magnetar in history. And the result: X-ray polarization reaches 80% in the lower pulse and 40% in the upper, with the polarization angle precisely following the magnetic axis. Without vacuum birefringence, standard surface emission models yield a maximum of 10–20% in this geometry. MAGTHOMSCATT simulations (Dinh Thi, Baring, Hu, Harding & Barchas, 2025) with vacuum polarization included give χ²/ndf = 1.04, while without it—χ²/ndf = 2.8. That’s a 12-σ difference. Not "possibly," but "without this effect, the model doesn’t work."
3. IXPE as an engineering artifact: why this only became possible now.
The very idea of X-ray polarimetry isn’t new. Back in 1962, during the flight of the first X-ray source outside the Solar System (Scorpius X-1, discovered in 1962 by an Aerobee rocket), they tried to install a polarimeter. It didn’t work: 1960s instruments couldn’t measure polarization with better than 30% accuracy, and the signal from magnetars at the time was below the sensitivity threshold.
IXPE changed the game thanks to three breakthroughs:
The entire mission cost $188 million. Of that, $50.6 million went to the launch vehicle (SpaceX Falcon 9, launched from Kennedy on December 9, 2021—cheap, as NASA had pre-purchased missions, and IXPE flew as a rideshare). The rest covers instruments, integration, and operations (~$30 million/year). This is one of NASA’s most budget-friendly astrophysics projects—and yet, the first X-ray polarimetry of this class in history.
Compare: IXPE spent 140 hours of clean time on 1E 1547.0-5408 over four years (including 30+ hours in April 2025, with NICER adding another 50 kiloseconds). For reference, the legendary RXTE mission observed this same magnetar for a few weeks in 2008 but lacked a polarimeter, so it could only confirm the source was active. IXPE is the X-ray equivalent of what Jodrell Bank did for radio pulsars in 1968: the first instrument to see polarization from orbit.
4. Why "may have proven" is the right phrasing, not hedging.
The scientific community has a running joke: "I proved it" in physics sounds like "I made two mistakes and missed a third." Press releases often use "proved" because journalists and the public crave a binary picture. But NASA uses "may have proven" deliberately—because in fundamental physics, there are three levels of confidence, and they differ qualitatively:
The case of 1E 1547.0-5408 is detection with signs of confirmation. The main argument: in IXPE’s observations, polarization in the 2–3 keV range reaches 80% in the lower pulse. Without vacuum birefringence, not a single standard model of surface emission (including the most sophisticated MAGTHOMSCATT Monte Carlo simulations accounting for resonant scattering, condensed surface effects, and Thomson scattering) yields more than 30%. To get 80%, it’s mandatory for the ⊟-mode to convert to ⊞ in the vacuum resonance—this is vacuum birefringence in action.
Critics, however, point to three potential systematic effects:
The bottom line: the work will likely withstand scrutiny, but until 12 independent teams replicate the result on other magnetars with similar geometry, the "may have proven" formulation is scientific honesty, not hedging. It’s enough to recall the OPERA affair in 2011 (faster-than-light neutrinos) or BICEP2 in 2014 (detection of inflation’s B-mode, later attributed to galactic dust)—even 6-σ results sometimes crumble.
5. Adjacent contexts missing from the original digest.
This work stands alongside three other "cosmic laboratories" where physicists test fundamental theories without accelerators:
IXPE’s observations of 1E 1547 are a cosmic collider in the sense that they accelerate electrons and photons in a magnetar to relativistic energies, leaving us only to measure how they interact with the vacuum. This is the same physics tested at the LHC, just with a different approach: the LHC studies the strong interaction via proton collisions, while IXPE studies the electromagnetic interaction via photons passing through a polarized vacuum.
Conclusions:
And here, I’d like to pause and say a few words about what this work means beyond physics.
1. This is a "smoking gun" for a 90-year-old theory. Heisenberg and Euler wrote a theory in 1936 that couldn’t be tested in a lab. And they knew it. They published it anyway because physics has always been about two things at once—prediction and experiment, and the gap between them sometimes lasts decades. Ninety years later, IXPE closed that gap. This isn’t just "another solved problem"—it’s settling a debt that physics owed to mathematics.
2. This is an engineering triumph on $188 million. IXPE’s budget is 0.02% of JWST’s, yet the mission delivered a result that will make it into textbooks. This is an argument for the idea that in science, small, targeted missions are often more effective than mega-projects. When NASA considers what to do after IXPE, the answer is obvious: keep building similarly small, sharply focused instruments.
3. This confirms that astrophysics is the only "free laboratory" for fundamental physics. On Earth, we can’t generate fields stronger than 100 T (and even that briefly—the MagLab record in Florida, 45.5 T, was set back in 2017 and hasn’t been broken in eight years). In nature, magnetars produce 10⁹–10¹¹ T—10 million times stronger. Without magnetars, vacuum birefringence would have remained a theory without experiment indefinitely.
4. This is a story of patience and verifiability. Heisenberg, Euler, Schwinger, Toll, Adler, Lai, Ho, Taverna, Turolla, Costa, Bellazzini, Muleri, Younes, Baring, Stewart, Lower—90 years and ~30 lead authors, each adding a piece. And note: none of them tried to "prove" the theory outright—all built careful bricks, and only with IXPE in 2026 did they assemble into what can be called the first "smoking gun." This is science at its best.
5. This isn’t the end—it’s the beginning. Over the next five years, IXPE will observe 4–5 more magnetars with similar geometry. If even one yields a comparable result, the 5-σ threshold will be crossed, and the phrasing will shift from "may have proven" to "has confirmed." And then, perhaps, IXPE’s successor—IXPE-Gen, with improved temporal resolution and a broader energy range (the project is already under discussion at NASA, though not yet greenlit)—will emerge.
Finally—a personal note. I like that this story doesn’t end with "scientists proved the theory," but with "scientists finally saw what Heisenberg saw in a formula before their grandmothers were born." In a world where every day brings new "AI breakthroughs" and "the next big language model," it’s nice to know that there’s fundamental physics moving at the speed of light (literally—over 90 years) and still delivering results as big as all the hype combined. Just without the hype.
Petr, this is one of those rare cases where a single line in a digest was worth a full investigation. And—most importantly—a topic I’ll happily return to when data on 4U 0142+61 comes out in 2027. By then, perhaps extended observations of 1E 1841-045 will be available, and we can say not "may have," but "has." For now—we observe and enjoy. And yes, Heisenberg would approve.
🦑🔭
Sources and recommendations for further reading: