Paper check · The Sun
A 2026 review asks whether superflares on Sun-like stars and the extreme particle storms recorded in tree rings and ice are “two sides of the same coin” [1 p. 2]. It rules out a one-to-one link but can’t yet choose between the two other answers. We checked the infographic that sums it up and corrected its flare energies and charts.
Our analysis marks our own reasoning and calculations, and the verdicts are our assessment. Everything else comes from the numbered sources.
7 + 2
extreme solar particle events confirmed in tree rings and ice cores over 15,000 years, plus two candidates [1 p. 3]
The review’s claims
“hypothesis (i) is rejected, with high confidence, by the observed data”
The review, on a one-to-one link between superflares and extreme particle events [1 p. 14]
It leans on the authors’ own estimate that superflares come about once a century on Sun-like stars [2], about 12 to 15 times as often as these events, or 15 to 20 times counting only the confirmed ones. At the other published rates, superflares would be between three times as frequent and four times rarer [1 p. 7]. See section 4.
“we cannot distinguish between hypotheses (ii) and (iii)”
The review, on the two remaining answers [1 p. 14]
A scenario the review calls “speculative but physically plausible” fits (ii) [1 p. 15], and a probabilistic study by two of its authors favours (iii) [3]; the data exclude neither. See section 4.
“hypothesis (ii) the current paradigm”
The review, on superflares that only rarely cause extreme particle events [1 p. 14]
This describes where the field stands, with two citations co-written by one or more of the authors; it isn’t a finding the review tests. The infographic reports it accurately. See section 4.
“roughly once per 1500 years”
The review, on how often extreme particle events happen [1 p. 3]
The review gets this by counting its two unconfirmed candidates with the seven confirmed events, as it says. The seven alone give about one per 2,000 years, in line with its own “one–two millennia” [1 p. 5]. See section 5.
“roughly one event per century for E ≳ 1034 erg”
The review, on superflares on Sun-like stars [1 p. 8]
It comes from one team’s analysis of Kepler data, by the authors and colleagues, which they argue is the best matched to the Sun [2]. None of the other studies the review lists finds a rate that high, and it calls the results “rather inconsistent” [1 p. 7]. See section 5.
1 · Two lines of evidence
Extreme solar particle events are known only for our Sun, from isotopes in tree rings and ice; superflares are seen on other Sun-like stars. The two have been “discovered and are studied by different methods on different objects” [1 p. 14].
Sudden spikes of carbon-14 in tree rings, and of beryllium-10 and chlorine-36 in ice cores, archives that can be dated independently [1 p. 3]. Their isotope pattern “unambiguously confirmed” a solar origin: energy spectra like those of modern solar particle events, but about a hundred times stronger [1 p. 4]. Seven are confirmed and two are candidates, over 15 millennia [1 p. 3].
| Event | Above 100 MeV | Above 200 MeV | Above 500 MeV |
|---|---|---|---|
| AD 775 | 114 × | 72 × | 34 × |
| 660 BC | 137 × | 83 × | 38 × |
| 7176 BC | 125 × | 92 × | 53 × |
| AD 994 | 54 × | 34 × | 16 × |
Our analysis Fluence of each reconstructed event above the energy shown, divided by that of the strongest particle event recorded on the ground, on 23 February 1956, read from the review’s Fig. 2 to about ±10% [1 p. 2][1 p. 5]. About a hundred times at 100 MeV for the three strong events, falling to tens of times at 500 MeV. Against the other strong events in Fig. 2, from 1972 and 1989, the ratio doesn’t fall with energy.
2 · When the Sun went extreme
The review names five of the seven confirmed events. The strongest, in 12,350 BC, was 18% stronger than the first one found, in AD 774–775 [1 p. 3].
Scroll sideways to see the whole timeline →
Searches for a particle signal after the strongest flare on record, in 1859, “failed – there was no detectable ESPE” [1 p. 14]; one found no sign in Alaskan tree rings [4]. Another, in polar tree rings, did find a statistically significant, short-lived offset in carbon-14 that its model couldn’t reproduce [5]. Our page on the Carrington storm in the ice looks at the ice-core side.
3 · Where a flare’s energy goes
Free magnetic energy builds up as an active region’s field is “sheared and twisted”, and reconnection releases it [1 p. 10]. How it is shared between radiation, ejected mass and particles likely depends on the region’s magnetic flux and topology [1 p. 1].
Soft X-rays, the band behind a flare’s X-class, add “perhaps another roughly 1%” to the less than 1% in harder X-rays; most of the energy comes out in the ultraviolet, visible and near-infrared, where it remains “poorly constrained” [1 p. 11]. In a confined flare, most of the energy “remains trapped”, so it can look bright but usually releases fewer particles [1 p. 12].
In eruptive events, “often more than half” of the released energy goes into the ejection’s motion and potential energy, and above about X4–X5 “nearly all flares are eruptive” [1 p. 12]. On active stars, the overlying field may stop ejections from escaping in flares above 1034–1035 erg [1 p. 12]; on the Sun, simulations suggest this may already happen above a few × 1032 erg [1 p. 15].
Reconnection in the flare accelerates particles fast, with hard spectra, for tens of minutes to hours; shocks driven by the ejection work more slowly but over large volumes, for several hours to a day [1 p. 13]. The most extreme events “likely require both”, and an open magnetic field so the particles can escape [1 p. 13].
“The strongest SEP events were not caused by the strongest flares and vice versa” [1 p. 13]: acceleration, magnetic topology and the particles’ path through space all matter [1 p. 14]. Flare size and particle output are related statistically for weak and moderate events, but for the strongest the relation “remains elusive” [1 p. 13].
4 · Three hypotheses
The review sets out three ways superflares and extreme particle events could be related [1 p. 14].
Hypothesis (i) Rejected
Superflares and these events would be nearly one-to-one, so their rates could be compared directly. The review rejects this because recent analyses, led by the authors, “demonstrated that superflares are significantly more frequent” on Sun-like stars than these events on the Sun, and because none followed the Carrington flare [1 p. 14].
Hypothesis (ii) Current paradigm
Extreme particle events come from superflares, but only when conditions are favourable [1 p. 14]. In a scenario the review calls “speculative but physically plausible”, strong overlying fields hold back ejections and particles in the biggest flares, so that escape grows less likely as magnetic complexity and flare energy grow [1 p. 15].
Hypothesis (iii) Not excluded
Most of the events may come from “strong but not yet extreme flares in the X1–X10 range” when conditions happen to be favourable, with superflares causing one only rarely [1 p. 14]. That is what a probabilistic study by two of the review’s authors concluded [3].
It rests mostly on rates. At the authors’ estimate of one superflare per century, superflares outnumber extreme particle events about 12 to 15 times, or 15 to 20 times counting only the confirmed events. At the other rates in the review’s Table 2, once per 500 to 6,000 years, they would be between three times as frequent and four times rarer [1 p. 7], a range that earlier work treated as close enough to allow a one-to-one link [1 p. 14]. The Carrington flare, at a few × 1033 erg, is below the 1034 erg used for the rates, though above where the review elsewhere says superflares begin [1 p. 11][1 p. 6]. Both arguments point the same way, but neither is decisive on its own.
5 · How often
Superflare rates come from other stars, extreme particle events from isotopes, and a third estimate from the Sun’s own X-ray record. They span a factor of ten thousand.
about a century Superflares above 1034 erg on Sun-like stars
about 1,500 years Extreme solar particle events
1 million years or more The Sun’s X-ray flare record, extrapolated to 1034 erg
Average years between events (logarithmic scale: each step is ten times the last)
The review stresses that the X-ray and stellar estimates rely on different observables, baselines and degrees of extrapolation, so “the two estimates are not directly comparable” [1 p. 9]. A different extrapolation of the Sun’s flare record, by energy rather than X-ray brightness, does “align with these stellar results” [1 p. 8].
6 · How big
Typical large solar flares release 1030–1032 erg. Estimates for the most extreme historical cases reach several times 1033 erg, and the largest sunspot group on record could have produced more [1 p. 11].
1030–1032 erg Typical large solar flares
(2–6) × 1032 erg The Halloween flares of 2003, measured
a few × 1033, up to 7 × 1033 erg The Carrington flare of 1859, estimated
6 × 1033 to several × 1034 erg What the great sunspot of April 1947 could have produced
about 1033 to 1036 erg Superflares on stars of the Sun’s type, up to 1034–1035 on the closest matches
Total radiated flare energy, erg (logarithmic scale)
In the review’s words, solar flares “could reach energies of several 1034 erg, approaching the lower end of the stellar superflare regime” [1 p. 11].
7 · What it means for Earth
An extreme particle event today would be felt above the atmosphere, not on the ground.
Past events “remained unnoticed by ancient societies”, because Earth’s magnetic field and thick atmosphere protect the surface [1 p. 6].
The authors write that today “the spacecraft fleet could be instantly destroyed by corpuscular radiation”, probably sparing satellites in low-altitude, low-inclination orbits [1 p. 6].
During geomagnetic excursions and reversals, several per million years and each lasting a few millennia, shielding is greatly reduced for long enough “to accommodate several ESPEs” [1 p. 6]. Our analysis At one event every 1,500–2,000 years, an excursion of a few thousand years would see one to three on average.
The PLATO mission will observe many more bright Sun-like stars, and spectroscopy can constrain stars’ ages and help reject unresolved binaries [1 p. 9]. More precise isotope measurements should find weaker events, “filling the observational gap” to the events measured today [1 p. 5].
We read the accepted manuscript on arXiv, because the publisher’s site blocked automated access; the published abstract shows small copy edits, so the wording elsewhere may differ slightly [1].
The highest estimates, about 7 × 1033 erg for the Carrington flare and several × 1034 erg for the 1947 sunspot, come from a study led by one of the authors and published in the same issue; we read only its abstract, which gives “a few ×1034 erg” for 1947 [6]. The lower estimates come from other teams the review cites [1 p. 11]. Of the other cited studies, too, we read only abstracts.
Bottom line: a one-to-one link looks unlikely. Whether every extreme particle event comes from a rare superflare, or most come from smaller flares, is still open.
The review is careful: it rejects only the simplest link and says the data can’t separate the other two. Its rejection leans on the authors’ own superflare rate, which other published samples put lower, and the scenario it offers for (ii) is, in its own words, “speculative but physically plausible”. More Sun-like stars from PLATO and isotope records sensitive enough to catch weaker events are what would decide it.
Bracketed numbers refer to this list; p. gives a page of the review’s accepted version. Values without a reference are our calculations from the cited inputs. Links checked October 19, 2026.
Vasilyev, V., Usoskin, I., and Krivova, N. (2026). Linking solar magnetism, extreme solar particle events and stellar superflares. Philosophical Transactions of the Royal Society A 384(2329), 20250261. CC BY 4.0. We read the accepted version, arXiv:2602.10243; p. gives its page.
Vasilyev, V., Reinhold, T., Shapiro, A. I., Usoskin, I., Krivova, N. A., et al. (2024). Sun-like stars produce superflares roughly once per century. Science 386(6727), 1301–1305. We read the abstract.
Koldobskiy, S., Usoskin, I., Vasilyev, V., and Shapiro, A. I. (2025). Solar flares versus solar energetic particle events: Probabilistic projections to extreme events. Astronomy & Astrophysics 702, A195. CC BY 4.0. We read the abstract.
Miyake, F., Hakozaki, M., Hayakawa, H., Nakano, N., and Wacker, L. (2023). No signature of extreme solar energetic particle events in high-precision ¹⁴C data from the Alaskan tree for 1844–1876 CE. Journal of Space Weather and Space Climate 13, 31. CC BY 4.0. We read the abstract.
Uusitalo, J., Golubenko, K., Arppe, L., et al. (2024). Transient offset in ¹⁴C after the Carrington event recorded by polar tree rings. Geophysical Research Letters 51(5), e2023GL106632. CC BY 4.0. We read the abstract.
Krivova, N., Chatzistergos, T., Kazachenko, M., and Işık, E. (2026). Empirical flare energy limits for the largest historical sunspots. Philosophical Transactions of the Royal Society A 384(2329), 20250290. CC BY 4.0. We read the abstract.
Method
On October 19, 2026, we checked every statement, number and chart on the original infographic, 69 items, against the review’s accepted manuscript, its Crossref record and the abstracts of the studies it relies on. We read the review’s Fig. 2 to compare its extreme events with the strongest event recorded on the ground, measured where the original placed each mark on its timeline and charts, and redrew all three to scale. The check corrected several statements; the change log lists them.
The verdicts and the items tagged “Our analysis” are our own assessment, built on the cited sources: the readings of Fig. 2, the rate arithmetic and our reading of the review’s claims. The review is open access under CC BY 4.0 [1]; the timeline and both charts are drawn by factchecked.science from its values.
After the page was built, a second reviewer who hadn’t seen the drafting re-read every quotation against the review, re-read Fig. 2 from the PDF’s vector paths, which agreed with our table within about 1%, converted every chart mark back to a value and checked every reference. The changes that followed: hypothesis (iii) is no longer described as no link at all, since in it superflares still cause an event now and then; the title now says the two don’t appear to be one-to-one, as the review’s abstract does; the review’s label for its scenario, “speculative but physically plausible”, is quoted in full; superflares outnumber the events about 12 to 15 times, or 15 to 20 times counting only confirmed events; the review’s two lower ends for superflares are both stated; the study behind the highest flare estimates, now published, is cited; and several locators and wordings were corrected.
This page was researched and written with Claude, an AI assistant made by Anthropic, and then re-checked against its sources in a separate verification pass.
Corrections
If we get something wrong, we fix it and record the change below with the date. If a correction changes a verdict, the log says what the old verdict was.