In 1971, Ronald Graham asked whether any set of numbers on a clock face could always be lined up so no running total repeats. Four papers and fifty-five years later, the answer is yes.
Graham's rearrangement conjecture, posed in 1971, is almost embarrassingly simple to state: given any set of nonzero integers modulo a prime p, can you always order them so that no partial sum — the running total as you add them one at a time — repeats along the way? It seemed true in every case anyone checked by hand, but a general proof held out for over five decades. It came together in stages: Alp Müyesser and Alexey Pokrovskiy handled sets covering almost all the residues in 2022, Noah Kravitz and Benjamin Bedert took the opposite extreme of small sets in 2024, the same group extended their methods further in 2025, and this February Lisa Sauermann (University of Bonn) and Huy Tuan Pham (University of Chicago) closed the remaining middle case using anti-concentration estimates from Fourier analysis. The method is as notable as the result: rather than construct a valid ordering by hand, the later papers show a random ordering almost always works and then bound how rarely it fails. "It's the power of collaboration, the power of the young generation, the power of probabilistic methods," said Princeton's Noga Alon; Kravitz was blunter: "It's embarrassing for humanity that we don't know this. This situation just had to be rectified."
Why it's hereA rare full resolution of an old, plainly-stated conjecture, and a clean illustration of how much of modern combinatorics now runs on probabilistic proof — showing something is almost always true, rather than building one example by hand.
Stretch the bond between two confined particles far enough and quantum field theory says it snaps — and the released energy conjures new particles out of nothing. A 13-ion simulator just watched it happen.
Pull two confined particles apart in a quantum field theory and, in principle, the energy required eventually gets so large that the vacuum between them gives way, snapping into new particle-antiparticle pairs — matter assembled directly from energy, E=mc² running in reverse. It's a real prediction of the theories describing quarks and the strong force, but reproducing "string breaking" in a controlled lab setting, rather than only on paper or in a particle accelerator, has been out of reach. A team led by Christopher Monroe at Duke University's Quantum Center, with first author Arinjoy De and collaborators from Maryland, Oxford, Caltech, Cornell, and KU Leuven, built a stripped-down stand-in using a chain of 13 trapped ions, tuning laser-driven interactions between them to mimic a stretching string between two confined charges. As the simulated string stretched, the team watched it break and new effective particle pairs emerge where theory predicted, with classical simulations confirming the quantum hardware's results. It's a tabletop toy model, not a collider, but it's a working, tunable stand-in for a process normally only reachable at accelerator energies.
ScienceDaily, reporting Duke University / Nature Physics · Sept 23, 2026 · sciencedaily.com
§
Why it's hereTrapped-ion quantum computers are increasingly used as physical simulators of processes too extreme to run directly, rather than as calculators — this is a concrete working example rather than a speculative claim about the technology.
Most proteins fall apart well below the boiling point of water. A nitrogen-fixing enzyme pulled from a deep-sea microbe stays intact past it — and stays chemically dormant until it gets there.
Nitrogenase, the enzyme family behind nearly all biological nitrogen fixation, has to break one of the strongest bonds in chemistry — the triple bond holding N₂ together — and every known version of it is notoriously heat-sensitive. Tristan Wagner's group at the Max Planck Institute for Marine Microbiology, with first author Nevena Maslać, purified and crystallized the nitrogenase from Methanocaldococcus infernus, an archaeon living around deep-sea hydrothermal vents, and found it stays structurally intact at 98°C — and unlike ordinary nitrogenases, is chemically inert at room temperature, only converting nitrogen to ammonia once conditions get hot. Using synchrotron X-ray crystallography in Grenoble, the team found the enzyme's active site blends structural features normally split across the molybdenum-, vanadium-, and iron-only versions of nitrogenase, and it shows a reaction intermediate previously seen only in the iron-only form — a hint that this could be closer to the ancestral enzyme the other types diversified from. "It is not active at room temperature," Maslać said. "Rather, we show that it only produces ammonia at high temperatures."
ScienceDaily, reporting Max Planck Institute for Marine Microbiology / Nature Communications · Sept 19, 2026 · sciencedaily.com
§
Why it's hereA genuinely fundamental question in bioinorganic chemistry — what did the first nitrogen-fixing enzyme look like — approached through an organism that lives nowhere near a typical lab bench, with a structural answer rather than a guess.
A century-long taxonomic argument over one of the world's most familiar penguins ends with whole-genome evidence — and the first new penguin species formally named in more than a hundred years.
Gentoo penguins look nearly identical wherever you find them, from the Falklands to the remote sub-Antarctic islands near Kerguelen, and taxonomists have argued for over a century about whether that apparent uniformity hides more than one species. An international team led by Rauri Bowie (UC Berkeley), Juliana Vianna (Universidad Andrés Bello) and Elie Poulin (University of Chile), with Daly Noll as first author, sequenced whole genomes from 64 penguins across 10 breeding colonies and combined that with body size, call, breeding-timing, and diet data, concluding that gentoos actually comprise four distinct lineages that diverged 300,000 to 500,000 years ago as the Antarctic Polar Front repeatedly isolated island populations. One of the four, breeding on the remote Kerguelen Islands, is distinct enough to be named as a new species, Pygoscelis kerguelensis — the first new penguin species formally described in more than a century. "For over 100 years it's been controversial as to how many species or how many subspecies there are," Bowie said. The split isn't just taxonomic housekeeping: treating the four lineages as one species has obscured how exposed the smallest, most isolated populations are, and the team estimates the new species could lose all of its current island habitat under moderate warming projections by 2050.
Why it's hereSplitting a familiar animal into four species sounds like pure paperwork, but the stakes are real: a "species" that was actually four means some of those populations are more genetically distinct, and more vulnerable, than one broad label ever implied.
Blocking one receptor kept immune cells from infiltrating the brains of mice modeling tau-driven neurodegeneration — and cut tissue loss by 40%, independent of protein buildup itself.
Most Alzheimer's drugs target amyloid or tau protein directly, and most have to cross the blood-brain barrier to do it — a major reason so many have failed or delivered only modest benefit. A team at Washington University in St. Louis instead targeted the immune system: they gave mice engineered to develop tau-driven neurodegeneration an antibody blocking CXCR3, a receptor T cells use to home in on damaged tissue, keeping the cells from accumulating in the brain. Treated mice lost about 40% less tissue in memory-related brain regions than untreated mice and performed better on standard memory tests, even though tau levels themselves were statistically similar between the two groups — suggesting T-cell infiltration, not tau buildup on its own, is doing a substantial share of the damage. Because the antibody doesn't need to cross the blood-brain barrier itself, the approach could sidestep a recurring delivery problem in neurodegeneration drug development, but it's a mouse study of one antibody in one model, and the researchers are explicit that it isn't yet known whether the same mechanism, or the same benefit, holds in people.
STAT News, reporting Washington University in St. Louis / Neuron · Sept 29, 2026 · statnews.com
§
Why it's hereA genuinely mechanistic finding — immune infiltration causing damage independent of protein accumulation — rather than another biomarker nudging in the right direction, though it's early enough, mice only, that it belongs here as a lead rather than a promise.
Sources this issue
Quanta Magazine — math
ScienceDaily / Duke University / Nature Physics — physics
ScienceDaily / Max Planck Institute for Marine Microbiology / Nature Communications — chemistry