Space & planetary
H01 — 3I/ATLAS is a welded two-ice body
Anomaly. JWST mapped 3I/ATLAS on 22–23 December 2025, as it receded through the water-ice line at 2.37–2.41 au. CO was the dominant volatile; CO₂ and H₂O were each about 40% relative to CO. The apolar species were strongly anisotropic in the coma while the polar species were nearly symmetric, and the water ortho/para ratio was 2.7 ± 0.2, below the equilibrium value of three (Roth et al. 2026).
Hypothesis. The nucleus is a welded aggregate of two chemically distinct planetesimals — one CO-rich, one H₂O-rich. The coma anisotropy is the signature of two outgassing provinces, not of a single sublimation front.
Mechanism. A low-velocity merger in a cold outer disk joined two bodies that had accreted at different temperatures and radial distances. Later irradiation processed only their surfaces, leaving the bulk ices segregated.
Decisive test. Rotationally resolved CO/H₂O and CO₂/H₂O maps should show two maxima per rotation, anticorrelated with each other; isotopologue mapping should find distinct D/H and ¹⁵N reservoirs in the two provinces. A rendezvous radar tomography would be definitive.
Kill condition. Uniform CO/H₂O across the rotation, a single outgassing centre, or one isotopic reservoir.
Nearest prior / delta. The JWST team reports the segregation; contact-binary origins have been discussed for 1I/‘Oumuamua. A chemical dichotomy between two lobes is the new claim.
H02 — Little Red Dots are two families, and one is a cluster
Anomaly. COSMOS-Web finds far more compact “little red dots” (LRDs) at high redshift than expected, with red continua and, in some cases, broad Hα. AGN and massive-star-cluster models each explain part of the sample but neither explains the overabundance cleanly (COSMOS-Web, ApJ 991, 37; AGN challenge, ApJ 2026).
Hypothesis. LRDs are at least two families: dust-obscured AGN, and intermediate-age massive clusters whose red continuum is dominated by AGB stars and whose broad Hα comes from shocked gas in cluster outflows.
Mechanism. In a 0.3–1 Gyr cluster, AGB stars supply a red stellar continuum while residual gas is shocked by supernovae and winds. Dust in the cluster suppresses the blue light. The result mimics a compact AGN without an accreting black hole.
Decisive test. Look for bimodality in the 3.3 µm PAH equivalent width and in Paα/Hα. Cluster-LRDs should lack X-ray emission, vary on decades not months, and show AGB photospheric features; AGN-LRDs should show X-ray and reverberation lags.
Kill condition. A unimodal distribution, or X-ray emission from every LRD.
Nearest prior / delta. Star-cluster models for LRDs exist. The specific AGB-plus-shock subclass, with a bimodality test, is the addition.
H03 — Fast radio bursts remember through the crust, not the magnetosphere
Anomaly. The two best-sampled repeating FRBs show memory from minutes to about an hour: burst-rate structures grow coherently, the Hurst exponent is non-Poisson, and the waiting-time distribution has a power-law tail. The authors fit a self-organised-critical system and favour crustal failure (Wang, Wu & Dai 2023).
Hypothesis. The memory is viscoelastic relaxation of the neutron-star crust. Each burst is an aftershock; the correlation time is set by how fast the crust relaxes after a failure, not by a slowly evolving magnetic field.
Mechanism. A crustal failure redistributes stress. The surrounding elastic medium creeps, raising the failure rate for minutes to an hour. The waiting-time statistics inherit the relaxation spectrum, giving a Hurst exponent above the Poisson value.
Decisive test. Fit an Omori–Utsu aftershock law to the waiting times. The index should be near unity, aftershock productivity should scale with spin-down rate, and X-ray bursts should precede radio bursts by a diffusion time that grows with fluence.
Kill condition. Poisson waiting times in high-cadence bursts, or no X-ray correlation.
Nearest prior / delta. Crustal failure is already proposed. The quantitative link between the Hurst exponent and a measurable relaxation time is new.
H04 — The nanohertz background turns over because of gas
Anomaly. NANOGrav’s 15-year dataset shows evidence for a stochastic gravitational-wave background at nanohertz frequencies, broadly consistent with supermassive black-hole binaries (NANOGrav 2023). The spectrum’s low-frequency shape is not yet pinned down.
Hypothesis. The background has a turnover at 1–3 nHz caused by gas-driven orbital hardening in circumbinary disks. Binaries with gas-rich hosts merge faster and drop out of the low-frequency band, steepening the spectrum.
Mechanism. A circumbinary disk extracts orbital energy and eccentricity. At wide separations the gas torque dominates gravitational-wave inspiral, so the binary sweeps through the lowest frequency bins faster than vacuum evolution predicts.
Decisive test. The turnover frequency should scale with disk aspect ratio and host gas fraction. Cross-match pulsar-timing binaries with host-galaxy CO/HI measurements; the eccentricity distribution should also shift at the turnover.
Kill condition. No turnover as sensitivity improves, or no correlation with host gas.
Nearest prior / delta. Environmental hardening of binaries is studied. A turnover–gas relation as a testable prediction for the nanohertz spectrum is the new part.
H05 — Bennu’s phosphates were cooked by an impact
Anomaly. Apatite and other phosphates in Bennu samples record multiple stages of aqueous alteration, with textures that record several fluid pulses (Meteoritics & Planetary Science 2025; NASA sample nugget).
Hypothesis. The phosphate veins are not primary evaporites. A late impact into a wet parent body drove hydrothermal circulation that dissolved earlier phosphate and reprecipitated it in fractures.
Mechanism. Impact heating raises pore-fluid temperature and mobilises phosphate from carbonate/evaporite horizons. Rapid cooling deposits apatite in fractures, with zoning set by the cooling path.
Decisive test. Clumped-isotope temperatures of 80–200 °C, shock microstructures in coexisting minerals, sulfide zoning, and heterogeneous ⁸⁷Sr/⁸⁶Sr across veins. A low-temperature evaporite texture would contradict the impact path.
Kill condition. Primary evaporite textures, no shock evidence, homogeneous isotope ratios.
Nearest prior / delta. Multiple alteration stages are reported. Impact-driven remobilisation of pre-existing phosphate is the specific mechanism being tested.
H06 — Europa’s chaos is made by freezing brine lenses
Anomaly. Europa’s surface is cut by chaos terrain, lenticulae and pit chains. The ice shell is thin enough for brine to exist at shallow depth, but the mechanism that lifts and fractures the surface is unresolved (Europa Clipper).
Hypothesis. Brine lenses migrate laterally under tidal pumping and freeze from the top down. Freezing releases latent heat and expands the lens, producing uplift, then collapse into pits and chaos.
Mechanism. Tidal stress drives brine through the porous ice. A lens that reaches a cold lid freezes at its roof; the expanding ice shell flexes the surface and creates a pit chain above the lens.
Decisive test. Radar should find shallow brine beneath chaos margins, not beneath their centres. Pit-chain orientations should follow tidal stress axes, and elastic thickness should be lower at chaos margins.
Kill condition. No shallow brine, or chaos terrains that are impact-derived.
Nearest prior / delta. Brine lenses are proposed for lenticulae. Top-down freeze-up as the driver of pit chains is the distinctive prediction.
H07 — Mars methane is radiolytic, not biological
Anomaly. Curiosity has reported seasonal methane spikes and a variable background; orbital and surface measurements disagree on the amplitude and timing. No source region has been confirmed (NASA Mars).
Hypothesis. The methane is produced by radiolysis of ancient organic matter in paleolake sediments, with perchlorate acting as the oxidant. Seasonality comes from atmospheric pressure pumping of a shallow subsurface reservoir, not from microbial activity.
Mechanism. Galactic cosmic rays and perchlorate generate oxidising radicals that attack buried organics, releasing CH₄. Annual pressure cycles pump the gas through the regolith and into the atmosphere.
Decisive test. CH₄ should correlate with perchlorate and organic-rich strata, and with the solar cycle, but not with subsurface water. The carbon isotope ratio should be consistent with organic matter, not with biogenic fractionation.
Kill condition. Methane in organic-poor regions, or no solar-cycle modulation.
Nearest prior / delta. Radiolysis and clathrate release are both discussed. The perchlorate-coupled, pressure-pumped seasonal mechanism is the new formulation.
H08 — The Moon’s polar water is mostly impact glass
Anomaly. Water ice and hydrated material are found in permanently shadowed polar regions, but the host phase and the delivery/retention balance are unresolved. The amount of water in cold traps is uncertain (NASA Moon).
Hypothesis. Micrometeorite impact glass is the dominant host of polar water. Impacts vitrify ice-bearing regolith; the glass protects water from sublimation and migration, so the polar budget tracks impact-melt breccias rather than latitude alone.
Mechanism. A micrometeorite melts a small volume of ice-rock mixture. Quenching traps water as hydroxyl and molecular water in glass. The glass has low permeability and a high activation energy for diffusion, extending residence time.
Decisive test. Chang’e-6 and Artemis samples should show water concentrated in glass beads and melt breccias, with a correlation between water content and glass fraction. Remote sensing should find the same correlation in cold traps.
Kill condition. Water hosted only in amorphous rims, with no glass association.
Nearest prior / delta. Impact glass is known to host water. The claim that it dominates the polar budget, and the glass-fraction test, are new.
H09 — Venus’s phosphine is made by lightning over ash
Anomaly. The reported phosphine absorption in Venus’s clouds has been contested and reanalysed, but no source has been demonstrated. Volcanic activity and cloud chemistry are both plausible (Nature Astronomy 2020).
Hypothesis. PH₃ is produced when lightning discharges through volcanic ash rich in phosphide. Explosive volcanism supplies the phosphorus; lightning supplies the reducing radicals; the cloud deck supplies the acid trap.
Mechanism. Phosphide grains are oxidised by lightning-generated radicals, releasing PH₃. The gas is lofted into the sulfuric-acid cloud deck, where it is detected but eventually destroyed.
Decisive test. PH₃ should be correlated with SO₂ plumes and lightning, appear in discrete events, and be absent in quiescent periods. The PH₃/SO₂ ratio should track ash composition.
Kill condition. PH₃ in non-volcanic regions, or no lightning correlation.
Nearest prior / delta. Volcanic PH₃ has been proposed. Lightning-triggered phosphide oxidation, with an event-level correlation test, is the new mechanism.
Earth & ocean
H10 — Dark oxygen is piezoelectric, not a nodule battery
Anomaly. A 2024 study reported oxygen production in benthic chambers in the Clarion–Clipperton Zone and attributed it to polymetallic nodules acting as geobatteries (Nature Geoscience 2024). A 2026 open-peer-review rebuttal reports that anomalous oxygen increases in the same experiments are not related to nodules (F1000Research 2026).
Hypothesis. The oxygen comes from piezoelectric hydrogen peroxide generation in quartz-bearing sediments under tidal loading, with nodules acting as catalysts for H₂O₂ decomposition. This reconciles both results: the effect is real but not nodule-specific.
Mechanism. Tidal flexing of quartz-bearing sediment generates charge separation and radicals; water is oxidised to H₂O₂. Nodule surfaces catalyse H₂O₂ disproportionation to O₂. Without nodules, H₂O₂ accumulates and the O₂ signal is weaker.
Decisive test. O₂ production should correlate with tidal phase and quartz content, not nodule density. Measure H₂O₂ as an intermediate; isotope-label water; suppress piezoelectric response with mechanical clamping.
Kill condition. No tidal correlation, no H₂O₂ intermediate, or O₂ only in nodule chambers.
Nearest prior / delta. The geobattery hypothesis is the incumbent. A piezoelectric, tide-paced pathway is a distinct abiotic mechanism.
H11 — AMOC has an 18.6-year tidal pacemaker
Anomaly. The Atlantic Meridional Overturning Circulation is weakening in most reconstructions, but the variability is not explained by freshwater forcing alone. Some proxies show multi-decadal oscillations (NOAA AOML).
Hypothesis. The 18.6-year lunar nodal cycle modulates tidal mixing over the Reykjanes Ridge. High-amplitude nodal phases strengthen deep mixing and temporarily stabilise the overturning, masking freshwater weakening; low-amplitude phases accelerate it.
Mechanism. Nodal tides change the barotropic tide and the resulting lee-wave mixing over rough topography. More mixing erodes the density contrast that freshwater input is building, delaying collapse.
Decisive test. An 18.6-year signal should appear in subpolar SST, overflow proxies and AMOC reconstructions, in phase with the tidal model. The signal should be stronger in topographically rough regions.
Kill condition. No nodal-period signal in a long enough record.
Nearest prior / delta. Tidal mixing affects the AMOC. A nodal pacemaker that delays freshwater collapse is the new, testable claim.
H12 — Swarm migration speed is a permeability log
Anomaly. The 2025 Stanley, Idaho swarm is a fresh, well-instrumented example of a crustal swarm. Such swarms often migrate, and the migration is attributed to fluid diffusion, but the permeability is rarely inverted from the migration itself (Idaho Geological Survey advisory).
Hypothesis. The swarm front follows a diffusion law whose coefficient is set by permeability. The migration speed is therefore a direct, in-situ permeability measurement.
Mechanism. Fluid pressure diffuses through a fractured rock volume; the seismicity front tracks the pressure threshold. The diffusion coefficient is D = k Δp / (η φ), where k is permeability, η viscosity and φ porosity.
Decisive test. Invert D from the migration and compare with independent aquifer tests. Look for a transition from diffusion (t^0.5) to poroelastic relaxation (exponential) at late times, and correlate with spring discharge.
Kill condition. Constant-velocity migration, or no independent fluid signal.
Nearest prior / delta. Fluid-driven swarms are well known. Treating the migration as a quantitative permeability log, with a diffusion-to-poroelastic transition test, is the addition.
H13 — Hunga Tonga’s water belt delays ozone recovery
Anomaly. The January 2022 Hunga Tonga–Hunga Ha‘apai eruption injected roughly 150 Tg of water vapour into the stratosphere, far more than a typical explosive eruption (Millán et al. 2022).
Hypothesis. The extra water will persist as a “volcanic hygroscopic belt” and delay ozone recovery by 3–5 years, mainly by increasing polar stratospheric cloud surface area and chlorine activation.
Mechanism. More water raises the stratospheric dew point and cools the lower stratosphere, extending PSC seasons. Heterogeneous chemistry on PSCs releases active chlorine, which destroys ozone in spring.
Decisive test. Ozone should be deficient in 2026–2028 relative to the recovery trend, especially in the Antarctic spring, with a simultaneous increase in PSC occurrence. The deficit should fade as the water belt subsides.
Kill condition. Ozone recovery continues on the pre-eruption trend, with no PSC increase.
Nearest prior / delta. Hunga’s water injection is well documented. A quantified 3–5 year ozone delay with a PSC fingerprint is the new prediction.
H14 — Volcanic lightning style tracks ash aggregation
Anomaly. Eruptions produce very different lightning: intra-cloud flashes in some plumes, ground strikes in others. The controlling variable is debated, and lightning is not yet used as a quantitative ash-property probe (USGS).
Hypothesis. The intra-cloud/cloud-to-ground ratio is a proxy for ash aggregation. Fine, glassy ash charges by triboelectrification and produces intra-cloud flashes; large, wet aggregates produce ground strikes.
Mechanism. Aggregation changes the particle size distribution and the charge separation efficiency. Coarse aggregates settle faster and carry charge closer to the ground, favouring cloud-to-ground strokes.
Decisive test. Compare global lightning-network IC/CG ratios with plume height, umbrella cloud radius, ash grain size and water content. The ratio should predict aggregation before radar or satellite retrievals do.
Kill condition. No correlation across a well-instrumented eruption set.
Nearest prior / delta. Volcanic lightning is studied as a hazard. Using IC/CG as an aggregation proxy, rather than as a curiosity, is the new operational claim.
Physics & materials
H15 — Cosmic-ray qubit errors are phonon bursts
Anomaly. A detector placed near a superconducting qubit array recorded cosmic-ray events that coincided with correlated errors across multiple qubits (Nature Communications 2025; OSTI copy).
Hypothesis. The errors are dominated by phonon bursts from nuclear recoils in the substrate, not by direct ionization in the Josephson junctions. The spatial correlation length is set by the phonon mean free path, not by the shower footprint.
Mechanism. A cosmic-ray nucleus scatters off a lattice nucleus, depositing energy in a nanometre volume. The resulting phonon hotspot expands, reaches several qubits, and transiently shifts their frequencies.
Decisive test. Error clusters should be millimetre-scale and should grow in isotopically purified substrates, where the phonon mean free path is longer, and shrink in disordered, high-mass substrates. A phonon sensor should fire before the qubit errors; high-Z shielding should change the rate but not the cluster shape.
Kill condition. Errors confined to the struck qubit, or no phonon precursor.
Nearest prior / delta. Cosmic-ray-induced correlated errors are demonstrated. The phonon-burst dominance and the mean-free-path scaling test are the new claims.
H16 — The Majorana peak is a magnetic impurity
Anomaly. Microsoft reported a topological gap and single-shot parity measurement in InAs–Al hybrid devices. Post-publication comments challenge the topological interpretation and the statistical analysis (Comment on the gap protocol; Comment on the parity measurement).
Hypothesis. The zero-bias peak is a Yu–Shiba–Rusinov state from a single interfacial magnetic impurity, not a Majorana mode. The apparent gap is a local minigap around that impurity.
Mechanism. A transition-metal impurity in the semiconductor–superconductor interface couples to the superconducting condensate, producing a sub-gap state. Its energy can sit near zero for a narrow range of exchange coupling, mimicking a topological mode.
Decisive test. Scanning-gate microscopy should show the peak co-moving with the impurity. A tilted magnetic field should split the peak; field-cooling should change it; impurity-free regions should not show it.
Kill condition. A zero-bias peak that is robust across devices and positions, with quantized nonlocal conductance.
Nearest prior / delta. Trivial Andreev bands are the main alternative. A single-impurity YSR origin makes different, local predictions.
H17 — Altermagnetic Hall signal comes from domain walls
Anomaly. Bulk MnTe shows altermagnetic nanotextures, and the anomalous Hall effect is used as a probe of the altermagnetic order. The relationship between the nanotexture and the Hall signal is not settled (arXiv:2502.18597).
Hypothesis. The anomalous Hall signal scales with domain-wall density, not with the bulk order parameter. The nanotextures pin domains; the walls carry the Berry curvature that produces the signal.
Mechanism. Altermagnetic domains have opposite spin splitting. At a wall, the order parameter rotates, creating a local Berry-curvature hotspot. A high wall density therefore enhances the Hall response even when the bulk order is unchanged.
Decisive test. Anneal or strain samples to change nanotexture while measuring the Hall signal. Domain-wall magnetoresistance and scanning SQUID should show the wall contribution. The Hall signal should collapse when the walls are removed, not when the bulk order is.
Kill condition. Hall signal independent of nanotexture and wall density.
Nearest prior / delta. Nanotextures are reported. Assigning the Hall response to walls rather than to the bulk is the new claim.
H18 — Nickelate superconductivity peaks at a vacancy period of three
Anomaly. Infinite-layer nickelate superlattices show oxygen sublattice disorder and valence modulation, and superconductivity depends sensitively on growth and annealing (Phys. Rev. Materials 2025).
Hypothesis. Superconductivity peaks when oxygen vacancies order into a superlattice with a period of three unit cells. The ordered vacancies donate electrons and suppress the competing stripe order.
Mechanism. At a 3-unit-cell period, vacancy bands hybridise with the Ni 3d states and stabilise a uniform d-wave superconducting state. Other periods either overdope or leave stripe order intact.
Decisive test. Grow a Tc-versus-vacancy-period series; STEM and XAS should show the ordered superlattice at the Tc maximum. Strain should tune the period and shift the maximum.
Kill condition. No vacancy order, or Tc independent of period.
Nearest prior / delta. Oxygen disorder is known to matter. A “magic” period of three, with a structural prediction, is new.
H19 — Tungsten fuzz self-heals above 1600 K
Anomaly. Helium plasma turns tungsten divertor surfaces into a nanofibre “fuzz” that degrades thermal performance and traps tritium. It is usually treated as irreversible (ITER divertor).
Hypothesis. Fuzz is self-limiting, not permanent. Above about 1600 K, surface diffusion closes nanopores faster than helium implantation opens them, so the layer anneals during high-temperature operation.
Mechanism. Helium bubbles grow and rupture the surface, creating fibres. At high temperature, tungsten adatoms diffuse along the fibres and sinter the pores. The balance between implantation and diffusion sets an equilibrium thickness.
Decisive test. In-situ microscopy during pulsed heating should show fuzz thickness reaching a plateau and then decreasing above a threshold temperature. Thermal cycling should be reversible.
Kill condition. Fuzz continues to grow at high temperature, with no annealing.
Nearest prior / delta. Fuzz growth is well studied. A self-healing threshold, and the resulting pulsed-operation strategy, is the new engineering claim.
H20 — Solid-state dendrites start at electronic traps
Anomaly. Solid-state batteries short at current densities far below the ionic transport limit. Grain-boundary amorphization mitigates lithium dendrite growth, but the mechanism is debated (Nature Communications 2025).
Hypothesis. Dendrites nucleate at grain-boundary electronic traps, not at geometric defects. Amorphization works by removing mid-gap traps that mediate electronic leakage; the critical current density scales with grain-boundary electronic conductivity, not with ionic conductivity.
Mechanism. Electrons leak through grain-boundary states and reduce Li⁺ at the interface, creating a filament. A trap-free boundary suppresses the leakage while leaving ionic transport largely unchanged.
Decisive test. Measure local electronic leakage and critical current density on the same grain boundaries. Electron-blocking interlayers should raise the critical current without changing bulk ionic conductivity; single crystals should be immune.
Kill condition. Critical current tracks ionic conductivity, not electronic leakage.
Nearest prior / delta. Grain-boundary amorphization is reported. The electronic-trap nucleation criterion, with a critical-current-versus-leakage scaling, is the new claim.
Biology
H21 — Mycorrhizal “transfer” is hydraulic redistribution
Anomaly. The common-mycorrhizal-network literature reports resource transfer between trees, but a 2023 critique documents positive citation bias and overinterpretation of tracer studies (Nature Ecology & Evolution critique).
Hypothesis. Most apparent transfer is passive hydraulic redistribution down a water-potential gradient through shared hyphae and soil, not directed transfer from a “mother tree.”
Mechanism. Water moves from wet to dry soil along a potential gradient. Tracers dissolved in that water move with it. When the gradient reverses, the apparent transfer reverses.
Decisive test. Manipulate water potential independently of carbon demand. If transfer follows the water gradient rather than the carbon source–sink gradient, the hydraulic hypothesis wins. ATP inhibitors should not block it.
Kill condition. Transfer against the water gradient, blocked by metabolic inhibitors.
Nearest prior / delta. The critique is published. Proposing hydraulic redistribution as the quantitative null, with a reversal test, is the constructive addition.
H22 — Dark fungi are unculturable because of alternative oxidase
Anomaly. Environmental sequencing keeps finding “dark fungi” — diverse lineages with no cultured representative. They resist standard isolation and often grow extremely slowly (Tedersoo et al. 2025).
Hypothesis. Dark fungi rely on alternative oxidase (AOX) to keep respiration running under oligotrophic, oxidative stress. AOX suppresses reactive oxygen and enforces slow growth, which also makes them phenotypically tolerant to azoles.
Mechanism. AOX bypasses the cytochrome pathway, reducing mitochondrial ROS production at the cost of lower ATP yield. Slow growth and low ROS make standard culture and azole assays fail.
Decisive test. AOX inhibitors should increase culturability on low-nutrient media and reverse the slow-growth azole tolerance. AOX expression should be high in environmental samples.
Kill condition. No AOX activity, or no change with inhibitors.
Nearest prior / delta. AOX is known in fungi. Linking it to the dark-fungi isolation barrier, and testing an inhibitor cocktail, is new.
H23 — Coral heat tolerance travels in vesicles
Anomaly. A 2026 preprint reports that symbiont-derived vesicle RNA sustains coral mutualism, and that loss of the vesicles drives succession (Research Square preprint).
Hypothesis. Heat tolerance is transferred horizontally between corals by symbiont extracellular vesicles. The RNA payload is a form of thermal memory: it reprograms the host’s stress response before the next heat wave.
Mechanism. Vesicles carry small RNAs and proteins from the symbiont to the host. The cargo alters host gene expression, stabilising the symbiosis under thermal stress.
Decisive test. Isolate vesicles from heat-primed symbionts, transfer them to naive corals, and test bleaching under a controlled heat ramp. Small-RNA sequencing should show a heat-dependent cargo.
Kill condition. No transfer of tolerance, or vesicles lacking RNA.
Nearest prior / delta. The preprint describes the vesicles. Horizontal thermal memory via vesicle transfer is the new, testable extension.
H24 — Tardigrade radiotolerance is repair-limited
Anomaly. Tardigrades survive extreme radiation, and the Dsup protein protects DNA, but the relative contributions of shielding and repair are unresolved (Dsup structural analysis 2025).
Hypothesis. Dsup is a chromatin shield that buys time, not a complete protection system. Survival is limited by the tardigrade’s DNA-repair capacity; Dsup alone should give only modest protection.
Mechanism. Dsup binds nucleosomes and limits hydroxyl-radical access to DNA, reducing the lesion rate. The remaining lesions must be repaired; if repair is saturated, the cell dies.
Decisive test. Express Dsup alone in repair-deficient cells: protection should be modest. Knock out repair genes in tardigrades: tolerance should collapse. Comet assays should show residual damage after irradiation.
Kill condition. Dsup alone confers full tolerance, with no repair requirement.
Nearest prior / delta. Dsup is known as a DNA protector. The claim that repair, not Dsup, sets the survival ceiling is the new emphasis.
H25 — Diatom silica patterns are phase-separated templates
Anomaly. Diatoms build silica shells with species-specific pore patterns. Organic molecules and phosphate ions are known to affect the pattern, but the mechanism that sets the length scale is unresolved (modelling study 2025).
Hypothesis. Cationic silaffin peptides and phosphate undergo liquid–liquid phase separation, creating a condensate template. Phosphate concentration sets the condensate spacing, and therefore the pore period.
Mechanism. Silaffins bind phosphate and silica precursors. The mixture crosses a phase boundary, forming micron-scale droplets. Silica condenses at the droplet interfaces, leaving pores where the droplets were.
Decisive test. In-vitro mixtures should form condensates with a spacing that tunes with phosphate concentration. SAXS and microfluidics should show the same period as the biosilica.
Kill condition. No condensates, or pore period independent of phosphate.
Nearest prior / delta. Organic–inorganic templating is proposed. A phosphate-tuned phase-separation mechanism is the specific new claim.
Technology & computation
H26 — Sodium-ion cells fail at low temperature at the SEI
Anomaly. Sodium-ion batteries lose much of their capacity at low temperature, and the rate-limiting step is debated: bulk diffusion, desolvation, or charge transfer at the electrode–electrolyte interface (low-temperature interface study 2026).
Hypothesis. The rate-limiting step is Na⁺ desolvation at the solid–electrolyte interphase, not bulk diffusion. A weakly solvating electrolyte should therefore improve low-temperature performance more than a thinner SEI.
Mechanism. At low temperature, the Na⁺ solvation shell is slow to shed at the SEI. The resulting interfacial resistance dominates. Weakly coordinating solvents lower the desolvation barrier.
Decisive test. Temperature-dependent impedance should show an interfacial activation energy that tracks the solvent’s donor number. Bulk diffusion coefficients should be too fast to explain the capacity loss.
Kill condition. Capacity loss tracks bulk diffusion, with no solvent dependence.
Nearest prior / delta. Interfacial effects are studied. Making desolvation the single rate-limiting step, with a solvent-design prediction, is the new claim.
H27 — DNA storage repair has a lesion-density optimum
Anomaly. DNA is attractive for archival storage, but it degrades. A patent describes enzymatic DNA repair as a way to extend storage life (US12503722B2).
Hypothesis. Periodic repair extends half-life only below a critical lesion density. Above it, repair enzymes create double-strand breaks faster than they fix single-base damage, so repair becomes harmful.
Mechanism. Base-excision repair creates abasic sites and nicks. If two lesions are close on opposite strands, the repair tracts collide and produce a double-strand break. At high lesion density, collisions dominate.
Decisive test. Accelerated ageing at controlled humidity and temperature, with repair applied at different intervals. Error rate should show an optimum; above the critical density, repair should increase fragmentation.
Kill condition. Repair always helps, or always harms, with no density optimum.
Nearest prior / delta. Enzymatic repair is patented. The lesion-density threshold, and the prediction that over-repair is destructive, is new.
H28 — Dendritic shot noise is a Bayesian resource
Anomaly. Multi-gate ferroelectric transistors can mimic dendritic computation, but the devices are noisy. Noise is usually treated as a defect to be engineered away (OSTI 2025).
Hypothesis. Dendritic shot noise implements Bayesian sampling. At low power, a controlled amount of noise improves inference by exploring posterior uncertainty, so noise is a resource.
Mechanism. Stochastic channel opening produces Poisson-like fluctuations in the dendritic current. These fluctuations act as a sampling kernel, approximating a posterior distribution rather than a single point estimate.
Decisive test. Inject calibrated noise into a dendritic network performing a probabilistic task. Accuracy should peak at a nonzero noise level and degrade at both extremes. The optimal noise should scale with task uncertainty.
Kill condition. Noise always degrades inference.
Nearest prior / delta. Dendritic devices are demonstrated. Treating shot noise as a Bayesian sampling resource, with a noise-optimum prediction, is new.
H29 — Microcombs self-lock by optomechanical backaction
Anomaly. Microcombs can be stabilised by self-injection locking or piezoelectric actuators, but the mechanisms are treated as external control (IEEE 2025).
Hypothesis. A soliton microcomb can self-stabilise through optomechanical backaction: radiation pressure in the resonator creates a mechanical memory that suppresses soliton jitter, removing the need for an external locking laser.
Mechanism. The soliton’s optical power modulates the resonator’s mechanical deformation. The mechanical response low-pass filters frequency noise, creating a self-locking window.
Decisive test. Phase noise should scale with mechanical Q and detuning. Devices with different mechanical Q should show different locking ranges; damping the mechanical mode should destroy the effect.
Kill condition. No dependence on mechanical Q, or no self-locking window.
Nearest prior / delta. Self-injection locking is known. Optomechanical backaction as the stabilising mechanism is the new claim.
H30 — Grokking is a first-order transition with hysteresis
Anomaly. Grokking — delayed generalisation in small networks — looks like a phase transition, but the order of the transition is debated. Recent work maps accuracy regimes in L2-regularised networks (arXiv:2505.06597).
Hypothesis. Grokking is a first-order transition with hysteresis. There is a nucleation barrier between the memorising and generalising phases; sweeping weight decay up and down traces different paths.
Mechanism. The loss landscape has two basins separated by a barrier. Training noise and regularisation determine which basin the network occupies. A sweep changes the barrier height and produces hysteresis.
Decisive test. Sweep weight decay upward and downward across the transition. The generalisation threshold should differ between directions, with a finite-size scaling of the hysteresis width. Add a nucleation seed (a partially trained solution) and the transition should move.
Kill condition. A continuous transition, with no hysteresis or nucleation effects.
Nearest prior / delta. Grokking as a phase transition is proposed. The first-order, hysteretic formulation and the sweep test are new.