Does matter remember its history?
Every equivalence-principle test ever run has compared primordial matter with primordial matter. NuGrav will build a gravitational torsion balance whose test bodies were forged in nuclear fission — and ask, for the first time, whether provenance changes how matter attracts.
Can nuclear fission demonstrate gravitational memory effects?
Is provenance physical?
Quantum field theory says two nuclei of the same isotope are identical, full stop. A ⁹⁸Mo nucleus assembled in a star five billion years ago and one assembled by nuclear fission should fall towards the fission sibling mass at exactly the same rate. No experiment has ever checked this asymmetry.
Read the mission →Fission-born molybdenum
Electrorefining of spent metallic reactor fuel leaves anode fines rich in noble-metal fission products. The molybdenum in them is the most stable and handleable macroscopic matter on Earth whose nuclei are paired via fission with a nearby attractor mass.
See the science →Disposal is scheduled
The parent material is slated for waste disposition in 2028. Once it is stabilized and buried, no comparable sample will exist anywhere. There is an urgency dictated by the circumstance of our experimental design.
See the timeline →Test whether the interaction history of matter changes how it gravitates.
NuGrav is a time-limited nonprofit research organization with a single deliverable: the first direct test of the equivalence principle across particle provenance.
The blind spot
The universality of free fall is among the best-tested facts in physics. Torsion-balance experiments and the MICROSCOPE satellite have compared how different materials fall to parts in 10E13 – 10E15 , constraining new forces that couple to composition — baryon number, lepton number, neutron excess.
But every test mass in every one of those experiments shares the same biography. Its protons and neutrons were bound into nuclei in the Big Bang and in stars, billions of years ago. If a gravity-like force couples not to what matter is but to how and when it was made — a charge carried by provenance — the entire experimental record to date would be silent on it.
The opportunity
Pyroprocessing of spent metallic reactor fuel concentrates noble-metal fission products — molybdenum chief among them — into electrorefiner anode fines. The molybdenum there was synthesized by fission during reactor operation: nuclei tens of years old, not billions.
Separated and purified, this material can be machined into test bodies for a rotating torsion balance and compared, gram for gram, against natural molybdenum. A differential acceleration toward the Sun, the galaxy, or a local source would be the signature of a provenance-coupled force.
The custodial inventory is scheduled for waste disposition in 2028. NuGrav exists to recover a scientific instrument from a waste stream before the door closes.
Built like an FRO because nothing else fits the ambition or the clock.
Too fast for grants
Federal proposal cycles routinely take 12–18 months from submission to funds-on-hand. Against a 2028 disposition deadline, that timeline forfeits the experiment before it starts. Philanthropy can move in weeks.
Too focused for academia
This is a single, theoretically-syncretic, engineering-heavy measurement campaign. The FRO model was designed for exactly this shape of problem.
Too fundamental for a startup
The deliverables are a published result, open data, open hardware drawings, and an archived set of isotopically characterized test masses held for future science. There is nothing to sell and everything to learn.
The anyonic nucleus, and the balance built to judge it.
The experiment first, in three steps: recover fission-born molybdenum from electrorefiner anode fines, machine it into pendulum bodies beside non-fissogenic controls, and read the twist on a rotating torsion balance. Then, for those who want it, the idea in three moves: statistics is topology, the nucleus is a topological object, and how a nucleus was assembled can leave a trace.
1 · The material
During electrorefining of spent metallic fuel, uranium is dissolved into molten salt and recovered at the cathode. What stays behind in the anode baskets — the anode fines — is enriched in noble-metal fission products: Mo, Ru, Rh, Pd, and Tc.
NuGrav's separations campaign extracts the molybdenum fraction and purifies it to instrument grade. Technetium, actinides, and activation products removed to background, verified by independent assay. What remains is chemically ordinary molybdenum metal.
Fission is pairwise, so this material has siblings. Every event that made a molybdenum nucleus on the light peak also made a heavy partner in the Xe → Cs chains — and the partners never left. They decayed inside the fuel within minutes, and their cesium now sits dissolved in the electrorefiner's salt, meters from where the molybdenum is recovered (FIG. 1). The test mass and the attractor mass were born in the same events.
Alongside the fission-born bodies, the pendulum carries controls machined from non-fissogenic molybdenum — natural-origin metal put through the same chemistry, the same furnace, and the same lathe. Whatever twist the balance finds must then be laid at provenance's door, not the material's.
FIG. 1 — Thermal-fission mass-yield curve of ²³⁵U (schematic, log scale). Fission is pairwise: one light and one heavy fragment per event, A + A′ + ν ≈ 236 — e.g. ²³⁵U + n → ⁹⁵Sr + ¹³⁹Xe + 2n, whose chains end at ⁹⁵Mo and pass through ¹³⁹Cs. NuGrav takes the light-peak molybdenum chains (A = 95–100) out of the anode fines and machines them into test bodies; every one of their heavy partners (A ≈ 133–139, Xe → Cs) stays behind, dissolved in the fuel and salt — which is why the electrorefiner is the attractor mass.
2 · The measurement
The instrument is a classic rotating torsion balance: a composition dipole of Mo test bodies hangs from a fine fiber inside a vacuum chamber on a continuously rotating turntable. An autocollimator reads the pendulum's twist at the nanoradian level.
A provenance-coupled force sourced by the Sun, the galactic center, or a laboratory attractor mass would appear as a twist signal at the turntable rotation frequency, with a phase that tracks the source. Rotation moves the physics away from slow drifts; the non-fissogenic control pair, run in the same apparatus, kills instrument systematics.
The strongest conceivable local source is the electrorefiner itself: its fission-product-laden salt is provenance-charged matter in bulk. A near-field run beside it — extra pull on the fission-born group, none on the natural group, a net twist about the fiber — is the sharpest test of all, and the reason the apparatus is designed to travel.
Target: differential acceleration sensitivity η ≲ 10⁻¹¹ in the first science run, with an instrument path toward 10⁻¹². A null result at that level sets the first limits ever placed on provenance-dependent interactions. A signal would mean quantum field theory's most basic assumption — that identical particles are identical — has an exception.
FIG. 2 — The near-field configuration, in plan view looking down the fiber (twist exaggerated). Four fission-born bodies group on one side of the rotor, four natural bodies on the other, inside a vacuum chamber on a continuously rotating turntable beside the electrorefiner — whose salt holds the Xe → Cs partners of the very fission events that made the test bodies (inset: the two peaks of the yield curve end up facing each other across the gap). If provenance couples to provenance, the fission-born group feels an extra pull δF toward the salt, the natural group does not, and the pendulum twists at the rotation frequency.
3 · Indistinguishability has fine print
Quantum mechanics allows an exchanged pair of identical particles to differ from the original state by a phase at most, and in three dimensions the standard argument permits exactly two: +1 or −1. Swap twice and you have done nothing at all; the permutation group leaves no room in between. Bosons, fermions, full stop — and with that, the claim that a ⁹⁸Mo nucleus carries no memory of its birth.
The plane is different. In two dimensions an exchange path can wind, winding is counted by the braid group rather than the permutation group, and the exchange phase e^{iθ} may sit anywhere on the circle. Leinaas and Myrheim saw this in 1977; Wilczek named the resulting particles anyons in 1982; fractional quantum Hall interferometers finally caught them braiding in 2020. Fractional statistics is not a fantasy. It is geometry.
Now read the three-dimensional argument closely: it leans on one quiet assumption — that the particles are points. Extended objects obey different topology. Loops in ordinary 3-space can link, knot, and pass through one another; their exchanges are classified by the loop braid group; and Baez, Crans & Wise showed that string-like excitations of a 4d BF field theory really do carry exotic exchange statistics — in three spatial dimensions, with no contradiction anywhere. The door is not locked. It is merely closed to points.
FIG. 3 — Exchange for extended objects, drawn after Baez, Crans & Wise (2007): the worldsheet of a moving loop is a tube. Left: two loops trade places by passing around one another — the swap, the only exchange available to points, giving ψ → ±ψ. Right: a loop can also pass through a loop — the slide, an exchange with no point-particle counterpart. Swaps and slides together generate the loop braid group, and the slide may carry any phase e^{iθ}. Nuclei, on our hypothesis, exchange like this.
4 · Four clues, one hypothesis.
Each clue below is established physics — flux tubes, topological field theory, θ-vacua, the double copy. The assembly of them into a single hypothesis is ours, and the assembly is what the balance will judge.
The nucleus is not made of points
The string picture of hadrons is older than QCD itself: lattice simulations show quarks bound by string-like gluon flux tubes, and in the Skyrme picture a nucleon is a topological soliton whose baryon number is literally a winding number. A nucleus is a dense tangle of extended chromodynamic structure. The precondition for statistics beyond Bose and Fermi — extension — is already present in every nucleus on Earth.
BF theory supplies the phases
BF theories are field theories with no local dynamics at all — only topology — and in four dimensions they endow their particles and flux loops with exotic braiding statistics; Baez and collaborators made this exact. 't Hooft's dual-superconductor picture of confinement equips QCD with precisely this kind of two-form structure. A hidden statistical sector inside nuclear matter is the sort of thing BF-like physics is built to host.
θ-vacua make it writable
The QCD vacuum carries a topological angle, θ, and Peccei–Quinn dynamics promotes it from constant to field — a quantity free to evolve over cosmic time. Our hypothesis: a nucleus freezes in the ambient θ-like phase of its gluon field at assembly. Nearly every nucleus on Earth was forged in stars before the Sun existed; a fission fragment was assembled in a reactor within living memory. If the phase has drifted over the billions of years between, birth date — provenance — becomes a quantum number.
The double copy lets gravity read it
Bern, Carrasco & Johansson found that gravity amplitudes are gauge amplitudes squared — an exact, term-by-term dictionary between gluons and gravitons. If a topological label lives in a nucleus's gluon field, the double copy is the pipeline by which it could surface on the gravitational side, as a small, sector-selective addition to how that nucleus attracts. Which is precisely the quantity a torsion balance measures.
5 · The hypothesis, assembled
Chain the clues. Nuclei are extended topological objects (01), of exactly the kind that can carry exchange phases beyond ±1 (02); such a phase can plausibly be written at formation (03); and gravity possesses a documented channel for reading gluon-level structure (04). The conclusion we hypothesize is the anyonic nucleus: two ⁹⁸Mo nuclei, identical in every tabulated quantum number, distinguishable by topological sector — and coupling to matter of their own sector with a small additional force.
The corollary: pairs
Fission is a violent, topology-changing event, and both fragments of a given event share it. The molybdenum NuGrav recovers and the Xe → Cs partners fission gave it (§1) are not merely material of the same vintage — they are opposite halves of the same events. If sector couples to sector, the largest signal within reach is between our test bodies and the fission-product salt they were separated from: the near-field attractor run of §2 (FIG. 2) exists because the hypothesis demands it be tried first.
What this is — and what it isn't
It is a conjecture, not a derivation. We cannot yet compute the η this mechanism predicts, and the mainstream expectation is a clean null: quantum field theory as written says identical is identical. Our claim is that the assembly above is coherent enough to be worth a falsification — not that it is right.
Existing tests of statistics — Pauli-violation searches, exchange-symmetry spectroscopy of molecules — constrain electrons, photons, and rotational states. None ties statistics to formation history, and none reads it out gravitationally. That is the gap the balance walks into. If the twist is there, this page becomes the outline of a research program. If it is not, the hypothesis dies in public at η ≈ 10⁻¹¹ — which is what hypotheses are for.
READING — LEINAAS & MYRHEIM 1977 · WILCZEK 1982 · SKYRME 1961 · 'T HOOFT 1978 · PECCEI & QUINN 1977 · BAEZ, CRANS & WISE, EXOTIC STATISTICS FOR STRINGS IN 4D BF THEORY, 2007 · BERN, CARRASCO & JOHANSSON 2008
Safety, in one paragraph
Molybdenum's fission isotopes are stable; the hazard in anode fines comes from their neighbors, chiefly ⁹⁹Tc and residual actinides. All separations occur at the custodial laboratory under its licensing, by qualified radiochemists. Test bodies are released for metrology only after assay confirms activity indistinguishable from background. The finished pendulum is, radiologically, a lump of metal.
Open by default
NuGrav commits to publishing the full dataset, analysis code, instrument drawings, and the isotopic assay of every test body. The characterized masses themselves will be archived and made available to future experiments — a permanent scientific asset recovered from a waste stream.
Big idea. Small team.
NuGrav runs lean: a projected core staff of four to five, with an external advisory board for theory and analysis review.
Noah Harris
Founder & CEOExperimental gravitation. Ten years on torsion-balance tests of the equivalence principle; leads instrument design, science strategy, and the blind-analysis protocol.
Charles Fleck
Chief EngineerPrecision mechanics and vacuum systems. Responsible for fiber metrology, the rotating platform, thermal control, and getting the chamber below 10⁻⁵ Pa and keeping it there.
Dalton Jones
Theory AdvisorPhenomenology of fifth forces and Yukawa interactions. Advises on source modeling — solar, galactic, and local attractor — and on how to parameterize a provenance charge.
Open roles
Join the campaignHiring in 2026–27: a metrology postdoc (torsion-balance instrumentation) and a contract precision machinist experienced with refractory metals. Write to join@nugrav.org.
Profiles shown are placeholders for this concept site. Founding-team announcements begin Q4 2026.
Thirty months, fixed endpoint.
The schedule runs backward from a date NuGrav does not control: the 2028 disposition of the electrorefiner material. Every milestone below exists to make the science runs happen before that door closes.
-
01Q3 2026 — Now
Formation & design
NuGrav incorporated; seed gifts received; conceptual design review complete. Fundraising open and material-access negotiations underway with the custodial laboratory.
In progress · July 2026 -
02Q4 2026
Close funding · sign agreements
Full $1.2M committed. Material-transfer, licensing, and hot-cell work agreements executed. Long-lead procurement begins: fiber stock, turntable, autocollimator.
-
03Q1 2027
Separation campaign
Molybdenum fraction extracted from anode fines and purified at the custodial laboratory. Independent isotopic and radiological assay; material released for fabrication.
-
04Q2 2027
Test-body fabrication
Machining of the fission-born bodies and their non-fissogenic controls to matched mass, geometry, and surface finish. Balance assembly begins in parallel.
-
05Q3–Q4 2027
Commissioning
Fiber characterization, chamber pump-down, turntable servo tuning. Systematic-error budget measured and locked: gravity gradients, thermal, tilt, magnetic.
-
06Q1–Q2 2028
Science runs I & II
Two blinded data campaigns spanning distinct source geometries (solar and galactic phases, plus local attractor). Analysis proceeds behind a blind on the twist-signal amplitude.
-
072028
Material disposition deadline
The remaining electrorefiner inventory is scheduled for stabilization and waste disposition. After this point, no comparable fission-born material will be recoverable — anywhere, at any price.
Hard deadline -
08Q3–Q4 2028
Unblinding, publication & archive
Results unblinded and submitted for publication. Full dataset, analysis code, and drawings released. Characterized test masses transferred to a permanent archive for future experiments. NuGrav winds down, on schedule.
$1.2 million buys the whole experiment.
Not a program. Not an institute. One instrument, one measurement campaign, one published answer — fully funded, start to finish, for less than the cost of a single year of a typical lab.
LAST UPDATED JULY 2026 · COMMITMENTS FROM 11 DONORS · AMBER NEEDLE MARKS FUNDS IN HAND
Where the money goes
- Personnel — 30 months, ~3.5 FTE average$565,00047%
- Apparatus & instrumentation — balance, chamber, turntable, autocollimator$285,00024%
- Material separation, transport & compliance$170,00014%
- Facility & operations — lab space, vacuum, utilities$95,0008%
- Contingency & end-of-life disposition$85,0007%
What a gift buys
A spool of tungsten fiber stock — the thread the whole measurement hangs from, literally.
One test body: purified, machined, assayed, and engraved with its isotopic certificate.
The autocollimator — the instrument's eyes, reading twist angle at the nanoradian level.
The entire separations campaign: the step that turns a waste stream into a physics experiment.
How giving works
NuGrav operates under a 501(c)(3) fiscal sponsor while its own determination is pending; gifts are tax-deductible to the extent the law allows. We accept major gifts, donor-advised-fund grants, and foundation support. Finances are audited annually and the budget above is published and versioned.
A technical prospectus — full instrument design, error budget, and separations flowsheet — is available to prospective donors on request.
Frequently Asked Questions
What is a Focused Research Organization?
A time-limited nonprofit built to produce one well-defined scientific output that doesn't fit academia's incentives or a startup's economics. NuGrav follows the pattern strictly: fixed budget, fixed schedule, a single deliverable, then a planned wind-down.
Is the material radioactive? Is this safe?
The fission isotopes of molybdenum itself are stable. The hazards in raw anode fines come from co-located species — chiefly ⁹⁹Tc and residual actinides — which is why all separations happen at the custodial laboratory, under its license, by qualified staff. Test bodies finish chemical processing only after independent assay shows activity indistinguishable from background. The apparatus NuGrav proposes contains, radiologically speaking, ordinary metal.
Didn't MICROSCOPE already test the equivalence principle to 10E-15?
Yes — for composition. MICROSCOPE compared titanium against platinum; torsion-balance experiments have compared beryllium, titanium, aluminum, and more (to test varying proton-ratio). Every one of those test masses is primordial matter. NuGrav varies a different axis entirely: chemically identical molybdenum bodies that differ in interaction history. Existing experimental limits don't constrain a force that couples to provenance (interaction history).
What would a positive signal even mean?
That the indistinguishability of identical particles, a foundational assumption of quantum field theory, fails at the level of gravitating fission fragments. This would offer a practical foothold towards using nuclear waste to engineer spacetime structures for technological applications. It would also ignite theoretical development on the interface between string theory, QCD, and gravity.
And if you see nothing?
Then we publish the first experimental limits ever set on provenance-coupled interactions, publish the data and experimental procedure, and place a permanent archive of isotopically characterized fission-born test masses in trust for future experiments. A clean null from a well-built instrument is a real result and reference point for any others who might explore these ideas.
Why molybdenum, of all elements?
It sits at the peak of the light fission-fragment yield curve, so spent fuel is genuinely rich in it; its fission isotopes are stable, so purified metal is safe to handle and non-emissive; it's a refractory metal that machines into excellent, dimensionally stable test bodies; and natural molybdenum is cheap and plentiful for control masses. No other fission waste element checks all four boxes.
Why can't this wait, or use other material later?
The electrorefiner inventory is scheduled for stabilization and disposal in 2028, and no other separable, macroscopic stock of fission-born metal exists outside active waste streams. Producing fresh material from scratch would require irradiating and reprocessing fuel — politically, legally, and financially out of reach for a physics experiment. This is a golden yet compressed window of opportunity.
How are funds governed?
Through a 501(c)(3) fiscal sponsor with annual independent audit, a published budget, and quarterly donor reports against the milestone schedule on the Timeline page. Unspent contingency at wind-down is returned to the sponsor's charitable pool or redirected per donor agreement.