From radioactive tritium to maps of distant galaxies, neutrino experiments are narrowing the possibilities. Their results become most powerful when the differences between their measurements are understood.
How do you weigh a particle that almost always passes through your apparatus without interacting? For neutrinos, physicists have developed several answers. They measure radioactive decay, track changes in particle identity, search for extraordinarily rare nuclear transitions and reconstruct the growth of cosmic structure. Each approach reveals something different about mass, and each brings its own assumptions.
That diversity is becoming more valuable as the numerical limits tighten. A laboratory result and a cosmological constraint cannot simply be placed on the same league table. One concerns a particular combination of neutrino masses, another their sum, while a third depends on whether neutrinos and antineutrinos are fundamentally the same particles. Europe’s contribution is therefore best understood as part of a connected measurement programme, with experiments capable of challenging one another’s interpretation.
What oscillations establish
Neutrinos are produced and detected in three familiar flavours: electron, muon and tau. These are mixtures of three states with definite masses. As neutrinos travel, the mixture changes, producing the phenomenon called oscillation. The pattern depends on differences between squared masses rather than their absolute starting point. Oscillation experiments can establish that the masses differ without determining the mass of the lightest state.
Daya Bay provides a precise example. Its final reactor antineutrino analysis used 5.55 million candidate events collected over 3,158 days. For the normal ordering of masses, the collaboration obtained a mass-squared difference of 2.466 ± 0.060 thousandths of an electronvolt squared. The result describes spacing within the neutrino spectrum; it is not a measurement of an individual neutrino weighing 2.466 thousandths of an electronvolt.
An analogy is knowing the distance between floors without knowing how high the building stands above sea level. Better measurements of the spacing sharpen the problem. They cannot supply the missing reference level by themselves.
KATRIN examines the missing energy
Germany’s KATRIN experiment approaches that reference level through tritium decay. The available energy is shared among the decay products. Very close to the electron spectrum’s endpoint, the neutrino’s mass changes the shape of the distribution. KATRIN measures that tiny distortion rather than capturing and weighing the escaping antineutrino.
The collaboration’s 2025 Science paper analysed 36 million electrons from 259 measurement days. It reported an upper limit of 0.45 electronvolts divided by the speed of light squared, or eV/c², at 90% confidence. This constrains an effective electron-antineutrino mass: a weighted combination of the three squared masses, followed by a square root. It does not establish that the neutrino mass equals 0.45 eV/c².
The distinction matters when setting expectations. Improving sensitivity can produce a tighter upper limit even if the true mass remains below reach. A measurement programme can therefore advance substantially without yet reporting a positive mass determination.
New instruments change the measurement problem
Project 8 takes a different route to measuring electron energy. An electron moving through a magnetic field emits radiation whose frequency depends on its energy. The collaboration used this cyclotron radiation emission spectroscopy to analyse tritium decay, reporting a Bayesian upper limit of 155 eV/c² at 90% credibility in its first neutrino-mass result. This was much less restrictive than KATRIN, but demonstrated a new measurement principle on the relevant radioactive source.
The proposed progression includes atomic tritium. Molecular decay can leave energy in excited molecular states, complicating the endpoint spectrum. Atomic sources could remove that particular complication, but creating, cooling and containing sufficient atomic tritium introduces new difficulties. The collaboration’s development programme identifies 40 millielectronvolts as a sensitivity goal, not an achieved result.
HOLMES instead embeds radioactive holmium-163 in very cold microcalorimeters and measures the heat released after electron capture. Its 2025 result used roughly 70 million decays collected over two months and placed a 27 eV/c² upper bound at 90% credibility. The detector array achieved an average energy resolution of 6 eV. The result tests a scalable calorimetric approach, while leaving considerable work before competitiveness at the sub-electronvolt scale.
These developments make detector technology a scientific choice. A radio-frequency signal, an electrostatic energy filter and a thermal pulse expose different sources of error. Their value lies partly in making future agreement harder to attribute to a shared instrumental weakness.
Cosmology weighs a population
Cosmology examines how neutrinos influence cosmic expansion and structure. Combining DESI observations with cosmic microwave background data gave a 95% upper limit of 0.064 eV on the neutrino mass sum under the standard Lambda cold dark matter model. Allowing a two-parameter evolving dark energy model relaxed the bound to 0.16 eV. The model therefore changes the inferred limit.
The comparison has implications for how different research communities communicate. An instrument team can publish a well-characterised observational dataset without uniquely settling every physical interpretation of it. A cosmologist can then test that dataset against several frameworks, while a laboratory team investigates a related particle property independently. The strength of this arrangement is the possibility of revisiting one part of the argument without discarding the others. Keeping measurements, statistical analysis and physical interpretation distinguishable makes the combined programme more durable.
This does not diminish their scientific value. It changes the question they answer. A disagreement with a future direct laboratory measurement would require investigators to examine astrophysical modelling, experimental systematics and possible new physics together. Choosing whichever number looks most precise would discard the opportunity that the disagreement creates.
A rare decay asks a further question
Neutrinoless double-beta decay searches add another dimension. In the familiar two-neutrino process, a nucleus emits two electrons and two antineutrinos. The hypothetical neutrinoless version would emit the electrons without accompanying neutrinos. It would violate lepton-number conservation and provide evidence that cannot be obtained from an endpoint mass limit alone.
The complete KamLAND-Zen analysis combined successive xenon datasets to set a lower bound of 3.8 × 10²⁶ years on xenon-136’s neutrinoless double-beta decay half-life at 90% confidence. Under the standard light-Majorana-neutrino interpretation, this translates into upper limits on an effective Majorana mass ranging from 28 to 122 millielectronvolts. The range reflects different nuclear matrix-element calculations, which describe the nuclear transition.
That effective mass combines neutrino contributions coherently, allowing cancellations. Its interpretation also assumes a mechanism for the decay. A null result therefore cannot be read as a universal upper bound on each neutrino’s mass, or as proof that neutrinos are different from their antiparticles. The particle question and the nuclear calculation remain intertwined.
Where the comparisons become decisive
The experimental landscape becomes clearer when each result retains its own label: an oscillation spacing, a direct effective mass, a cosmological sum or a decay half-life interpreted through a particular mechanism. Their numerical values need not coincide. What must eventually fit is a common physical account of the neutrino states and their interactions.
For research organisations and technology suppliers, this argues for complementary capabilities. Better cryogenic readout can advance a calorimetric programme; improved source control can strengthen endpoint spectroscopy; more reliable nuclear calculations can make rare-decay results easier to interpret. Progress in one area does not eliminate the need for the others.
Consider a future situation in which a laboratory reports a mass signal but a cosmological analysis retains a lower upper limit. The first task would be to translate between the effective mass and the mass sum using oscillation information. Researchers would then test whether the apparent disagreement survives the uncertainty in both results. Only after that comparison could they examine which assumptions might need revision. Conversely, agreement would strengthen confidence in both the laboratory result and the cosmological framework. This is why preserving the distinctions now is practical preparation for discovery, rather than merely a matter of terminology. The same discipline prevents a proposed instrument’s sensitivity goal from being mistaken for evidence that the relevant mass range has already been explored.
The compelling prospect is an overconstrained answer: enough independent information that neutrino mass can no longer move freely within every explanation. Reaching that point requires more than announcing the smallest number. It requires understanding precisely which quantity each apparatus has brought within reach.
