Ghost particles: the neutrinos passing through you right now

Hold up your thumb. In the second it took to read that, about 65 billion particles from the Sun went straight through every square centimeter of you, an area about the size of your thumbnail [Reference 1]. They were made a little over eight minutes ago, in the Sun’s core. Tonight, with the Sun below the horizon, they will keep coming, up through the floor, after crossing thousands of kilometers of rock. You will feel nothing. These particles are neutrinos: tiny particles with no electric charge that almost never touch anything [Reference 2,Reference 3].

Almost never is not never: by our own estimate, from a real detector’s counting rate, about once a week one of those solar neutrinos bumps into an electron somewhere inside your body. This article is about how we know, and why, for more than thirty years, up to two-thirds of the expected ones seemed to be missing [Reference 4,Reference 3].

Made in the core

The Sun shines because of nuclear fusion: light atomic nuclei joining to make heavier ones and releasing energy as they do. Its core is at about 15 million °C [Reference 5]. That is so hot that hydrogen atoms have lost their electrons, leaving bare protons in a super-hot, electrically charged gas called a plasma [Reference 5].

The first step is the hard one. Two protons repel each other and cannot form a nucleus on their own. Even when the heat slams them together, they only stick if one of them turns into a neutron at that instant. Only one force in nature can do that: the weak interaction, the force that can change one kind of particle into another. When it succeeds, the pair becomes a deuteron, a heavy-hydrogen nucleus made of one proton and one neutron. Two new particles fly out: a positron, the antimatter twin of the electron, and a neutrino [Reference 6]. Neutrinos come in three types, named after the electron and its two heavier, short-lived cousins, the muon and the tau. This one is an electron neutrino; the Sun makes only electron neutrinos [Reference 2].

The weak force lives up to its name: a given proton in the solar core waits, on average, about ten billion years before it takes part in this reaction [Reference 6]. The Sun is only about 4.6 billion years old, so most protons in its core are still waiting [Reference 5].

After a few more steps, four protons end up as one helium nucleus; two protons must become neutrons, so in the main branch the first step happens twice. That releases 26.73 million electronvolts. An electronvolt (eV) is the energy an electron picks up crossing one volt; a million of them is written MeV. Physicists also give masses in these units, using E = mc²: an electron weighs 511,000 eV. Between them, the two neutrinos carry off about 0.63 MeV on average and leave the Sun for good. The rest heats the Sun and eventually leaves as sunlight. By our calculation from the reaction energies and the solar model, neutrinos take away about 2.4% of the Sun’s total power (0.63 of every 26.73 MeV) [Reference 6,Reference 1].

That bookkeeping is enough to count them. Each helium nucleus leaves about 26 MeV behind as heat and comes with two neutrinos, so each neutrino is like a receipt for about 13 MeV of sunshine. The Sun radiates 3.828 × 10²⁶ watts of light [Reference 7]. Divide that power into 13-MeV packets and you find the Sun must make about 1.8 × 10³⁸ neutrinos per second. Spread them over a sphere the size of the Earth’s orbit, and roughly 65 billion cross each square centimeter every second [Reference 7,Reference 6].

The standard solar model, physicists’ best computer model of the Sun’s interior, agrees on the total, as it must, since the model is tuned to match the Sun’s brightness. What it adds is the breakdown: which steps make which neutrinos [Reference 1]. About 91% come from that first proton–proton step; physicists call the whole sequence it starts the proton–proton chain. About 7.5% come from a later step involving beryllium-7 (the number counts the protons and neutrons in the nucleus). Fewer than one in ten thousand come from boron-8, an unstable nucleus that breaks apart. Those are among the most energetic, and they will matter later in this story [Reference 1].

The genuinely independent check is a count. Deep underground in Italy, the Borexino detector counted the collisions of proton–proton neutrinos; allowing for the known rate at which neutrinos change type (more on that later), that corresponds to about 6.1 × 10¹⁰ per square centimeter per second, give or take roughly 10%, against a predicted 6.0 × 10¹⁰ for this proton–proton part of the 6.5 × 10¹⁰ total [Reference 8]. Borexino’s measurements also confirm that about 99% of the Sun’s energy comes from the proton–proton chain [Reference 9,Reference 10]. The remaining 1% or so comes from a second route, the CNO cycle, in which carbon, nitrogen and oxygen nuclei act as helpers [Reference 11].

Two ways out of the Sun

Inside the Sun, light cannot travel far. A ray of light goes only a tiny distance before it is absorbed and sent off again in a random direction [Reference 13]. Physicists call that distance the mean free path: the average distance something travels before it hits something. For light in the Sun it is about a tenth of a millimeter near the center, and stays below a few millimeters until close to the surface [Reference 14,Reference 13].

So the light does not march outward. It wanders. This is a random walk, and random walks are slow: to get ten steps away from where you started, you need about a hundred steps; to get a hundred steps away, about ten thousand [Reference 13]. Crossing a star about seven hundred thousand kilometers in radius in steps of a millimeter or so takes a staggering number of them [Reference 7].

How long, then? It depends on what you time. Follow the light first. The answer depends on the step length: make the steps ten times shorter and you need a hundred times as many of them, each taking a tenth of the time, so the walk takes ten times longer. Estimates based on realistic models of the Sun mostly fall between about 10,000 and 170,000 years to reach the outer layer, where boiling motion, called convection, takes over [Reference 13,Reference 14]. The most quoted figure, 170,000 years, comes from a 1992 calculation that NASA’s pages use [Reference 14,Reference 5,Reference 13]; our own check suggests its formula overstates the time about sixfold, which would bring it nearer 30,000 years (details in the panel below).

No single particle of light makes the whole trip. Deep in the Sun the light is mostly X-rays, far too energetic for our eyes (our calculation from the core temperature), and it is absorbed and new light emitted over and over [Reference 15,Reference 13]. Now follow the heat instead. Think of the hot gas as a huge tank of heat and the light as a thin trickle carrying it out. By our calculation, at the center the light holds only about a thousandth as much energy as the gas. Each bit of light gets out in tens of thousands of years, but it carries so little that draining the tank takes a few hundred times longer: around ten million years, by our calculation. A published estimate of how long energy stays in the Sun, based on a rough estimate of the Sun’s stored energy, gives about 8 million years [Reference 16]. So the sunlight on your face left the Sun’s surface about 8 minutes ago, but the heat behind it left the core millions of years ago.

Neutrinos skip all that, crossing the Sun’s radius in 2.3 seconds and the gap to Earth in another 8 minutes and 17 seconds [Reference 7]. Those going through you now were made a little over eight minutes ago.

Why neutrinos are ghosts

Light is stopped by matter because it interacts with electric charge, and atoms are full of charge. A neutrino has no electric charge, so light and electricity have nothing to grab. Apart from gravity, the only force it feels is the weak interaction [Reference 4,Reference 2].

Physicists measure how likely a collision is with a number called a cross section: the effective target size one particle presents to another. For a proton–proton neutrino hitting an electron, it is about 1.2 × 10⁻⁴⁵ square centimeters [Reference 17], a number with forty-four zeros after the decimal point.

Turn that into a mean free path and the numbers get absurd. In solid lead, 11.3 grams per cubic centimeter [Reference 5], the average distance before one of these neutrinos hits an electron is hundreds of light-years: about 340 for electron-type ones, and about 500 averaged over the mix that actually reaches Earth (more on that below) [Reference 17]. A light-year, the distance light travels in a year, is nearly ten trillion kilometers. Neutrinos can also nudge whole atomic nuclei, which happens more often (counting those, the distance shrinks to several light-years), but each nudge gives a lead nucleus only a few electronvolts of energy [Reference 18,Reference 19].

The Earth, by comparison, is thin. Along a path straight through the middle of the planet, about 3 in 100 billion proton–proton neutrinos scatter, most of them by nudging iron nuclei in the core; only about 2 in a trillion bounce off an electron [Reference 20,Reference 21,Reference 17]. None is stopped: a scattered neutrino keeps going. Counting every kind of solar neutrino, fewer than one in ten billion scatter on the way through, so the planet’s rock removes essentially none of them. As we will see, there are signs that the Earth leaves one tiny fingerprint on them.

How often does one hit you?

A popular line says one neutrino might interact in your body in your whole lifetime. We found no source for it. It may mean rarer reactions that change a nucleus; for neutrinos simply bouncing off electrons, our arithmetic says otherwise.

Here is the reasoning. Borexino’s liquid gives a flash when a particle hits it, and from the pattern of flashes the experiment works out that solar neutrinos bounce off electrons about 190 times a day in every 100 tonnes [Reference 8,Reference 12]. That liquid [Reference 11] has almost the same number of electrons per gram as water, and so do you (our calculation). A 70-kilogram person is 1/1,429 of 100 tonnes, so expect about 191 ÷ 1,429 ≈ 0.13 collisions a day: about 49 a year, roughly one a week, or about 4,000 in an 80-year life. A second calculation, from the solar model’s fluxes and textbook cross sections, gives 49–51 a year [Reference 1,Reference 17].

This is our calculation, not a published result; details are in the panel. Each collision hands one electron a tiny amount of energy, almost always under 1.5 MeV [Reference 17].

The number merely passing through is enormous: if you present a target of about a tenth of a square meter with the Sun overhead, about 65 trillion solar neutrinos cross you each second; side-on, with the Sun low, a few hundred trillion. Over a lifetime: about 2 × 10²³ to 10²⁴ [Reference 1]. The Nobel committee’s “thousands of billions” every second is a safe understatement [Reference 3].

Catching ghosts

How do you detect them? With an enormous target, deep underground where rock blocks the rain of particles from space (cosmic rays), and patience [Reference 2].

The first to catch neutrinos from the Sun was Ray Davis, a chemist by training [Reference 23]. In the Homestake gold mine in South Dakota, 1,500 meters down, he filled a tank with 615 tonnes of perchloroethylene, a dry-cleaning fluid rich in chlorine [Reference 24]. Very rarely, a solar neutrino turns a chlorine-37 atom into radioactive argon-37. Solar neutrinos made about one argon atom every other day. But argon-37 itself decays, half of it within 35 days, so when Davis flushed the tank every couple of months he recovered only about 17 atoms, fished out from among some 2 × 10³⁰ chlorine-37 atoms [Reference 24,Reference 4].

Later detectors watched for light instead. When a neutrino knocks an electron hard enough, the electron can travel through water faster than light itself travels in water, which is about 75% of its speed in vacuum. It then gives off a faint cone of blue light called Cherenkov radiation, the optical cousin of a sonic boom [Reference 2]. At boron-8 energies the electron flies off within a few tens of degrees of the neutrino’s path, so the light points back toward where the neutrino came from [Reference 18]. Japan’s Kamiokande, built by a team led by Masatoshi Koshiba, used this to show that the neutrinos really come from the direction of the Sun [Reference 24]. Davis and Koshiba shared half of the 2002 Nobel Prize in Physics “for the detection of cosmic neutrinos”, which included the Sun’s [Reference 23]. Kamiokande’s successor, Super-Kamiokande, is a tank 40 meters tall and as wide, holding 50,000 tonnes of ultra-pure water watched by more than 11,000 light sensors [Reference 2].

Canada’s Sudbury Neutrino Observatory, SNO, used 1,000 tonnes of heavy water, in which each hydrogen atom carries an extra neutron, two kilometers underground [Reference 25,Reference 2]. Water detectors could only see neutrinos above a few MeV; Kamiokande’s threshold, the lowest energy it could see, was about 8 MeV [Reference 24]. So they, and largely Davis’s tank too, were counting the rare, energetic boron-8 neutrinos [Reference 4]. In Italy, Borexino used about 280 tonnes of a liquid that glows when a charged particle passes through it, which let it reach the proton–proton neutrinos, all below 0.42 MeV [Reference 11,Reference 17].

The missing two-thirds

Davis’s first result, in 1968, was already a surprise. He saw so few that he could only say the rate was no more than 3 “solar neutrino units” (SNU; think of it as a score where bigger means more neutrinos caught) [Reference 24,Reference 26]. The solar model published alongside it predicted about two and a half times as much [Reference 27]. Davis’s experiment then ran for a quarter century. Its final answer was 2.56 SNU, with an uncertainty of about 9%, roughly 30% of the 8.5 SNU that later solar models predict [Reference 28,Reference 4]. The prediction had shifted over the years as the model and nuclear measurements improved, but it never came near the measurement [Reference 27]. Kamiokande confirmed a shortfall too: it saw about half the expected number [Reference 24], rather than a third; why it saw more than Davis became clear only later (its method also picks up other neutrino types, about six times more weakly) [Reference 4].

This was the solar neutrino problem, and it lasted more than thirty years [Reference 4]. By 1978 alone, over 400 papers had proposed solutions [Reference 24]. Two explanations stood out. Perhaps the solar model was wrong and the core slightly cooler than calculated, which would make far fewer boron-8 neutrinos, the ones these detectors mostly saw. Or perhaps the neutrinos were made as predicted but changed into another type on the way [Reference 4]. The Sun makes only electron neutrinos. If some turned into muon or tau neutrinos, detectors built to catch electron neutrinos would miss them [Reference 2].

SNO was built to settle it, and heavy water was the key. It let SNO count boron-8 neutrinos two ways: one reaction happens only for electron neutrinos, another equally for all three types [Reference 29,Reference 4]. In 2002 the answer came in. The number of electron-type neutrinos arriving (the flux) was 1.76 million per square centimeter per second. The flux of all types was 5.09 million, consistent with the solar model. The non-electron part, 3.41 million (from a joint fit), was 5.3σ above zero [Reference 29].

That “σ” (sigma) is how physicists say how sure they are: the bigger the number, the less likely the result is a fluke of chance. Particle physicists usually want a significance of at least 5σ before calling something a discovery [Reference 30], and Super-Kamiokande and LZ, for example, call their 3.2σ and 4.5σ results “evidence” [Reference 31,Reference 32]. Smaller numbers are cheap: by chance alone, a bump of 2σ or more turns up about once in every 44 tries.

The neutrinos had never been missing. About two-thirds of the boron-8 electron neutrinos had simply arrived as another type [Reference 4,Reference 33]. The Sun was fine.

Neutrinos that change identity

Changing type in flight is called neutrino oscillation, and it only works if neutrinos have mass [Reference 3]. Each type of neutrino is really a mixture of slightly different masses. Each ingredient behaves like a wave, and ingredients of different mass wiggle at slightly different rates, so as the neutrino travels they drift in and out of step, a bit like two notes played together that beat louder and softer; the type you would catch swings back and forth with distance [Reference 2]. That only happens if the masses differ, so at least some of them cannot be zero. That came as a shock, because the Standard Model, physicists’ long-tested theory of the basic particles of matter, had treated neutrinos as massless [Reference 3].

Super-Kamiokande had already seen the same behavior elsewhere. Cosmic rays hitting the upper atmosphere make muon neutrinos all around the planet. Those arriving from straight above have traveled about 15 kilometers; those from below have crossed the Earth, about 13,000 kilometers. In 1998 Super-K reported that muon neutrinos coming up through the Earth were missing in a way that depended on the distance traveled: the longer the trip, the more time to change [Reference 34]. Takaaki Kajita of Super-Kamiokande and Arthur B. McDonald of SNO shared the 2015 Nobel Prize in Physics “for the discovery of neutrino oscillations, which shows that neutrinos have mass” [Reference 3].

For the Sun, there is a twist. For the energetic boron-8 neutrinos, most of the change happens inside the Sun itself, and not through collisions. The Sun is full of electrons but has no muons or taus, so electron neutrinos, and only they, feel a tiny, steady extra push from the whole crowd of electrons. Nothing collides and nothing slows down, but the push shifts the timing between the mass ingredients and changes how they mix, steering the neutrino smoothly into another type as it climbs out. The push matters more the more energetic the neutrino: by our calculation it dominates for boron-8 neutrinos in the core but is only a small correction for proton–proton ones. This is the MSW effect, after Mikheyev, Smirnov and Wolfenstein [Reference 4]. So “two-thirds missing” describes the boron-8 neutrinos. Theory predicts that low-energy neutrinos arrive as electron neutrinos about 55–60% of the time, and high-energy ones about 30% [Reference 31]. Borexino, taking the solar model’s fluxes as given, found 0.57, give or take 0.09, for proton–proton neutrinos [Reference 35,Reference 10], and SNO about 0.32 for boron-8 [Reference 33].

And the night-time fingerprint? Crossing the Earth stops none of them, but its matter can turn a few back into electron neutrinos, again without any collision [Reference 4]. Super-Kamiokande sees electron neutrinos more easily than the other types, so it counts about 3% more solar neutrinos at night than by day [Reference 36]. In its 1996–2018 data this difference stands 3.2σ from zero, which the collaboration calls evidence for the Earth’s matter effect [Reference 31]. Borexino works at lower energies, where theory predicts a negligibly small effect, and sees none [Reference 37].

How heavy are neutrinos? Not very. The KATRIN experiment has set an upper limit of 0.45 eV on the electron neutrino’s effective mass (a kind of average over its mass ingredients), less than a millionth of the electron’s mass [Reference 38]. Anything with mass travels slightly slower than light, but a typical solar neutrino trails a beam of sunlight by less than a billionth of a second over the whole trip to Earth.

Open questions

What is the Sun made of? Astronomers call every element heavier than helium a “metal”. Reading the Sun’s surface light one way gives fewer heavy elements, but then models disagree with helioseismology, the study of how the Sun vibrates [Reference 40,Reference 41]; a 2022 reanalysis found 26% more, and its authors argue this removes the clash [Reference 42]; others find it does not, once the models are made to match other solar data [Reference 43]. Neutrinos can referee, because the CNO cycle’s output depends on the carbon and nitrogen in the core [Reference 11]. Borexino’s final result favors the higher values [Reference 12]: a hint, not a verdict.

Do neutrinos from the Sun and from nuclear reactors, which also give them off, behave the same way? Almost. Solar data alone prefer a slightly smaller mass difference between two mass ingredients than reactor experiments do, a 1.5σ gap that could be chance [Reference 31]. China’s JUNO detector, running since August 2025, has already measured the reactor value more precisely than all earlier experiments combined, and agrees with them, so the gap remains (our comparison) [Reference 39].

A new way to see the Sun. Giant xenon detectors built to search for undiscovered particles also catch boron-8 neutrinos nudging xenon nuclei. After first indications in 2024 (2.73σ and 2.64σ) [Reference 44,Reference 45], LZ reported 4.5σ and XENONnT 3.3σ, both published in 2026, still short of 5σ [Reference 32,Reference 46].

Next. Hyper-Kamiokande in Japan, with a usable volume eight times Super-K’s, is under construction and planned to start operating in 2028 [Reference 47]. In our judgement, that size is what the day/night question needs.

Five Sigma, “Ghost particles: the neutrinos passing through you right now,” Five Sigma (12 October 2026), https://thefivesigma.com/articles/ghost-particles. Version of 12 October 2026.