draft 4.0/ 24 May 1998, updated 10/2015 jgl / jgl a Sci. Am. level introduction


The SuperKamiokande Collaboration is making a major statement at the 4-9 June Neutrino '98 Conference in Takayama, Japan. (The XVIII International Conference on Neutrino Astrophysics and Astrophysics, web site http://www-sk.icrr.u-tokyo.ac.jp). A paper is being submitted at the time of this release to the premier physics journal, the Physical Review Letters.

In the following we review the situation in some greater detail. Outline:


In 1998 we claimed the discovery of neutrino oscillations therefore mass, now borne out by many further results. In short, we observed a deficit of muon neutrinos coming from greater distances and at lower energies, from their production by cosmic rays high in the atmosphere to the detector buried deep underground. The behaviour of this deficit as a function of energy and arrival angle tells us that muon neutrinos oscillate, which is to say that they alternatingly change from one type of neutrino to another as they travel at close to the speed of light. We have examined many alternative explanations ranging from detector problems to alternative physics interpretations, and none come even close to fitting the results we find. However, the data behaves perfectly consistently with expectations for muon neutrinos undergoing oscillations with another type of neutrino, now shown to be the tau neutrino.

The most dramatic manifestation of this peculiar quantum mechanical phenomenon in our data is that muon neutrinos coming from the far side of the earth are depleted by about one half compared to those coming from the atmosphere overhead. Unlike some earlier (and now dismissed) claims of finding evidence for neutrino oscillations, the new observations are robust, in that the effect is large and depends only upon ratios of measured quantities. The results are consistent with all other related data.

What we found is that the muon neutrino is maximally "mixed" with another neutrino, in 1998 it was allowed to be either the tau neutrino or a new sterile neutrino. The latter possibility was eliminated over the course of further analysis and other experiments during the early 2000's, and the oscillating partner was shown to be almost entirely the tau neutrino. Neutrinos and oscillations are explained in more detail below. Oscillations depend upon a least one neutrino having mass. Measuring an oscillation tells us the difference between neutrino masses, so the number we find is only a lower limit on the actual neutrino mass. We are measuring a mass of 0.07 +/- 0.04 eV, about one ten millionth of the mass of the electron. But since there are around 50 billion neutrinos per electron, there is much more total mass in neutrinos than in electrons in the universe, and maybe even more than all normal matter.

The new understanding resolves an anomaly first uncovered in the IMB experiment in 1985, then confirmed and elaborated by Kamioka, and confirmed more recently by the Soudan experiment in Minnesota. (The latter was important, though with much smaller statistics, because it is a different type of detector). The present data base comes from the first 537 days of analyzed neutrino interactions inside the inner 22.5 kilotons of water, data taken between 1 April 1996 and 15 January 1998.


Finding nonzero neutrino mass is, of course, big news for elementary particle physics. In the "Standard Model", which is consistent with all elementary particle data so far presented, the neutrino masses are zero. The mixing is also very different from what we see elsewhere in the elementary particles (eg. the quarks and charged leptons). Hence the Standard Model needs revision. We hope that the insight gained from the peculiar mixing we find between neutrinos may open an intellectual channel for making progress towards a unified theory, one which explains the generations or flavors, and which predicts particle masses. The fact of non-zero masses is extremely important in the mathematics of the theory. What we now understand is that the masses of the building blocks of nature (the fundamental Fermions in physics jargon, the quarks and lepotons, particles with one half integer spin) are all non-zero.

We can also infer that the total mass of neutrinos in the universe must thus be significant. At a minimum the neutrinos amount to a significant fraction (10-100%) of the baryonic mass of the universe (that is ordinary matter, of which stars and people are made). On the other extreme, if the neutrino masses are of several eV total (and what we are measuring is only the small difference between two states with nearly the same mass), then neutrinos could be the dominant mass in the universe.

Neutrinos are not currently the favorite of cosmologists as solutions to "the dark matter problem" (strong astronomical evidence that the visible stars in galaxies only account for a small fraction fo the total mass of the galaxy), since in their calculations neutrinos do not cluster enough. (Fashions in cosmology have changed rapidly in the past however, so some would not give that statement as much weight as would cosmologists.) In any event, it is now the case that neutrinos cannot be neglected in the book keeping of the mass of the universe. Indeed there are some theoretical calculations which indicate that neutrinos may have played a crucial role in the production of an excess of matter over anti-matter, and are thus intimately linked to our very existence.

The balance sheet for the universe makes the crucial distinction between a universe that will expand forever into a slow cold death, or on the other extreme one which will fall back upon itself in the "big crunch", possibly to be recycled. Because neutrinos have mass they thus play an important role in the formation, structure and ultimately in the fate of the universe.

Neutrinos also play important roles in astrophysical events such as supernovae, which occurs when an old massive star collapses after running out of nuclear fuel. During the collapse the star literally becomes a neutrino star in that neutrinos totally dominate in numbers of particles for a few seconds, and carry off most of the energy from the implosion, more energy than radiated during the entire life of the star. (The neutrino flash from such an event was actually observed in Supernova 1987A, as discussed below.) Neutrinos are produced wherever violent events take place in the cosmos, and play a generally hidden but perhaps decisive role in energy transport, since with their short range force they escape easily from all but the most dense objects. The finding of oscillations and mass for neutrinos strongly conditions the nature of these effects.

Clearly the data reported here is the single most important finding about neutrinos since their discovery in 1956 by Fred Reines and Clyde Cowan, the discovery of muon neutrinos at Brookhaven National Laboratory in 1961, and the discovery of the tau particle at SLAC (Stanford California) in 1974. (The tau neutrino has not been directly detected, because we do not know how to make a tau neutrino beam, but it is inferred in the same way the electron neutrino was originally hypothesized, from the missing energy in the decay of the tau particle). According to some experts, it is also probable that the results reported in Takayama will be rated the single most important result (so far) of the decade in elementary particle physics.


Neutrinos are the least massive subatomic elementary particle in the set of building blocks of nature, the quarks and leptons. Neutrinos have no charge, and are the family of neutral leptons. They also do not feel the strong force which binds quarks into protons and neutrons, and protons and neutrons into nuclei. The energies of relevance here are about 1 GeV, giga-electron-Volts or one billion electron Volts, close to the rest-mass energy equivalent of the proton, the energy an electron would acquire after being accelerated accross a one billion Volt gap in vacuum. Neutrinos of the energies can travel through the earth with only about one in a million striking a quark (interacting via the short range "weak force").

There are three kinds, or "flavors" as they are called, of neutrinos: electron, muon and tau. There are also three anti-neutrinos of the same flavors. The neutrinos get their names from their charged lepton brethren, the electron, muon and tau, with masses 0.511 MeV, 106 MeV and 1777 MeV respectively. The proton (mass 983 MeV) is composed of up and down quarks. There are also the down and up, the strange and charmed, and bottom and top quarks, of progressively heavier masses. There has been no theory of neutrino masses, and many theoreticians have thought them to be zero. Why we are finding such small values, about 1/5,000,000 of the electron mass, is both a mystery and undoubtedly a clue.

Aside from the building blocks, all of which obey the exclusion principle, which says that two particle cannot occupy the same state simultaneously and which are called Fermions, there are particles which are perfectly content to be in the same state, called Bosons. Bosons are the carriers of the forces that hold particles together, or allow them to decay. These are the photons (which carry the electromagnetic force and interact with particles which have electric charge), the W and Z particles (which carry the weak force), the gluons (which carry the strong force), and the graviton (which carries the weakest force, gravity). Actually the "weak" force is somewhat of a misnomer, as this force is not so weak as it is short range, acting only through the very massive W and Z particles (100 proton masses). The range of a force is inversely proportional to the mass of the particle carrying the force (the uncertainty principle), and this is why electromagnetism and gravity have infinite range (the force carriers have zero mass).


The SuperKamioka detector consists of a huge 50,000 ton double layered tank of ultra pure water, observed by 11,146 twenty inch diameter photomultiplier tubes (a photomultiplier is a device which converts light, photons, into useable electrical signals). This amounts to a total of about 1 acre of photocathode, more than a factor of ten more light detection area than heretofore assembled. Located in a specially carved out cavity in an old zinc mine 2000 feet under Mount Ikena in the Japanese Alps, near the town of Kamioka, the project has been collecting data since 1 April 1996. The detector is reached by driving in a 2 km long tunnel under the mountain. Aside from the tank and photomultipliers there is a huge reverse osmosis water filtration system, a calibration electron accelerator, five trailers full of electronics, the main control room, preparation areas, and so on.

Energetic charged elementary particles travelling at close to the vacuum speed of light (300,000 km/sec), exceed the speed of light in water, which is 3/4 of that in vacuum. This results in the optical equivalent of a sonic boom, which is called Cherenkov radiation, a flash which is emitted in a 42 degree half angle cone trailing the particle. This nanosecond directional burst of blue light allows one to detect the particles with photomultipliers. The pattern, timing and intensity of light hitting the detectors surrounding the inner volume allow one to determine the particle's direction, energy and identity.

Most of the results we are discussing here are deduced from the cases (2/3 of the time) when a neutrino produces either a single electron or a single muon. The muon is the heavy brother of the electron. A short track from a neutrino interaction in the detector, typically several meters long, projects a ring of Cherenkov light onto the wall.

The data is aquired at a high rate (about 100 triggers per second) and is partially processed as it arrives, and then sent to the laboratory outside the mine via fiber optics. There data is archived and in parallel is filtered into different analysis streams for immediate processing. The single ring events of interest for the presently reported study come at a rate of about 5.5 per day.

The detector was designed to search for proton decay, measure solar neutrinos and study the atmospheric neutrinos which are the subject of this story. No proton decay has yet been found. Interesting solar neutrino results are just beginning to emerge, but that is another story.


The Collaboration consists of a team of about 100 physicists from Japan and the US. The major US collaboratoring groups are from Boston University (BU), the University of California, Irvine (UCI), the Universty of Hawaii (UH), Louisiana State University (LSU), the State University of New York at Stony Brook (SUNY-SB), and the University of Washington, Seattle (UW). Other collaborators are from Brookhaven National Laboratory, the California State University, Dominguez Hills, Los Alamos National Laboratory, the University of Maryland, and George Mason University. The lead group in Japan is at the University of Tokyo, Institute for Cosmic Ray Research, whose Director, Professor Yoji Totsuka, is the Spokesman for the Collaboration. The US team coordinators are Professors Hank Sobel of UCI (head of the old Reines neutrino group), and Jim Stone of BU. Other Japanese institutions are Gifu University, the High Energy Research Organization (KEK), Kobe University, Niigata University, Osaka University, Tohoku University, Tokai University, and the Tokyo Institute of Technology.

Many of the US Collaborators are veterans from the very successful IMB experiment, now disassembled, which was the first of this type of massive underground nucleon decay search instrument and neutrino detector. Built in the early 1980's, IMB is most famous for seeing the burst of neutrinos from Supernova 1987A (as also did the contemporanious Kamioka detector, the earlier 1/40 th scale version of the present instrument). The IMB experiment was located 2000 feet below ground in a Morton Salt mine near Cleveland Ohio, and was about 1/10 the size of the new SuperKamiokande detector. The photomultipliers from the IMB experiment were taken to Japan three years ago, and now form an outer guard layer around the SuperKamioka inner detector to tag incoming particle background.


Neutrinos are produced in many circumstances. The ones we are concerned with here are the result of cosmic rays hitting the earth's atmosphere. The primary cosmic rays make a spray of secondary particles, all travelling close to the same direction and at nearly the speed of light. Some of these secondaries (mostly pi and K mesons, and tertiary muons) decay, resulting in neutrinos. The charged particles and photons get absorbed in the atmosphere or ground. There are quite a few of these neutrinos, despite being down the decay chain from the incoming cosmic rays: about 100 or these cosmic ray induced neutrinos from the atmosphere per second, of the energies we are concerned with here are passing through you. (Yet there is only about 10% chance that one will hit a nucleon in your body in a lifetime!)

The neutrinos generally go right through the earth unscattered, but ocassionally one will interact in the SuperKamiokande detector. When the neutrino interacts it typically strikes a quark in a nucleus of oxygen (the water molecule being two hydrogens and an oxygen atom) and snatches a charge becoming either a muon or an electron (of either plus or minus charge). Depending upon energy, that charged particle travels some distance in the water. For example, a 1 GeV particle travels for about 5 meters. As the particle moves at high speed, it radiates Cherenkov light which we detect with the photmultipliers. Muons travel relatively straight and produce a rather clean ring image on the wall. Electrons are distinguished as they scatter and make fuzzier images, which can be recognized with about 98% accuracy. We catch about one atmospheric neutrino every 1.5 hours on average.


Because of the well known nature of the neutrino production, we knew that the there should have been twice as many muon neutrinos as electron neutrinos from the atmosphere. In the observations first made more than ten years ago in the IMB and Kamioka detectors, we found that the ratio of muon neutrino interactions to electron neutrino interactions was closer to one, not two. This became known as the "atmospheric neutrino anomaly". Many explananations were proposed. In the past we could not tell if the electron neutrinos were more abundant than expected (as from some unexpected though seemingly unlikely extraterrestrial source or even from nucleon decay); or if there was some problem with our calculations of the neutrino flux, or with the calculations of neutrino interaction rates; or if there was some problem unique to water detectors; or if some other hypothesis could explain the peculiar facts. People were suspicious that oscillations were the cause, but there were no compelling, exclusive arguments for oscillations. Indeed, even some of the experimenters thought that the problem would eventually be resolved as an experimental artifact.


From the new SuperKamioka data we now know that the anomaly is due specifically to a deficit in the muon neutrinos. The major breakthrough has come from benefit of the larger detector which allows us to catch higher energy muon neutrinos, and with much increased numbers. We now unambiguosly observe the zenith angle dependence of the muon neutrino flux, with a larger deficit in the up coming direction. In the previous experiments there was no significant energy or angle dependence, due to more restricted energy sensitivitity and lessor statistics.

Our analysis shows this can only be the result of the muon neutrino oscillating into another type of neutrino during long flight paths. Those neutrinos coming through the earth (13,000 km) travel further and have greater distance in which to change from a muon neutrino to another neutrino, and back again. At typical energies, some neutrinos may change back and forth many times. Those neutrinos coming from overhead (20 km) have not had time to oscillate before reaching the detector. Those neutrinos of higher energies oscillate more slowly, due to relativity. (As one may recall from high school physics class, clocks on board a passing space ship will appear to us to run more slowly as the ship's velocity approaches the speed of light). The net result is that we see muon neutrinos disppearing in proportion to their flight distance and inversely proportionally to their energy, and this is the hallmark of the previously hypothesized oscillation phenomenon.


Neutrino oscillations are a peculiar quantum mechanical effect, for which it is hard to find a good macroscopic analogy, as it has to do with the particle-wave duality of fundamental matter.

We only "know" or identify what a particle is by the way it is produced or interacts; that is how we name it. When a pion (a pi meson, composed of a quark-anti-quark pair, a particle produced abundantly in collisions of protons and nuclei) decays it results in a muon and a muon (anti)neutrino; when a neutron decays it results in a proton, an electron and an electron (anti)neutrino. When a muon is produced by a neutrino we know it was produced by a muon neutrino. And so on.

Another way to "know" or identify a particle is by weight, as expressed by speed given a certain amount of energy and also as it is attracted by gravity. This is the mass as in the famous F=ma, force equals mass times acceleration. Usually these identifications are the same for each particle: each particle with a given set of interaction signatures has a unique mass. Quarks do get a little confused about their identities, but not much. Muon neutrinos are apparently, from what we find, very mixed up.

We can make a muon neutrino beam at an accelerator for example, and, after passing the beam through a kilometer of earth and iron shielding to kill off all the charged particles, we see muons ocassionally produced in a detector, in the right direction and just after the particle beam pulse strikes the production target. Neutrinos are well known particles in this sense, and their interactions have been studied at the particle accelerators, underground and at reactors for more than 30 years.

The strange situation for neutrinos, different from all the other elementary particles, is that the state of the particle which we call the muon neutrino may not be the same as the particle mass state. Their identities are not the same: neutrinos are some sort of schizophrenic combination, a Dr.Jekyl and Mr. Hyde sort of affair. The muon neutrino is apparently composed of two different masses. Something like that never happens in macroscopic objects.

For example, the muon neutrino may be composed of half each of two states of slightly different mass. These massive neutrino states may be thought of as waves which have some specific periodicity for a given energy. The two mass states having different periodicities will oscillate in and out of phase with each other as they travel along (like the beats between to neighboring musical pitches). In one phase the pair may interact as a muon neutrino and when shifted by 90 degrees they may make a tau neutrino. In such a circumstance, if one could make a mono-energetic muon neutrino beam at an accelerator and had a moveable detector, then at first one would observe only muons being produced. Further one would see only taus. At twice the distance taus again, and so on. In between, one would see some fraction of each kind. (We find that the beat frequency of oscillation of muon neutrinos of one GeV energy is about a high D, if we could hear them!)

Experimentally we have not been able to do this at accelerators so far, because as it is turning out, the distance for oscillations has been too long to make a practical experiment (though new experiments are now being proposed based upon the information we are finding). With one GeV neutrinos one needs distances of hundreds of miles.

Another analogy which may help to understand oscillations is to compare the neutrino oscillations to the rotation of the plane of polarization of light when passing through some (so called optically active) materials. A sugar water solution has for example the property that polarized light changes the angle of the polarization as light passes through the solution. The more solution traversed, the more the rotation, and if one goes far enough, the rotation will come back to the original state. The analogy is made if we think of, say, vertially polarized light as the muon neutrino, which becomes horizontally polarized light after a while, which we call, say, the tau neutrino. The rotation of the polarization plane comes about because the light wave can be thought of as composed of right hand circular and left hand circular polarized photons, and in such media the photons of different handedness travel at slightly different speeds. In the case of the neutrinos, the neutrino waves oscillate at slightly different speeds because of their slightly different masses, and so the oscillations take place even when the neutrinos are flying through empty space.


Since all of the other building blocks (quarks and charged leptons) have mass it is natural to suppose that neutrinos would have some mass as well. Yet from their first hypothetical introduction by Pauli in 1930, in order to explain the energy missing in neutron decay, it has been clear that neutrinos had little or no mass. A peculiar principle that seems to work quite well in the elementary particle business, annunciated by Murray Gell-Mann, is that of "cosmic censorship": that which is not forbidden is required! There seemed to be no law of physics preventing neutrinos from having mass, though present models of elementary particles are notoriously without guidance about masses. Still, many take the neutrino mass to be zero, which is also seemingly allowable.

Neutrino oscillations were first hypothesized by Bruno Pontecorvo (most famous to the public for having defected from Canada to the Soviet Union) in 1957, and independently by Z. Maki, M. Nakagawa and S. Sakata from Nagoya, Japan in 1962. In fact there was no strong theoretical motivation (at least from a more fundamental model of elementary particles) for oscillations to occur, but it seemed it was not forbidden by any known rules.

The matter was largely an academic curiosity until the famous experiment of Ray Davis in the gold mine in South Dakota which revealed (and continues to reveal) that detected neutrinos from the, but at a rate much smaller than predicted. Neutrino oscillations were immediately suspected as the cause of the deficit, but it has not been proved even until the present. This 30 year old "solar neutrino problem", or deficit, as observed now in four terrestrial detectors measuring the neutrinos produced in nuclear burning in the solar core. For years people have not known whether the deficit of observed solar neutrinos was due to faulty solar models, or to faulty experiments, or to peculiar neutrino physics or to neutrino oscillations. Though oscillations of electron neutrinos are the main suspect in that mystery, that case is not closed. All of these experiments have been of the absolute rate measuring type. in which one measures a number to compare with a calculated number. There are many steps in these calculations, so people have not been sure as to the solution of the solar problem. (The SuperKamiokande results on solar neutrinos are beginning to reveal other things, such as the shape of the neutrino spectrum and temporal variations, or lack of them. The latter tests are less sensitive to complex calculations, but the results are not yet firm).

The mass of neutrinos and the possibility of their oscillations, has eluded researchers for many years. Accelerator based experiments and others using reactors and radioactive sources have so far only yielded upper limits on neutrino masses, and no oscillations have been firmly observed. Several shaky claims have come and gone over the years, and there is one peculiar contemporay observation (not in conflict with the reported results herein) from an experiment at Los Alamos, which remains to be confirmed.

Many experiments have sought to directly measure neutrino mass, which is very hard. They have only produced upper limits of a few eV in neutrino mass. Indeed cosmology tells us that the sum of the neutrino masses cannot be more than about 100 eV or the universe would have collapsed already. So we know the neutrinos are light, far less than the mass of the electron, which weighs in at 511,000 eV, but we do not know how light. Indeed many theoreticians have thought the neutrino mass would indeed prove to be zero.

It is noteworthy that the present observation is not a tiny effect. It is a nearly 50% deficit with angle variation of muon neutrinos compared to electron neutrinos. It is also an effect which has been seen in previous experiments, though with inadequate resolution to nail down the cause of the anomaly, and it is thus not a peculiarity restricted to one experiment. The results also do not need complex calculations nor comparison of observations with absolute number predictions to understand the physical source of the phenomenon. An example is the ratio of upcoming to downgoing neutrino events versus energy, for muon neutrinos and electron neutrinos. One sees clearly that the muon neutrino events have a deficit of events coming upwards, and those with higher energies, while electron neutrino events do not. Of course one needs a model to extract the neutrino mass from these data, but the qualitative effect is visible in the raw data.

In sum, until now there has been no smoking gun for neutrino oscillations and mass.


We have now examined every proposed alternate hypothesis for explanantion of the atmospheric neutrino anomaly (detector systematics, input physics, alternative physics explanations) and we rule out *all* of them, definitively. We have spent the last year carefully examining every possibility and looked for hints of problems in the data which might confound our result, and none of which have been found. *Only* the hypothesis muon neutrino oscillations fits the data, and it fits very well.

We are reporting, in a paper submitted to the Physical Review Letters, the results based upon analysis of 4700 neutrino interactions collected over a time of 537 days. We find that without oscillations the data fits the expectations at a probabilility much less than one in a million. (Actually that number is understated. The no-oscillations hypothesis is rejected at about the equivalent level of sixteen standard deviations. People usually demand at least five standard deviations for extraordinary claims, so our data greatly surpasses this requirement). With oscillations to a mass difference of 0.07 eV, the probability of getting a better fit by chance is about one half, as one expects from a correct hypothesis.

Parameters characterizing systematic effects (such as the neutrino spectrum magnitude and slope, particle identification efficiency, and so on) have been examined and all lie within expectations. We have tried fitting the data to expectations with various event selection criteria, varied data sets, fitting algorithms, and multiple independent analyses. The results are robust in that they are insensitive to any such choices. This is not a small effect, and we have found no way to make it go away or even be severely distorted.


We cannot yet resolve whether the muon neutrinos oscillate into tau neutrinos or a new sterile neutrino.

The betting would be generally on tau neutrinos, for reasons of economy, not introducing new unmotivated hypotheses (Occam's Razor). Tau neutrinos only produce tau particles at energies higher than we have much data in SuperKamiokande (because the tau is a very heavy particle, almost twice the mass of a proton). Hence, as yet, we have not yet detected the presence of taus. Taus are hard to detect in any case as the tau immediately (3 x 10^-13 sec, after travelling 0.1 mm) decays after production, and generally into many particles in a shower which looks like an electron neutrino interaction in our detector. (Indeed no experiment has yet directly detected tau neutrinos.)

By a sterile neutrino we mean one which does not even partake of the weak force, one which is only accessable to us via oscillations, possibly oscillating into a mirror world, completely out of detection from our world. There are some models however which do predict such strange new neutrino species.

One way which we shall surely be able to distinguish between taus and a new sterile neutrino is via the production of neutral pions. The normal observation of electron or muon events follows after a charge is exchanged (the W particle one of the carriers of the weak force, a fat charged photon as it were, carries a charge from a quark to the neutrino). There are also less frequentl neutral exchange events (the Z_0, a fat photon). One clear, though not frequent signal (1 in 40) is from the distinctive signature of pi-zero production. Pi-zeros decay almost immediately (8 x 10^-17 s) to two gamma rays, and these gammas make a characterisitic overlapping double ring pattern on the detector wall.

All ordinary (electron, muon and tau) neutrinos make the same rate of neutral events. So, if muon neutrinos are oscillating to taus, there will be no change in the total rate of neutral events, and thus there will be no up/down asymmetry in the pi-zero events, as there is with muons. Also the total rate of pi-zero events will agree with expectations, relative to the number of electrons.

On the other hand, if the muon neutrinos are oscillating to sterile neutrinos then there will be an up/down asymmetry and a decreased rate, since the sterile neutrinos will not make pi-zero events.

Right now we do not have enough pi-zero events to discriminate (we have about 200), but the initial look slightly favors taus. There are some hints elsewhere that slightly favor sterile neutrinos, so the case is open. We should be able to make the distinction definitively with SuperKamiokande within the next year or so.

Note that the data we are reporting tells us unambiguously that muon neutrinos are oscillating, though we cannot be sure with what other neutrino state. Yet we do know that the muon neutrinos are not oscillating into eletron neutrinos at any significant rate. In any case the conclusion about neutrino mass and the existence of oscillations does not rest upon that distinction.


The following section is of most interest to afficionadoes. If you do not want to read it, the summary is that there is no data from any other source which gainsays the results reported, and indeed there is some supportive evidence.

For the very same reason we are just acquiring these results, there is not much other information which directly confronts this data. The effects are too small to be observed in laboratory experiments (so far). Of course the previous underground nucleon decay experiments had uncovered the atmospheric neutrino anomaly, so our data is in agreement with those results. Previous data had been fitted to the oscillation hypothesis, and generally gave somewhat larger masses. The muon angular distribution we now see pulls the masses to the lower end of the range of previously allowed values. (This is a painful conclusion for the long baseline experiments proposed for accelerators, such as at Fermilab, as the lower the mass the longer the oscillation distance and the harder the experiment).

One related observation is that of the rate and angular distribution of muons produced by higher energy neutrinos (10-100 GeV) in the rock underground, and which cross the detector going upwards or sidewards. (Downwards there are cosmic ray muons made in the atmosphere, which penetrate to even the deepest mines). These measurements have been carried out in many detectors over 30 years.

One can observe the rate of muons going through ones detector, what fraction stop in the detector, and the angular dependence of these. Because the neutrino energies sampled are higher (100 GeV typically), the muon neutrinos will have oscillated less than the neutrinos which make the fully contained interactions discussed above (typically 1 GeV neutrinos), and the oscillation effects are less dramatic.

There are several difficulties with this data. First, the absolute neutrino flux predictions are imprecise to a level of perhaps 30%, so comparison with expected rates has not been fruitful. Secondly the rate is not very high, being only 1.4/day through SuperKamiokande, so we do not get much statistical power. Making things more difficult is the fact that oscillation length we find is such that the neutrino deficit sets in for the throughgoing muon sample near the horizon, just where there may be downgoing events to confuse matter.

Nonetheless, with the new SuperKamioka data exceeding the previous instruments in number, resolution and particularly in the number of muons stopping in the instrument, we can profitably search for oscillations effects. In results not ready for publication yet, we do find that the angular distribution is indeed consistent with the oscillations, and so also is the rate of stopping muons compared to throughgoing muons. The upcoming muon data itself is not as statistically compelling as the contained event data, but is perfectly consistent with it.

Our first comparisons of the stopping rate of these muons (120 stopping events) compared to through-going muons yields a stopping rate of 22%. Calculations, which are still being tuned as they are sensitive to systematics, predict a 35+/-5% with no oscillations. First results are that introducing oscillations to the predictions gives consistent results with the contained event data.

There was some analysis of data from the IMB experiment which apparently conflicts with the present results. However, we now understand that with changes in flux predictions (from updated calculations by others) and new neutrino interaction calculations (taking better account of low energy interactions), those limits disappear. Some initial (unpublished as yet) reanalysis of the old IMB muon data indeed yields results consistent with the new SuperKamiokande oscillations analysis, favoring muon neutrino oscillations with the same mass range.

The SuperKamiokande results being discussed have little impact upon the solar neutrino situation. This is because the solar oscillations (if indeed they are) involve electron type neutrinos, and due to the low energies (<1% of the energies relevant here) and long distance to the sun those oscillations inivolve much smaller mass differences.

The present results are not in direct conflict with the appearance of electron events reported from the LSND experiment at Los Alamos. It is the case that if LSND is correct and solar neutrinos oscillate, that all three results force the introduction of more than three neutrinos (that is sterile neutrinos). If the LSND results are a fluke (some undetected background, for example), then the atmospheric results plus the solar results make a simple explanation with a hierarchy of neutrino masses which resembles that seen in the other particles.


The local members of the SuperKamiokande Collaboration are graduate students Atsuko Kibayashi and Dean Takemori, and Professors John G. Learned, Shigenobu Matsuno, and Victor J. Stenger. Learned was one of the original seven people who formed the IMB project, prior to coming to Hawaii in 1980. Learned has been one of the leaders in the SuperKamiokande Collaboration in the drive to understand the data and the implications of oscillations, and is a member of the team which wrote the presently discussed paper. Matsuno has been also long involved with the IMB project, and was resident physicist in Cleveland for two years. He is leader of one of the SuperKamiokande analysis groups now. Stenger has been working on neutrino physics for many years, and has carried out many neutrino calculations. All three faculty were involved in the DUMAND Project, attempting to start neutrino astronomy under the ocean near Hawaii (that project is now being carried on in the Mediterranean). Over the years the IMB, DUMAND and SuperKamiokande projects have produced 14 PhDs from the University of Hawaii, and brought in about $8M in research funds to UH.

The UH team were the leaders in the IMB experiment in finding the neutrino burst from Supernova 1987A (23 February 1987), which was termed by many as the major high energy physics observation of the decade. This burst of neutrinos coming from the collapse of a star in the nearby small galaxy called the Large Magellanic Clouds (150,000 light years away), gave the first direct evidence that massive stars actually do end their lives in gravitational collapse to a neutron star, with the emission of a staggering pulse of neutrinos. More than 150 publications used these observations to extract many new facts about neutrinos, impossible to glean from laboratory or accelerator experiments, including significant upper limits on neutrino masses (about 20 eV).

The Hawaii group has been involved in the SuperKamiokande experiment since the US IMB group joined with SuperKamiokande in 1994. In fact this merger was initiated by Hawaii graduate Steven T. Dye, then a researcher at BU and now Associate Dean at HPU. Recent (December 1997) Hawaii graduate John Flanagan (now working at the KEK laboratory in Japan) wrote the first dissertation on SuperKamiokande using the contained neutrino interaction data discussed here, including the first neutrino oscillations analysis in a SuperKamiokande dissertation. Former Hawaii graduate student Robert Svoboda, now a Professor at LSU is also one of the SuperKamiokande analysis group leaders. On the human interest side, the project has produced not only good physics: both Flanagan and Svoboda met their new wives in the SuperKamiokande Project, and both couples were married in Hawaii last year.

An interesting historical connection to Hawaii comes from the neutrino astronomy workshop held at UH in 1976. Leaders of that Workshop included Professors Learned (then U. Wis.) and Stenger, now retired Professors Vince Peterson, Arthur Roberts (Fermilab), Hugh Bradner (UCSD), and Nobel Laureate Fred Reines (UCI). Other important people in neutrino research at that meeting were Professors Saburo Miyake (the first Director of the ICRR at U. Tokyo), Larry Sulak (then Harvard), David Cline (UCLA), and Dick Davisson (U. Washington). Fred Reines and Clyde Cowan discovered the electron neutrino in 1955. Cowan passed away, but Reines got the Nobel prize for this discovery, and other work, in 1995. Reines, now retired from UCI and in ill health, was also the leader of the IMB experiment and remains nominally part of the SuperKamiokande experiment.

There were several interesting spinoffs from that meeting, but the most fruitful was the design of a low energy neutrino detector. The results of the design principles and the tentative configurations worked out at UH in 1976 were applied to the three instruments proposed and built in the early 1980's in Cleveland, Utah and Kamioka. Hence in some sense the SuperKamiokande detector has a heritage linking back to a workshop held in Hawaii 22 years ago.

The UH elementary particle theory group has also been actively engaged in studying neutrino phenomenology, particularly as it relates to SuperKamiokande. Professor Sandip Pakvasa is one of the world's experts on neutrino phenomena and neutrino oscillations in particular. Professor Xerxes Tata and Adjunct Professor Walter Simmons also have participated in various calculations relating to neutrino phenomena relevant to the SuperKamiokande Project. There is a long standing and active interest in the study of neutrinos in the UH Department of Physics and Astronomy, which makes it an exciting and stimulating place to work, for those thrilled by neutrinos!


The research discussed herein was mostly funded by the Mombusho (Ministry of Education, Science, Sports and Culture) in Japan, who supplied most of the construction funding, and in the US the Department of Energy, Division of High Energy Physics. The project could not have been carried out without significant support from the Kamioka Mining and Smelting Company. Many other corporations and numerous individuals have made significant contributions.

Locally, we particularly acknowledge significant support from the University of Hawaii, particularly the Department of Physics and Astronomy, and the Dean of the School of Physical Sciences. Many colleagues in the Department, the Institute for Astronomy, and the School of Ocean and Earth Sciences have contributed to this work in various supportive ways over many years.

Additional Material:

Contacts : Professor John Learned, Phys.&Astron., U.H.; 808-956-2964 (UH); jlearned@hawaii.edu

Web page references: under construction http://www.phys.hawaii.edu/~superk

Graphics and Photos: Cheryl Ernst, Public Information, U. Hawaii; ernst@hawaii.edu; 808-956-5941