= Super-Kamiokande atmospheric-neutrino experiment
{c}
{title2=1998}
Super-Kamiokande used $535$ days ($33.0$ kiloton-years) and $22.5$ fiducial kilotons of <water>. Inner <photomultiplier> times and charges reconstructed interactions; the outer <anticoincidence detector> vetoed entering <particles> and marked exits. Contained single-ring events used <likelihood> classification: <Electron> <particle showers> blur rings; <muons> give sharper edges. The $1.33\,\mathrm{GeV}$ visible-<energy> sample division is not a universal detection threshold. Simulation estimated classification efficiency above about $98\%$, distinct from total <detection efficiency>; stopping <muons> and decay <Electrons> checked classification.<Evidence for oscillation of atmospheric neutrinos by Y. Fukuda and others>[PDF pp. 2–3, Table I: apparatus and classification]{ref}
Downward <neutrinos> travel about $15\,\mathrm{km}$; upward ones about $13000\,\mathrm{km}$. The multi-GeV <muon>-like up/down asymmetry was $-0.296\pm0.048\pm0.010$; <Electron>-like events lacked a comparable effect. Calibration constrained directional bias. This was disappearance evidence; <tau neutrino> appearance was untagged and sterile conversion remained possible.<Evidence for oscillation of atmospheric neutrinos by Y. Fukuda and others>[PDF pp. 2–5, Fig. 1: zenith test and interpretation]{ref}
Writing $U$ and $D$ for the selected upward and downward populations, the asymmetry is $A=(U-D)/(U+D)$: a negative value means fewer upward events. An absolute source-normalization error alone rescales both populations and leaves this ratio unchanged. A direction-dependent response error need not cancel.
Back to article page
