At 07:35:35 UT on February 23, 1987, the Kamiokande II detector recorded 11 electron events of 7.5 to 36 MeV within 13 seconds, the first two pointing back to the Large Magellanic Cloud; IMB registered eight and Baksan five in the same window, roughly two to three hours before SN 1987A's visible light. The neutrinos escaped the collapsing core directly while the shockwave still had to fight its way to the star's surface.
A fourth detector, Mont Blanc's liquid scintillator, recorded five events four and a half hours earlier, a burst most analyses now discount as unassociated with the supernova. The anomaly lives in the literature as a reminder that in a brand-new science, even the first dataset contained a false dawn.
Koshiba's share of the 2002 Nobel recognized the detection, per the Nobel summary, of 12 of the ten-to-the-fifty-eight neutrinos the supernova emitted. The ratio is the point: an instrument the size of a house caught a dozen particles out of a number with fifty-eight zeroes, and it was enough to confirm how stars die.
The star's next-of-kin were informed by a particle that left three hours before the explosion became visible. What else has already arrived that we lack the tank to see?