When the ground moves somewhere, it sends out two kinds of wave. The first kind travels faster and arrives first. The second is slower and arrives behind it. Every seismic station records both, and the gap between the two arrivals is the useful part: the further away the source, the more the fast wave has pulled ahead, so the wider the gap.
From that gap a station can work out how far away the earthquake was. What it cannot do is say where. A single station knows one thing only — the source lies somewhere on a circle drawn around the station, with that distance as the radius. Every point on the circle fits the reading equally well.
A second station, somewhere else, draws its own circle. The two circles usually cross at two points, and the earthquake is at one of them. It takes a third station, and a third circle, to say which. Where the three meet is the epicentre, and that is why seismic networks are built as networks rather than as single instruments.
The method is old, careful and quietly beautiful. Nothing is guessed; each station narrows the possibilities, and the answer is whatever survives all three. Distance alone is a circle. Two distances leave a choice. Three leave a place.
The same reasoning runs, in three dimensions, inside every satellite navigation receiver in every telephone — spheres instead of circles, and rather more of them.