integrated luminosity and the inverse femtobarn

Think of luminosity as the brightness of a beam — how intensely two clouds of particles are colliding right now, this instant. But for a discovery what matters is not the brightness at one moment; it is the total amount of collision you have racked up over the whole run, the way a long-exposure photograph gathers faint starlight over hours. Integrated luminosity is exactly that running total: luminosity added up over all the time the collider has been delivering collisions. It measures, in one number, how much data you have collected.

Integrated luminosity has the strange-looking units of inverse area, almost always written as the inverse femtobarn, abbreviated fb^-1 and spoken as 'inverse femtobarn.' The point of these odd units is a beautifully simple rule: the number of times a given process happened equals its cross section (in femtobarns) multiplied by the integrated luminosity (in inverse femtobarns). So if a process has a cross section of one femtobarn and you have collected 100 inverse femtobarns of data, you expect roughly 100 of those events. The unit is built so the barns cancel and you are left with a plain count of events.

Integrated luminosity is therefore the single most important measure of an experiment's reach into rare physics: the more inverse femtobarns you accumulate, the rarer the process you can dig out of the noise. The Higgs discovery in 2012 rested on only about 10 inverse femtobarns per experiment; later runs have piled up well over a hundred. An honest caveat: 'delivered' luminosity (what the accelerator provided) and 'recorded' luminosity (what the detector actually saved, after downtime and dead time) differ, and measuring the absolute luminosity precisely is one of the largest systematic uncertainties in many cross-section results.