Fig. (2) Zero-bias data sample at s =1960 GeV with an exponential fit: (left) efficiency of event selection (probability that the whole detector is empty) vs beam-bunch instantaneous luminosity for a single bunch (the L quoted in the text is 36 times larger, as there are 36 colliding bunches); (right) detector-noise levels for “interaction” and “no-intezoomraction” events. The vertical dashed lines show the cuts used to define “empty” detectors or “noise” (the K^+K-K background in this area, measured with K0K0 events, amounts to a few %).