We discuss the problem of the detection of gravitational waves (GW) signals with small energy signal to noise ratio (SNR). We consider coincidence experiments between data processed by optimum filters matched to delta-like bursts. It is shown, by calculation and by simulation, that, because of the noise, the “event” lists produced by the same signals on different detectors, using the same filters, overlap only partially—about 30 percent for SNR close to the threshold used for defining the events. Furthermore, because of the noise, the correlation of the event energy between identical detectors is weak and cannot be used as a strong discriminator against noise in coincidence search, even for SNR = 10 or more.
