The need of a refined algorithm for recognizing transient events like GRBs, has risen from the difficulty in identifying them among the several events detected by the GRBM. Actually, the majority of the transient events triggering the GRBM on-board logic includes particle events, caused by the passage of high-energy charged particles through the detector units: since the mean rate of these ``false'' triggers is at least 10-20/day, the need of proper identification criteria capable of discarding them, stands out. Particularly in the case of the on-line search for GRBs, the need of automatically rejecting the false triggers, thus avoiding frequent false alert messages, is apparent.
Another motivation of this search resides in the possibility of detecting GRBs, that left their signal in the continuous data (1 s ratemeters in the two bands, 40-700 and 100 keV, respectively), without triggering the on-board logic. The set of these ``missed'' bursts can be split into different classes: first, those GRBs which, no matter about their brightness or peak flux, occur during the on-board trigger dead time of a previous event; second, the GRBs that are too faint to match the OBTCs (these form the richest class among the nontriggered GRBs). An important subset of the last class includes the GRBs coming from directions very close to the normal axis of one of the GRBM units: in this case, the probability of matching an OBTC in one of the other units, in addition to the unit facing it, is lower than it would be if the same burst would come from an off-axis direction. Since the WFC1 and WFC2 aboard BeppoSAX are co-aligned with the GRBM1 and GRBM3 (fig. ), respectively (these units also have the best sensitivity), it comes out that some non-triggered GRBs belonging to this subclass and facing one of these ``special'' units, might have a prompt X-ray counterpart in one of the WFCs; actually, this really happened in several cases, that are discussed later on.
These nontriggered bursts, for which, therefore, there are no HTR count rates, are important in any case, because of two main reasons: as regards the on-line search, they increase the number of GRBs detected within a few hours after they occurred, thus making the prompt localization feasible (when possible), or, simply, confirming the burst nature of events detected aboard other missions, like HETE-II, Ulysses, ASM/RXTE. As to the off-line search, the detection and the statistical analysis of these faint GRBs, also through a comparison with the results from other similar works on the BATSE GRB catalog, give the opportunity to study the GRBM sensitivity, the presence of possible biases and other general properties of the whole GRB sample produced by the quest.