REVISITING BELL THEOREM FOR A CLASS OF DOWN-CONVERSION EXPERIMENTS

A certain class of parametric down-conversion Bell-type experiments ha s the following features. In the idealized perfect situation it is in only 50% of the cases that each observer receives a photon; in the oth er 50% one observer receives both photons of a pair while the other ob server receives none. The standard approach is to discard the events o f the second type. Only the remaining ones are used as the data input to some Bell inequalities. This raises justified doubts about whether such experiments could ever be genuine tests of local realism. We prop ose to take into account these ''unfavorable'' cases and to analyze th e entire pattern of polarization and localization correlations. This d eparture from the standard reasoning enables one to show that the expe riments are indeed a true test of local realism. [S1050-2947(97)50612- 3].