Photons, randomly prepared in one of two non-orthogonal quantum states
, are used for a cryptographic key distribution. If the receiver tests
them one by one, he may either identify their state, or get an inconc
lusive (useless) result. If he tests them pairwise, he may also obtain
information about their parity (whether or not they have the same sta
te), without identifying each signal separately. While this procedure
does not give a higher rate of information transmission, the parity-ge
nerated bits are more sensitive to attacks by an eavesdropper than bit
s obtained from single photons.