Anandan et al. reply

We agree with Bhandari [1] that our mixed state phase Φ = argTr(U_iρ0) = 0 is undefined in the special cases,
Tr(U_iρ0) = 0
However, for the example in our paper [2] that Bhandari criticizes Tr(U_iρ0) = -1 ≠ 0. In this example of interferometry with unpolarized neutrons, where one beam is given a rotation of 2π radians, our mixed state phase shift is π (modulo 2π), in agreement with the experiments. But Bhandari claims that this phase shift is ‘‘indeterminate’’ because it could be π or -π; but these two phases differ by 2π. So, the only difference between Bhandari’s viewpoint and ours is that our phase is defined modulo 2π, whereas Bhandari argues that two phases that differ by 2πn, n integer, may be distinguished experimentally in a history-dependent manner.