It is an empirical question whether photons always obey Bose-Einstein statistics, but devising and interpreting experimental tests of photon statistics can be a challenge. The nonrelativistic cross section for Compton scattering illustrates how a small admixture ν of wrong-sign statistics leads to a loss of gauge invariance; there is a large anomalous amplitude for scattering timelike photons. Nevertheless, one can interpret the observed transparency of the solar wind plasma at low frequencies as a bound ν < 10 −25 if Lorentz symmetry is required. If there is instead a universal preferred frame, the bound is ν < 10 −14 , still strong compared with previous results.