Evidence for an anisotropy of the speed of light

By comparing with the most recent experimental results, we point out the model dependence of the present bounds on the anisotropy of the speed of light. In fact, by replacing the CMB with a class of preferred frames that can better account for the experimental data, one obtains values of the RMS anisotropy parameter (1/2 -beta + delta) that are one order of magnitude larger than the presently quoted ones. The resulting non-zero anisotropy can be understood starting from the observation that the speed of light in the Earth's gravitational field is not the basic parameter c=1 entering Lorentz transformations. In this sense, light can propagate isotropically only in one `preferred' frame.