Blue gravity waves from BICEP2?

We present new constraints on the spectral index n(T) of tensor fluctuations from the recent data obtained by the BICEP2 experiment. We found that the BICEP2 data alone slightly prefer a positive, "blue," spectral index with n(T) = 1.36 +/- 0.83 at 68% C. L. However, when a TT prior on the tensor amplitude coming from temperature anisotropy measurements is assumed, we get n(T) = 1.67 +/- 0.53 at 68% C. L., ruling out a scale-invariant n(T) = 0 spectrum at more than three standard deviations. These results are at odds with current bounds on the tensor spectral index coming from pulsar timing, big bang nucleosynthesis, and direct measurements from the LIGO experiment. Considering only the possibility of a "red" n(T) < 0 spectral index, we obtain the lower limit n(T) > -0.76 at 68% C. L. (n(T) > -0.09 when a TT prior is included).