The method consisted in the use of the "double x-ray spectrometer" for measuring the extremely small changes in angle produced by refraction in prisms, which were placed between two crystals. The rocking curves from the second crystal were from 6 to 10 seconds of arc wide at half maximum and, under favorable temperature conditions, a shift of.2 seconds could be detected. Two different wave-lengths of x-rays and prisms of various materials were used. The values of $\ensuremath{\delta}\ifmmode\times\else\texttimes\fi{}{10}^{6}$, where $\ensuremath{\delta}=1\ensuremath{-}\ensuremath{\mu}$, are as follows: These results agree with the Lorentz dispersion formula within experimental error. In the case of certain materials, notably graphite, the curves from the second crystal were extremely broad (20 to 50 seconds), caused by internal refraction due to the granular structure. This seems of interest with respect to the possibility of comparing internal arrangements of particles in various samples, though nothing of that sort was attempted here.