Abstract
Errors in the measurement of the submillimeter wave-length spectra of low loss dielectrics are thought to arise from three possible sources: a "light-pipe" effect, diffraction, and interface effects. It is shown that only the last of these effects is important except when the wavelength exceeds 0.5 cm. The corrections for interfacial reflections calculated from equations presented by Pardoe are shown not to be applicable for all wavelengths.