GaP was one of the first III-V compounds to be studied successfully by infrared spectroscopy. One reason for this was that many impurities which were of technological importance produce strong localised vibrational modes (LVMs) in spectroscopically convenient positions in this material. This is true both for simple substitutional impurities and for more complicated centres involving intrinsic defects.1,2 A more detailed infrared study of GaP is valuable for two reasons: first that it can provide useful information about corresponding defects in other III-V compounds such as GaAs which are technologically important but less amenable to spectroscopic study and second, it provides a wide range of data for the testing of theoretical models. This is because GaP, unlike GaAs has a gap in the one-phonon density of states so that a substitutional impurity such as B on the Ga site will exhibit three modes of vibration for each of its isotopes (a localised vibrational mode, gap-mode and a low frequency resonance). To predict the frequencies of these modes correctly, together with the correct frequency separation for the isotopes, is indeed an exacting test of theory3. Unfortunately, early spectra of gap modes in boron doped GaP were made with inferior spectral resolution and signal/noise ratio 1,2,4, with the result that information was misinterpreted,4 or the correct interpretation1,2 ignored, possibly due to the lack of convincing spectra. Our aim is to remedy this situation in particular as part of a presentation of improved spectral data recorded at 15K over a wide frequency range, and to give more precise values for the frequencies of modes commonly used in the testing of theoretical models. In the course of this we have been able to search for previously unreported LVMs.