The measurement of perceptually relevant information about textures has been approached through profilometry, vibrometry, and tribometry. Manfredi et al.  used a laser Doppler vibrometer to measure skin surface vibrations as a texture sample slides across a fingertip. In our work, we treat the Manfredi et al. measurements as a gold standard, and assess the performance of a simpler and more portable device: A high-bandwidth tribometer. The tribometer was used to measure shear and normal forces applied to each of six texture samples as a fingertip scanned across them. The collected data was used to build two classifiers: One based on features extracted from the spectra (which treats the data as stationary); and a second based on the first through fourth order statistics associated with a set of band-pass filters (which treats the data as nonstationary). The results indicate that tribometry, while not as effective as vibrometry, may nonetheless prove effective as a means of recording natural texture. Additionally, we find that the non-stationarity of skin vibrations may serve as means of texture classification. Ongoing work aims to couple tribometric recordings with texture rendering and playback via surface haptic devices, and to understand the perceptual significance of non-stationarity in vibrations.