All kinematic studies utilize data obtained from some type of measurement and data reduction system, which inherently adds noise to the spatial signal. Although this noise may not be visibly evident in the spatial trajectories, it can cause large inaccuracies when velocities and accelerations are determined by direct differentiation. Therefore, some form of data treatment, such as filtering, must be developed to decrease the noise content. The design of such filters requires knowledge of the frequency spectrum of the signal and noise.

Utilizing data obtained from a television tracking system a spectral analysis of the trajectories of 7 body markers in the saggital plane, and of the knee angle as determined by 4 of these markers, was performed on 21 runs (8 subjects) of 2 or 3 strides each. Results indicate that, for the marker trajectories, 99.7 per cent of the signal power lies below the 8th harmonic. For the knee angle, 99.7 per cent of the signal power is contained below the 6th harmonic. Suitable low pass digital filtering reduces the higher frequency noise to such an extent that meaningful velocities and accelerations can be calculated by direct digital differentiation.