This study describes an interpolation-based approach to the extraction of velocity measures, including peak velocities, from eye data with sampling frequencies from 62.5Hz to 1250Hz. The approach takes into account the known physical constraints of rotation through interpolation on the velocity profile under smoothness constraints and the application of classical mechanics. The result is accurate velocity profiles from all saccades, regardless of amplitude, above 31.25ms in duration at a sample rate of 62.5Hz. The method is tested by downsampling high speed data and comparing original peak velocities and main sequence with those interpolated from lower sample rates. SMI HiSpeed and SR Research EyeLink II high speed data smoothed using the Savitzky-Golay FIR smoothing filter, decimated to 62:5; 125; 250, and 625Hz, correlate with the original measured peaks; r = :95 to :99. This method represents a new approach in terms of processing eye data, emphasizing the spatial over the frequency domain. It requires good precision in the data, but reduces dependency on high sample rate for velocity measures. Results suggest that the spatial domain oers superior possibilities for the extraction of events from low frequency data, and that calculus based error metrics which can identifying physically unsound movements are more appropriate than unilateral filtering, especially in data recorded at low sample rate. Using classical mechanics to identify physically unsound movement in the data also offers tools for comparing data quality and noise characteristics at all sample rates, across various hardware, filters, individuals, calibrations or groups.