Observations of turbulent energy dissipation, ϵ, measured during a week-long mixed-layer study on the continental shelf off Nova Scotia are presented. This time series of dissipation measurements at a fixed site and with a wide range of wind speeds indicates that a constant fraction of the energy flux in the atmospheric boundary layer appears as dissipation in the mixed layer. Our measured velocity-shear spectra are consistent in shape with an isotropic-turbulence spectral form and simultaneous determinations of spectral level from two mutually perpendicular sensors are consistent with isotropy. Significant changes in turbulence levels between two profiles a few minutes apart are observed. These changes (often a factor of 10) emphasize the necessity of adequate space-time averaging to obtain good mean values of ϵ. Including data from measurements of the large-scale density and velocity fields, the generation of the observed turbulence is thought to be Richardson-number instabilities in the mixed ... Abstract Observations of turbulent energy dissipation, ϵ, measured during a week-long mixed-layer study on the continental shelf off Nova Scotia are presented. This time series of dissipation measurements at a fixed site and with a wide range of wind speeds indicates that a constant fraction of the energy flux in the atmospheric boundary layer appears as dissipation in the mixed layer. Our measured velocity-shear spectra are consistent in shape with an isotropic-turbulence spectral form and simultaneous determinations of spectral level from two mutually perpendicular sensors are consistent with isotropy. Significant changes in turbulence levels between two profiles a few minutes apart are observed. These changes (often a factor of 10) emphasize the necessity of adequate space-time averaging to obtain good mean values of ϵ. Including data from measurements of the large-scale density and velocity fields, the generation of the observed turbulence is thought to be Richardson-number instabilities in the mixed ...