For six stations in the northeastern United States, the spectrum of horizontal wind speed was analyzed using 10 yr of 1-min averaged, hourly surface reports. The fast Fourier transform technique was employed to estimate the spectrum between 1 cycle/2 hr and 1 cycle/2 yr. The kinetic energy spectra show two major spikes at periods of 24 hr and 1 yr. However, most of the energy is contained in the traveling cyclones and anticyclones with periods between 2 and 7 days. The apparent discrepancy between Van der Hoven's results and our results concerning the existence of an important diurnal cycle in the kinetic energy can be explained by Blackadar's theory of the diurnal wind variation with height. Van der Hoven's spectrum represents conditions near the top of the surface layer, while our data were taken well within the surface layer. A line-by-line investigation of the diurnal peak reveals a very sharp line at 2400 hr with two side lobes 3.9 min away from the main line. These side lobes are probably caused by an annual modulation of the diurnal cycle. The spectra tentatively corrected for aliasing give some indication of the existence of a spectral gap between small-scale turbulence and mesoscale phenomena.