An analysis is presented of low-level trajectory data obtained by means of constant level balloon flights from Cape Hatteras, N.C., during September and October 1959. An approximately constant floating level was obtained by flying the nearly constant volume Mylar balloons (tetroons) with an internal superpressure of about 100 mb. The metalized tetroons were positioned a t 1-min. intervals by means of a manually operated SP–1M radar. From knowledge of these positions, overlapping 5-min. average velocities were determined for flight durations of up to 5 hr. On some of the flights the radar return was enhanced by the addition of a radar reflective mesh to the tetroon. With the addition of this mesh, flights at altitudes of less than 5,000 ft. were tracked as far as 92 n. mi. from the radar, or approximately to the radar horizon. Spectral analysis of the velocity data obtained from the four best flights shows some evidence for a (Lagrangian) wind speed periodicity of 26-min. period, a vertical motion periodicity of 13-min. period, and a cross-stream velocity periodicity of 17-min. period. Cross spectrum analysis shows that, with the exception of oscillations of 45-min. period, the wind speed is a t a maximum ahead of the trough in the trajectory. Thus, if the large-scale air flow is nearly geostrophic, there is evidence that kinetic energy and momentum are transported down the pressure gradient by these small-scale oscillations. The maximum upward motion of the tetroon tends to take place near the trajectory trough line for oscillations of a period exceeding 18 min. and near the trajectory crest for oscillations of smaller period. Therefore, looking downstream, the longer-period tetroon oscillations tend to be counterclockwise in a plane normal to the mean trajectory while the shorter-period oscillations tend to be clockwise. However, until more information is obtained on the small-scale temperature field and the extent to which the tetroons follow the vertical air motion, any statement regarding “direct” and “indirect” air circulations is tentative. The ratio of one minus the cross-stream and one minus the along-stream autocorrelation Coefficients for these flights is approximately 0.6, suggesting a certain similarity between Eulerian space and Lagrangian autocorrelation coefficients. The tetroon data also indicate that initially one minus the cross-stream autocorrelation coefficient is proportional to the time, as would be anticipated from Lagrangian turbulence theory. These data tend to confirm that, for the scale of motion under consideration, the Lagrangian-Eulerian scale factor β of Hay and Pasquill has a value near 4.