The U.S. Geological Survey developed equipment to measure stream velocity simultaneously with 10 current meters arranged in a vertical and to measure velocity closer to the streambed than attainable with conventional equipment. With the 10 current meters, synchronous velocities were recorded for a period of 66 minutes at 10 different depths in one vertical of one gaging-station cross section. In addition, with a current meter installed on a special bracket to allow measurements to 0.5 foot above streambed, data were obtained at two to four verticals in four gaging-station cross sections. The mean velocity determined for the 66-minute period of record was 3.30 fps (feet per second). The graphic record of velocity was analyzed on a minute-by-minute basis. It was noted that the shape of the vertical velocity curves (plot of horizontal flow velocities measured in a vertical) changed from one minute to the next, but the change seemed to be random. Velocities obtained at different depths in the, profile fluctuated significantly, with the 1-minute velocities obtained at 0.05 depth (5 percent of total depths measured from the surface at indicated vertical) showing the smallest range--0.66 fps--and those at 0.55 depth the largest range--l.22 fps. The standard deviation, expressed in feet per second, of the velocity at each point in the vertical tended to increase with depth--from 0.16 fps at 0.05 depth to a maximum of 0.24 fps at 0.75 depth. The standard deviation, expressed as a percentage of the mean velocity, ranged from about 4 percent near the surface to 11 percent at 0.95 depth. In spite of the fluctuation in mean velocity that occurred during the 66 minutes and observation period of 4 minutes yields a mean velocity that differs from the 66-minute mean by less than one-half of a percent. Determining the mean velocity by averaging the 10-point observations of the 66minute run proved to be as accurate as by plotting the vertical velocity curvy (from the averaged 10 points) and then integrating the depth-velocity profile. In comparing the velocity obtained by integrating the depth-velocity profile with the 10-point mean velocity for other field data, collected beyond that obtained during the 66-minute run, the difference ranged from -1.3 to +1.7 percent and averaged -0.2 percent. Extension of the curve below the 0.95 depth by use of a power function proved to be fairly accurate (when compared with actual measurements within this reach made with the special current-meter bracket). However, the extension did not improve significantly the accuracy of the integrated-curve mean velocity. Both the one- and two-point methods were found to agree with the 10-point velocity. In computing mean river velocity, values determined by the two-point method ranged from -1.4 to +1.6 percent when compared with the base integrated-curve mean river velocity. The one-point method yielded results that ranged from -1.9 to +4.4 percent and averaged 40.1 percent. In determining river flow by use of the midsection and mean-section methods, the mean-section method uniformly yields lower flows for the same dart.. The range in difference is from -0.2 percent to -1.6 percent, with an average difference of -0.6 percent.