Normal type 1 radar echoes obtained from relatively large zenith angles have a power spectrum with a single narrow peak whose Doppler shift corresponds approximately to the acoustic velocity in the medium. On some occasions, however, this single maximum splits into two distinct peaks, separated in phase velocity on one occasion by 270 m/s. This bifurcation is most easily observed at large zenith angles during daytime when a narrow antenna beam is used. It has also been seen in a daytime experiment in which radars at Jicamarca and Huancayo simultaneously probed the same region from two different radar zenith angles. The bifurcation has been observed so far only to the west of Jicamarca, over the Pacific Ocean. This spectral splitting could be caused by vertical electron density gradients, such as those associated with ‘blanketing’ sporadic E layers. A sufficiently sharp (scale lengths of a few hundred meters or less) positive gradient on the underside of the layer and negative gradient on the topside would cause the type 1 velocity to be decreased and increased, respectively, during the day, by amounts as large as those observed.