During the early morning of February 7, 1983, with a VHF Doppler radar interferometer we observed a discrete auroral arc that was also photographed by an all-sky camera (ASC) at Fort Churchill, Canada. We determined horizontal drift velocities and hence the vector electric fields from the interferometer data. We measured both a weak ambient and stronger arc-associated polarization field (sometimes exceeding 100 mV/m) along the poleward boundary of the arc from where the radar echoes were received. This latter field was in the proper direction (roughly parallel to the electron density gradient) to excite the gradient drift plasma instability. Both the radar and the ASC observed large (20–50 km) structures, probably caused by a Kelvin-Helmholtz instability, along this same poleward edge. The radar measured both type I and type II Doppler spectra, and in addition, we also saw a few examples of a very narrow resonant spectral peak. Because of the Doppler shift (∼60–90 Hz) of this feature and the fact that it was associated with velocity shears (which imply a horizontally divergent electric field and field-aligned currents) we believe these latter echoes were from O+ electrostatic ion cyclotron (EIC) waves generated at an altitude of about 150 km. But linear kinetic theory indicates that such waves, with lengths as short as 3 m and propagating at the observed radar aspect angle (between k and B) of about 83°, could be directly excited only by an unreasonably large field-aligned electron velocity (greater than the electron thermal speed). More realistic velocities could, however, excite 20-m EIC waves propagating at this angle, and if these steepen due to some nonlinear process, they could perhaps explain our observations.