We present EISCAT (European Incoherent Scatter Radar Facility) measurements of >2000 K enhancements of the apparent ion temperature which occurred simultaneously over a latitude range of at least 100 km for brief periods (less than 1 min) in the auroral F region. One event occurred during a substorm onset and a second during passage of a westward-traveling surge. The apparent Ti increases showed significant anisotropy, with measurements oriented less parallel to B0 exhibiting the largest amount of apparent Ti increase. In these examples the vector electric fields measured by EISCAT were much too low to account for the temperature increases via frictional heating, and also too low to generate non-Maxwellian ion velocity distributions, which can cause errors in ion temperature estimates. We argue that the measured Ti increases are not real, and that both their magnitude and their anisotropy with respect to B0 can be satisfactorily explained by turbulent plasma flows with peak amplitudes of ~2 km/s but which could not be directly resolved by EISCAT, because they varied with a time scale less than the 10-s integration period, or possibly because their scale size was smaller than the ~3--5 km antenna beam width. While such unresolved but inferred turbulent flows can themselves cause ion frictional heating, we show that an equally important cause of high Ti estimates in our case was the distortion of the measured spectra by strongly varying Doppler shifts. We also present a counter example which shows a bright auroral arc in the radar common volume but with no large increases in the radar-measured Ti, indicating that not all auroral structures are associated with electric fields which vary with sufficient intensity to distort incoherent scatter radar spectra. ¿ American Geophysical Union 1993