An extensive series of spectral measurements has been made in the auroral E region with the Chatanika incoherent scatter radar. Because of the small scale length of variations of electron density, temperatures, and ion-neutral collisions we used the operating mode with the best possible range resolution-9 km. About 5% of the time the data exhibited an unusual spectral shape that was most pronounced at 105 and 110 km. Instead of being almost Gaussian with only a small hint of two peaks, the spectra are much wider, with two well-developed peaks. After carefully considering the validity of the measurements and their interpretation, we conclude that the unusual spectra are due to greatly enhanced electron temperatures. At 110 km, the electron temperature may increase from 250 K to 800 K, while the ion temperature remains near 250 K. This enhancement of the electron temperature extends from 99 km to at least 116 km. We show that the temperature increase is too large to be accounted for by auroral particle precipitation, though it coincides in time with ion temperature enhancements at altitudes above 125 km. Because these latter enhancements are believed to be due to joule heating, we deduce that electric fields of 24--40 mV/m are present and that the electrons are moving through the ions and neutrals at speeds of 500--800 m/s. Despite these velocities, we find that joule heating of the electrons also cannot account for the elevated electron temperatures. Several consequences of the elevated electron temperatures are discussed. One is that the rate constants for molecular recombination are reduced. Another is that during periods of significant joule heating, the deduced electron density profile, when fully corrected for temperatures, has a significantly lower peak altitude and greater density than that deduced under the usual assumption of equal electron and ion temperatures. Since conductivities, currents, ionization rates, and differential energy spectra are dependent upon the density profile, care must be taken to account properly for the temperature effects when deriving these quantities.