Coordinated EISCAT, Chatanika, and Millstone Hill incoherent scatter radar observations have revealed that in the auroral zone, the nighttime F region densities vary substantially with the longitude of the observing site: EISCAT's densities are the largest and Millstone Hill's are the lowest. The nighttime F region densities measured by the individual radars are not uniform: the regions where the densities are maximum are the so-called ''blobs'' or ''patches'' that have been reported previously. The observations are consistent with the hypothesis that the nighttime densities are produced in significant amounts not by particle precipitation, but by solar EUV radiation, and that they have been transported across the polar cap. The observed differences can be explained by the offset of the geographic and geomagnetic poles. A larger portion of the magnetospheric convection pattern is sunlit when EISCAT is in the midnight sector than when Chatanika is. In winter, when Millstone Hill is in the midnight sector, almost all the auroral oval is in darkness. This universal time effect, which was observed on all coordinated three-radar experiments (September 1981 to February 1982), is illustrated using two periods of coincident radar and satellite observations: November 18--19, and December 15--16, 1981. These two periods were selected because they corresponded to relatively steady conditions. Dynamics Explorer (DE) measurements are used to aid in interpreting the radar observations. De 1 auroral images show what portion of the oval was sunlit. DE 2 data are used to measure the ion drift across the polar cap. Because the altitude of the ionization peak was high, the decay time of the F region density was substantially longer than the transit time across the polar cap. The southward meridional wind that was observed coincidentally with the ionization patches at Chatanika and EISCAT contributed to the maintenance of the F region by raising the altitude of the peak. DE 2 Langmuir probe measurements of electron density clearly showed a UT dependence, the same as that in the radar measurements.