In this section, we describe our attempts to correct the ground-based response matrix for spatial and temporal gain variations described in the previous section (Prestwich et al., 1998). We used Bright Earth data to monitor the gain variation both as a function of position on the detector and time during the mission (see above). The Bright Earth data are dominated by scattered solar X-ray emission, primarily an atomic oxygen line at 0.525 keV, thus effectively providing a monochromatic flat-field. We have extracted the BE HRI map for 32 epochs throughout the first 5 years of the mission. Each map consists of a 32 x 32 array; 128 arcsec/pixel; with the pixel size determined by a compromise between the spatial resolution and the S/N. We fitted all 33,000 spectra with a Pearson function (Fraser 1992, file ``spec_resp_zero.tex'', available on GSFC anonymous ftp node legacy.gsfc.nasa.gov, in sub-directory ``/rosat/doc/hri'') and an additional exponential component at higher channels. This data set provides the gain appropriate for a given observation. In addition, we have written software to recover the detector position for a given source by de-applying the aspect solution. This allows the user to determine the range of gain states the source experienced as it is wobbled over the detector. This information allows a response matrix to be constructed which is weighted by the time spent in any given gain state.