In order to find evidence for any sensitivity variations with time, we used the first post-COSTAR observation of BD+28D4211 on SMS 94098 as a reference point for the 1200 Å data and divided all subsequent observations by this spectrum. Since no observations centered on 1500 Å were obtained on SMS 94098, we used the 1500 Å observations obtained on SMS 94164 for reference.
Plots of individual ratioed spectra obtained early in the cycle which include error bars were shown in GHRS ISR 69. This ISR also contains details on how the errors were calculated. The same technique was used here. Typical errors are less than +/- 1% for wavelengths above 1200 Å and around +/- 1-2% below 1200 Å.
Ratio of five sets of G140L spectra of BD+28D4211 taken on SMS 94098 (denominator) and SMSs 94164, 94213, 94294, 95014, and 95107 (numerator). The spectra were summed over wavelength intervals of 10 Å, starting at 1100 Å and continuing until 1348 Å (the last bin is slightly less than 10 Å).
Figure 1 is a plot of the five ratios obtained during Cycle 4. The baseline spectrum is the one obtained on SMS 94098, and the five subsequent spectra were divided by the baseline spectrum with binning intervals of 10 Å (for details see ISR 069). The plot confirms the trend found earlier in the cycle. The figure suggests a marginal sensitivity increase by ~1.5% between 1200 Å and 1325 Å over the cycle. The sensitivity variation is insignificant since systematic errors have the same magnitude.
In contrast, a clear drop of the sensitivity is visible below 1200 Å. The sensitivity has decreased by about 10% at 1140 Å during 1 year from SMS 94098 and SMS 95107. The trend is present in all 5 ratio spectra and is strongest in the most recent one.