Image motion not only shifts the zero point of the spectrum recorded by the HUT detector, but it can also lead to significant loss of light at the edges of the apertures if it is large enough. Since stable pointing was a persistent problem for Astro-1, achieving the highest photometric accuracy requires careful examination of the light lost due to the pointing jitter. We have adopted two approaches to making photometric corrections for image motion. The first relies on the availability of HUT and AST guide stars to explicitly track changes in the pointing direction. The second relies on the statistical distribution of the observed counting rate for an assumed steady source. The latter method is the only recourse for bright stars with no guide stars visible in the HUT TV camera.
The first method starts with the same basic data used to make corrections to
the wavelength scale for image motion. To derive a photometric correction,
a 
fit is done to the observed counting rate vs. offset position.
Using a Gaussian model
for the point spread function and an assumed slit profile, we fit the data to
determine the slit center, the width of the PSF, and the intensity of the
target. Only data from regions free of airglow line emission are used in
the fits. The derived intensity from this fit is used to renormalize the
flux-calibrated spectrum.
The second method assumes that the observed source is constant in intensity and should only display Poisson fluctuations in the count rate in spectral regions free of airglow lines. A first pass through the data eliminates regions with obviously low count rates when the source is clearly out of the aperture. The high end of the count rate distribution is then fit to a Poisson distribution to determine the best value for the source intensity. In the absence of satisfactory results for photometric corrections based on guide star data, this Poisson-derived count rate is used to renormalize the flux-calibrated spectrum to our best estimate of the true source flux. An obvious shortcoming of this method occurs for observations of targets with reasonably stable pointing that are poorly centered in the aperture -- the count rate may be steady and be a good match to a Poisson distribution, but it may be lower than the true flux. For this statistical method to succeed, the object must spend some time near the aperture center so that the highest count rates observed match the true source intensity. We have increased several of the aperture sizes for Astro-2 compared to those used on Astro-1 to improve the photometric accuracy of the data.