Observations made during the activation period were used to determine
the primary mirror position giving the best focus for both the
spectrograph and acquisition camera. The camera optics had been adjusted
during instrument assembly to make these two mirror positions coincide, but
on-orbit they were found to differ by 200 microns.
The cause of this change is not understood; it is presumably
due to shifting of some component during launch.
For most observations the mirror was set halfway between
these two positions, which gave camera images of 
FWHM. For some observations with faint guide stars, however,
the mirror was set at the position giving the best camera focus
(
FWHM). The spectrograph line spread function
is determined primarily by the fact that the flat detector surface does not
coincide with the curved Rowland circle. Changing the focus position of the
primary mirror shifts the point at which these two surfaces intersect.
The compromise mirror position resulted in minima in the line spread
function at 900 and
1600 Å, instead of the intended locations of 1050 Å and 1350 Å.
Measurements of the in-flight point-source spectral resolution utilized unresolved emission lines in stellar spectra. The initial determination of spectral resolution as a function of mirror position was done with the early observation of Capella; more detailed measurements were subsequently made with spectra of the symbiotic stars RR Tel, AG Dra, V1016 Cyg and Z And, obtained as part of the Guest Investigator program of B. Espey. The spectrum of RR Tel provided the bulk of the line measurements due to its high signal-to-noise and the presence of lines across the entire HUT bandpass ([Espey 1995]). Since gaussians were found to provide a good fit to single emission lines, gaussian deblending using the SPECFIT software ([Kriss 1994]) was used to obtain the location and linewidth of blended features. The relative strength of lines observed in IUE high resolution spectra were used where necessary to help in the deblending process. The resulting resolution, illustrated in Figure 1, was found to be 2 Å FWHM at 900 Å and 1600 Å with a broad maximum of about 4 Å FWHM over 1100--1250 Å.