We present and analyze Hubble Space Telescope observations of the eclipsing nova-like cataclysmic variable UX UMa obtained with the Faint Object Spectrograph .
Two eclipses each were observed with the G160L grating ( covering the ultraviolet waveband ) in August of 1994 and with the PRISM ( covering the near-ultraviolet to near-infrared ) in November of the same year .
The system was \sim 50 % brighter in November than in August , which , if due to a change in the accretion rate , indicates a fairly substantial increase in \dot { M } _ { acc } by \raisebox { -2.58 pt } { $ \stackrel { \raisebox { -0.86 pt } { $ \textstyle > $ } } { \sim } $ } 50 \% .

The eclipse light curves are qualitatively consistent with the gradual occultation of an accretion disk with a radially decreasing temperature distribution .
The light curves also exhibit asymmetries about mid-eclipse that are likely due to a bright spot at the disk edge .
Bright spot spectra have been constructed by differencing the mean spectra observed at pre- and post-eclipse orbital phases .
These difference spectra contain ultraviolet absorption lines and show the Balmer jump in emission .
This suggests that part of the bright spot may be optically thin in the continuum and vertically extended enough to veil the inner disk and/or the outflow from UX UMa in some spectral lines .

Model disk spectra constructed as ensembles of stellar atmospheres provide poor descriptions of the observed post-eclipse spectra , despite the fact that UX UMa ’ s light should be dominated by the disk at this time .
Suitably scaled single temperature model stellar atmospheres with T _ { eff } \simeq 12 , 500 - 14 , 500 K actually provide a better match to both the ultraviolet and optical post-eclipse spectra .
Evidently , great care must be taken in attempts to derive accretion rates from comparisons of disk models to observations .

One way to reconcile disk models with the observed post-eclipse spectra is to postulate the presence of a significant amount of optically thin material in the system .
Such an optically thin component might be associated with the transition region ( “ chromosphere ” ) between the disk photosphere and the fast wind from the system , whose presence has been suggested by Knigge & Drew ( 1997 ) .
In any event , the wind/chromosphere is likely to be the region in which many , if not most , of the UV lines are formed .
This is clear from the plethora of emission lines that appear in the mid-eclipse spectra , some of which appear as absorption features in spectra taken at out-of-eclipse orbital phases .