Following an outburst in 2001 July , the UV flux of WZ Sge has declined slowly toward pre-outburst levels .
Here , we describe new 1150-1710 Å HST /STIS spectra of WZ Sge obtained between 2002 April and 2003 March to follow the decline .
A combined analysis of these and spectra obtained in the fall of 2001 show that if log g=8.5 , then the white dwarf temperature decreased from 23,400 K in 2001 October , shortly after the steady decline of the system began , to 15,900 K 17 months later .
The UV flux in 2003 March was still 2.4 times higher than measured prior to the outburst and so the system was still recovering from the outburst .
During this period , the shape of the spectrum and the flux from the system are consistent with a log g=8.5 , 0.9 \ > M _ { \sun } white dwarf at the astrometrically-determined distance to WZ Sge .
Although the spectrum from 2001 September resembles that of a white dwarf with a temperature of 28,200 K , the implied radius is smaller than in the remainder of the observations .
If the entire white dwarf was visible in 2001 September , then either a second component , such as the disk emission , was distorting the spectrum , or more likely , the temperature on the white dwarf photosphere was not uniform then .
The metal lines in the spectra of WZ Sge have weakened with time .
Model fits that allow for material along the line of sight to the white dwarf photosphere are substantial improvements over models which assume all the lines arise from the photosphere .
This obviates the need to explain very unusual abundance ratios in the photosphere and complicates the determination of the rotation rate of the white dwarf .