Abstract

The dwarf nova U Gem is unique in having a direct measurement of the K velocity of its white dwarf from Hubble Space Telescope (HST) spectra ( from Long et al.). We present high-resolution optical spectra of U Gem in quiescence taken to test the accuracy to which the HST value can be recovered from optical data. We find that, even with data of very high signal-to-noise ratio on this archetypal system, we cannot recover Long et al.'s value to better than about 20 per cent by any method. Contamination by neighbouring emission lines seems a likely culprit. Our data reveal a number of new features: Doppler tomograms show emission at low velocity, close to the centre of mass, and a transient, narrow absorption feature is seen in the Balmer lines near the line centres at the time of eclipse. We suggest that stellar prominences, as previously invoked for the dwarf novae IP Peg and SS Cyg in outburst, may explain both of these features. The He ii 4686.75 line emission is dominated by the gas stream/disc impact region. Two distinct spots are seen in Doppler maps, the first being very narrow and showing a velocity close to that of the accretion disc in the impact region, and the second much broader and located between the velocities of the (ballistic) stream and the (Keplerian) disc. We present tentative evidence of weak spiral structure, which may support explanations for 'spiral shocks' based upon three-body effects. We find no evidence of stream-disc overflow in the system. Our data suggest an inclination angle >70°, favouring a mass for the white dwarf, in disagreement with measurements of its gravitational redshift. The mass donor is clearly seen in the Doppler maps, with emission concentrated towards its poles, and mainly on the side facing the white dwarf. This suggests irradiation with shielding by the disc from which we estimate a height-to-radius ratio (H/R) between 0.15 and 0.25. © 2006 RAS.