The System Parameters of DW Ursae Majoris

We present new constraints on the system parameters of the SW Sextantis star DW Ursae Majoris, based on ultraviolet (UV) eclipse observations with the Hubble Space Telescope. Our data were obtained during a low state of the system, in which the UV light was dominated by the hot white dwarf (WD) primary. The duration of the WD eclipse allows us to set a firm lower limit on the mass ratio, q = M2/M1 > 0.24; if q < 1.5 (as expected on theoretical grounds) the inclination must satisfy i > 71°. We have also been able to determine the duration of WD ingress and egress from our data. This allows us to constrain the masses and radii of the system components and the distance between them to be 0.67 ≤ M1/M☉ ≤ 1.06, 0.008 ≤ R1/R☉ ≤ 0.014, M2/M☉ > 0.16, R2/R☉ > 0.28, and a/R☉ > 1.05. If the secondary follows Smith & Dhillon's mass-period relation for CV secondaries, our estimates for the system parameters become M1/M☉ = 0.77 ± 0.07, R1/R☉ = 0.012 ± 0.001, M2/M☉ = 0.30 ± 0.10, R2/R☉ = 0.34 ± 0.04, q = 0.39 ± 0.12, i = 82° ± 4°, and a/R☉ = 1.14 ± 0.06. We have also obtained time-resolved I- and K-band photometry of DW UMa during the same low state. Using Bessell's spectral type versus (I-K) color calibration, we estimate the spectral type of the donor star to be M3.5 ± 1.0. This latter result helps us to estimate the distance toward the system via Bailey's method as d = 930 ± 160 pc. Finally, we have repeated Knigge and coworkers' WD model atmosphere fit to the low-state UV spectrum of DW UMa in order to account for the higher surface gravity indicated by our eclipse analysis. The best-fit model with surface gravity fixed at log g = 8 has an effective temperature of Teff = 50,000 ± 1000 K. The normalization of the fit also yields a second distance estimate, d = 590 ± 100 pc. If we adopt this distance and assume that the mideclipse K-band flux is entirely due to the donor star, we obtain a second estimate for the spectral type of the secondary in DW UMa, M7 ± 2.0. After discussing potential sources of systematic errors in both methods, we conclude that the true value for the distance and spectral type will probably be in between the values obtained by the two methods.