Astron. Astrophys. 358, 600-604 (2000)

4. Spectroscopy

Our spectra are from two observing runs. The discovery on 1994 May 1 UT occurred on the penultimate night of the first run; on the final night (May 2) we obtained seventeen 600 s spectra, spanning 3.2 hours. We returned to the object 1997 June 29 - July 2, and obtained 35 more exposures using the 2.4 m telescope, modular spectrograph, and a SITe CCD detector. These spectra covered 4000 - 7500 Å at 3.5 Å resolution, with considerable vignetting toward the ends. In order to minimize ambiguities in the period-finding caused by aliasing with the earth's rotation, we arranged our observations to cover a 7.1 h range of hour angle. The average spectrum from 1997 appears in Fig. 3; the 1994 spectrum is similar but at a stronger flux level. Spectral features are listed in Table 2, with their emission equivalent widths and full-widths at half maximum. The lines are single-peaked, except for a slight doubling at the top of HeI 5876; H has weak emission wings which can be traced to km s^-1 from the line center. The spectrum appears normal for a dwarf nova at minimum light. He II in particular is not detected; magnetic cataclysmics usually show this line rather prominently, so this is probably not a magnetic system.

Fig. 3. Mean spectrum obtained in 1997 June and July. The continuum level suggests , though this probably underestimates the flux

Table 2. Spectral Features

We measured radial velocities of the H line (Table 3) using convolution methods explained by Schneider & Young (1980). Four of the 35 spectra from 1997 were rejected from the analysis because of insufficient signal-to-noise. Period searches of the 1994 and 1997 data are shown in Fig. 4. Both data sets are consistent with a frequency near 7.00 d^-1; the longer time base of the 1997 data gives the more precise value. The 1997 search shows aliases at cycle d^-1 which reflect different choices of cycle count between the nights, but the Monte Carlo test of Thorstensen & Freed (1985) shows that these may be rejected with high confidence. Sinusoidal fits to the 1997 velocities of the form

yielded

where is the uncertainty of a single measurement inferred from the goodness-of-fit, and the uncertainties quoted are formal 1. While the uncertainty in P should be realistic, we caution that and (especially) K in cases where they can be checked, are often serious mis-estimates of the systemic velocity and radial velocity amplitude of the white dwarf, so they should be viewed only as fitting parameters. A fit to the 1994 data alone with the period fixed at 0.1429 d yielded km s^-1, km s^-1, and km s^-1. The improved probably reflects the somewhat brighter state of the star during the 1994 observations. The time interval between the 1994 and 1997 data sets is so long that there is no unique choice of cycle count between them, but if one assumes phase coherence, the two runs constrain the period to

is an integer. The uncertainty in N keeps P within standard deviations of the value above. Fig. 5 shows all the velocities folded on the best period,

Fig. 4. Period searches of the radial velocities from the two observing runs. These are generated by fitting a least-squares sinusoid at each trial frequency and plotting the inverse of the variance. The different appearance of the two panels reflects the different sampling; the greater height of the 1997 May peak results from the lower noise of those data

Fig. 5. Radial velocities folded on the best period. Squares are from 1994 May, and triangles are from 1997 June/July. All data are plotted twice for continuity. The period used for the fold was computed using an arbitrary choice of 8090 cycles between the observing runs. The sinusoid is fitted to the 1997 July data

Table 3. H Radial Velocities

© European Southern Observatory (ESO) 2000

Online publication: June 8, 2000
helpdesk.link@springer.de