2. Previous photometric observations and some interpretations
The helium white dwarf AM CVn was discovered to be a small amplitude, short period variable star by J. Smak in 1962 (Smak 1967). For a long time it had the record as the shortest period binary system, and has been intensively observed as a test of general relativity predictions of angular momentum loss. The light curve has been described as double humped, with a period of approximately 1051 s (951.5 µHz), as well as sometimes single humped, a change also observed in some CVs (Warner & Robinson 1972). Despite heroic observational efforts, there exists no consensus on the identity of orbital period is or its possible variation with time (Solheim et al. 1984, hereinafter SRNK; Patterson et al. 1992, hereinafter PSHR).
To identify the orbital period and address concerns about the phase stability of the photometric variations, SRNK ana- lysed new and old observations spanning 1968 to 1982, finding possible periods in the range of 1051.04 to 1051.212 s (951.438-951.283 µHz). The authors point out that all periods could be related to 1051.044 s (951.435 µHz) if corrections for possible cycle count errors arising from lunar aliasing are included. Based on this idea, SRNK found a period change of s s-1, but with a considerable scatter, or "phase jitter" up to , of a nonperiodic nature, as if the period drifted for several months, and is then readjusted to the average. Later observations (Seetha et al. 1990; Emanuelsen 1990) confirmed this for observations including the 1989 observing season, giving s s-1. PSHR found the scatter to be too large, and that their observations did not fit this ephemeris.
SRNK also included analysis of the densely observed 1978 data set (Patterson et al. 1979), for the first time calculating the Fourier transform (FT) of the light curve. The authors found no power at the expected period of 1051 s (951.4 µHz), but instead at 525.6 s and 1011.4 s (1902.6 and 988.7 µHz). They proposed that the changes in the light curve from double to single humped were due to beating between these two periods on a time scale of 3.7 hrs. Additional power was also found at 350.4 s (2854 µHz). SRNK suggested that the 525.6 s (1902.6 µHz) modulation might represent a physical period present in the system, and proposed it as the rotation period of the accretor, based on the negative , which they interpreted as sign of accretion-driven spin-up.
Evidence shows that power has been present at 525.6 s (1902.6 µHz) and 1011 s (988.7 µHz) in AM CVn's FT since it was discovered to be variable. Re-analysis of the earliest observations from 1962 (Kruzewski & Semeniak 1993), as well as high speed photometry from 1976, 1978, 1982, 1987, 1988, 1990, and 1992 (Provencal et al. 1995) find the dominant power always to be at 1902.5 µHz. The amplitude of this peak is stable within the observing error, as are peaks at integer multiples of 951.3 µHz which itself is never detected. A peak at 988.7 µHz (1011.4 s) is present in all data sets but show large amplitude variations on time scales from days to months, and sometimes from night to night (Emanuelsen 1990).
The alias problem, which occurs when periodic gaps are present in the data, for example, when one combines data from successive nights at the same observatory, are serious in AM CVn's case. Although few periods are present, the strong flickering and suspected frequency variations make a short light curve look almost aperiodic (Fig. 2). In the following sections we will report WET observations for this object, and investigate which of the earlier observed periods were present, at least during the period of our observations.
© European Southern Observatory (ESO) 1998
Online publication: March 30, 1998