Classical novae comprise 35% of all cataclysmic variables (CVs). However, their orbital period distribution is the least well known of all subgroups of CVs. This can be explained by an observational selection effect: While in eruption the light of such systems is emitted by an extended pseudo-photosphere which prevents the direct observation of the binary system, during quiescence most of them are too faint for time-resolved radial velocity measurements. Photometric observations can also be difficult because novae are concentrated in the galactic bulge and thus are usually found in very crowded fields. While some efforts have been made to understand the observed period distribution of the other types of CVs such as polars (e.g. Hameury et al. 1988) and dwarf novae (Shafter 1992), attempts to calculate the intrinsic nova period distribution via population synthesis have been performed by Kolb (1995), but no comparisons with observations are available as yet.
There are at least three selection effects which influence the observed distribution of nova periods (see Fig. 1 for a schematic representation). The first (S-1) is just related to the possibility to recognize in principle the existence of the object. In other words, it is related to the eruption of a system in the potential novae population independent of the fact whether it is actually observed or not. This "intrinsic" selection effect defines the sample of classical novae. It renders the number of observed systems proportional to the eruption frequency of the prenova binaries and is therefore closely related to the physics of the nova outburst. The second selection effect (S0) is a "visibility" function that takes into account the magnitude limit of the observed sample. The last effect (S1) is associated to the probability of detection of a particular orbital period in the sample of known galactic novae. We call this the technical selection function. When time-resolved photometric observations are used to measure periods this effect favours the detection of large modulations, short periods, and high inclinations. Such methods may be specially effective on novae which are observed during the decline to the quiescent state, when the illumination of the secondary by the still very hot and luminous primary may produce large continuum modulations (Patterson 1979). Of course, eclipses also play a prominent role. It is important to point out that more than 75% of the sample of novae with known orbital periods show orbital modulations or eclipses of the primary component. Many periods have been discovered in recent years in time-resolved CCD observations of faint remnants, emphazising the high potential of this technique. On the other hand, radial velocity measurements are more effective in revealing periods in systems where the inclinations are high (which also favours eclipses!), the period is short (but not too short, see Sect. 5), and the white dwarf mass is low. But such measurements have been less productive in finding periods and are considerably more expensive, requiring larger telescopes than photometric measurements. Moreover, the large majority of the known novae are too faint in quiescence to permit time-resolved spectroscopy even with the most powerful existing instruments.
Thus, if is the observed and the intrinsic period distribution ("parent" distribution), we may write:
In this study we aim at the quantification of the selection effects in order to derive the intrinsic distribution of the nova progenitors from the observed period distribution. This may then directly be compared to evolutionary predictions from population synthesis studies (e.g. de Kool 1992; Politano & Webbink 1990, and Kolb 1993).
© European Southern Observatory (ESO) 1997
Online publication: June 5, 1998