1 Introduction

The light variation of semiregular variables (SR) of subtype SRa and SRb is generally associated with the pulsation of these low and intermediate mass red giants located on the asymptotic giant branch (AGB) of the Hertzsprung-Russell diagram. Since the characteristic timescale of the variations is between 20 and 2000 days (usually hundreds of days), semiregulars are typical targets of amateur observers estimating visual brightnesses. There are very few high-precision light, colour, and/or radial-velocity measurement covering many cycles that confirm the pulsational origin of the variations. Very recently, Lebzelter and his co-workers (Lebzelter 1999a,b; Lebzelter et al. 1999) presented high-precision photometric and infrared radial-velocity data for a sample of bright SR stars. Their results both illustrate the ability of automatic telescopes to monitor this type of variable stars and support the assumption of pulsation (see also Percy et al. 1996 on involving amateur photoelectric observers into regular observations of small-amplitude red variables, SARVs).

However, one cannot ad hoc accept pulsation to be the main reason of variability. As has been noted by Percy & Polano (1998), virtually every M giant is variable in brightness. This has been also supported by the large-scale photometric surveys (Hipparcos, MACHO, OGLE) leading to the discovery of thousands of M-type variables. It has been pointed out by, e.g. Lebzelter et al. (2000) and Kiss et al. (2000), that specifying the physical reason responsible for the variation requires simultaneous light, colour, and radial-velocity data, as even consecutive cycles may show significant deviations. This kind of long-term observation is crucial when studying the possible mechanisms affecting the light changes of SR variables. Besides pulsation, other contributors to variability cannot be readily excluded, such as time dependent surface inhomogeneities due to large convective cells (Schwarzschild 1975; Lebzelter et al. 2000), ellipsoidal deformation due to the presence of a close companion, or spots on a relatively rapidly-rotating star. Most recently, Koen & Laney (2000) gave a detailed list of considerations concerning the possible reasons of variability and noted, that there is still no definite conclusion on this issue. Until the application of these mechanisms is better understood, each case must be studied separately.

AI Aurigae (HD 259342, $9\hbox{$.\!\!^{\rm m}$ }1 < V < 9\hbox{$.\!\!^{\rm m}$ }8$, $P=63\hbox{$.\!\!^{\rm d}$ }9$, spectral type M5III - Chuadze 1973) is a poorly-studied red semiregular (subtype SRa) variable star. In the General Catalogue of Variable Stars (GCVS - Kholopov et al. 1985-1988) a period of 63 $.\!\!^{\rm d}$9 is listed with an early epoch of maximum (JD 2426029, due to Beyer 1937). Since then there has been no other photometric measurement published for this star. Due to the relatively small range of light changes, it is also neglected by the international organizations of amateur variable star observers (American Assocation of Variable Star Observers - AAVSO, Association Française des Observateurs d'Étoiles Variables - AFOEV and Variable Star Observers' League in Japan - VSOLJ). The AAVSO International Database does not contain this star, AFOEV has 16 points which are wrongly identified with AI Aur (they are 2 mag brighter), and VSOLJ collected only 30 individual estimates between 1986 and 1999. Similar neglect is present concerning other observations: there was only one radial velocity monitoring ( $\langle v_{\rm r} \rangle =+63$ km s^-1, Feast et al. 1972), while Dickinson & Dinger (1982) listed AI Aur among the negative detections during their H₂O survey. There is no metallicity determination in the literature. Its infrared colours (see later) place the star among the "blue'' SRs (Kerschbaum & Hron 1992) and the presently available data suggest the star to be a regular member of this group.

The main aim of this paper is to present a continuous photometric monitoring of AI Aur between 1985 and 1996 which reveals the associated colour variations in addition to obtaining an accurate light curve. (We note, that the presented observations were acquired during a large-scale photometric survey of solar type stars, Lockwood et al. 1997, without any specific reason. Therefore, this study is a by-product of that long-term observing programme). The observed behaviour of AI Aur is very likely to be mainly due to radial pulsation, most probably in fundamental or low-degree overtone mode. The next section deals with the data aquisition, while the light- and colour-curves are discussed in Sect. 3 together with the physical parameters of the star. A summary of the conclusions is given in Sect. 4.