1 Introduction

In 1996, the EROS II collaboration started an observation program towards the Galactic Spiral Arms (GSA) dedicated to microlensing events. Since then, four regions of the Galactic plane located at large angles with respect to the Galactic Centre are being monitored to disentangle the disc, bar and halo contributions to the microlensing optical depth. Seven microlensing event candidates have already been published, based on three years (1996-98) of observations (Derue et al. 1999, 2001), (hereafter Papers I and II). The distance of the source stars used in these papers to compute the expected optical depths was deduced from a detailed study of our colour-magnitude diagrams. It was thus found that the source star population is located $\sim$7 kpc away, undergoing an interstellar extinction A(V) of about 3 mag (see Mansoux 1997 for more details). This distance estimate is in rough agreement with the distance to the spiral arms obtained by Georgelin et al. (1994) and Russeil et al. (1998), but its uncertainty is limiting further interpretation of our microlensing optical depth estimates. It was therefore desirable to seek more information on the distance distribution of the source stars - whether these stars belong to the disc or to the spiral arms - and on the reddening along our observation line of sights. This led us to perform a dedicated variable star search between April and June 1998, on a subset of our Galactic plane fields. The analysis was restricted to the brightest stars of this subset.

Among the wide variety of variable stars, periodic ones are of particular interest. The properties of Cepheids make them well suited to trace the Galactic spiral arms. Their reddening is measurable as well as their distance via the period-luminosity (PL) relation. RR Lyræ stars are old stars, well suited to trace the disc population. One can infer their mean dereddened magnitude and their absolute magnitude (Gould & Popowski 1998). The infrared PL(K) relation for Miras and Semi-Regular variable stars can be calibrated using a comparison of DENIS and EROS LMC giant stars (Cioni et al. 2001). Finally detached eclipsing binaries also offer the opportunity to measure their stellar parameters and their distance (Paczynski 1996).

This paper presents the results of this particular campaign that led to a catalogue containing a large number of new variable objects in the Galactic plane. Section 2 gives the basic features of the observational setup, Sect. 3 gives details on a new algorithm used to search for periodic variations of the luminosity. Section 4 describes the catalogue and the cross-identification process. In Sect. 5 we use the selected RR Lyræ to estimate the mean reddening of our fields and we give the distance distribution of these stars.