8 Conclusion

Besides being a by-product of the microlensing searches, the variable stars constitute their background and their understanding and discrimination are important (e.g. della Valle & Livio 1996). They need to be catalogued for any on-line microlensing searches (cf. MACHO and EROSII). In this paper, we have produced a comprehensive catalogue of 631 variable stars selected over a 0.25 deg² field in the LMC Bar with the Pixel Method applied to a 120-day window, with a sampling of about one measurement per day. For each variable star, we provide an astrometric position accurate within $1\hbox{$^{\prime\prime}$ }$, together with a finding chart. The study of their position in the CMD shows that this catalogue is dominated by a population of LT$\&$LPV, while a few "bluer'' variables have also been detected. Cross-identification with existing catalogues shows that 11% of them have already been studied elsewhere. We hence enlarge by a factor of 10 the number of LT$\&$LPV detected in this area of the LMC Bar.

The pixel analysis presented here allows a selection of variable objects independently of a photometry. It hence uses all the information present in the frames but does not provide a photometry for these objects. It is complementary to the image subtraction techniques. We have shown that not surprisingly the pixel light curves are significantly polluted by blending, especially in blue and that their conversion in magnitude can only be rough. Hence, for each variable star, we provide an indicator of blending in both colours. The preliminary study of these variable stars together with their cross-identification with previous works show that our photometry does not introduce significant bias on the overall distribution. Recent improvements of the image subtraction techniques developed for microlensing searches (Alard 1999; Alcock et al. 1999a,b) will have to be considered for the future to improve the photometry of the variable stars detected in such crowded fields (e.g. Olech et al. 1999). However, most of these objects are Long Period Variables better characterized with IR photometry than optical non-standard photometry anyway.

The limitation of this dataset is hence reached and it needs to be complemented by other data. In a companion paper, we will extend these selected light curves to the whole EROS I baseline (850 days window), with the 1992-94 data taken with a different set of filters. This will enable us to study the periodicity of these variable stars. In a forthcoming paper, we will cross-identify these variable stars with the IJK photometry of the DENIS catalogue.

Acknowledgements

We thank the EROS collaboration (http://www.lal.in2p3.fr/recherche/eros/erosa.html) for giving us their 91-92 data, and Marc Moniez for a careful reading of this manuscript. The U plate used for the astrometry was taken at the ESO Schmidt telescope at La Silla. We thank the MAMA team for scanning and reducing this plate. We acknowledge Dave Monet for the informations he kindly provided on the PMM/USNO catalogue. We are particularly grateful to C. Lamy for her useful help on data handling during this work, done in part on the computers of PCC-Collège de France. A.-L. Melchior has been supported by a European contract ERBFMBICT972375 at QMW. This research has made use of the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. The Vizier service (Ochsenbein 1997) at CDS has been used throughout this work.