The Sagittarius dwarf galaxy (Sgr hereafter) is a most interesting object. Located at only 24 kiloparsecs from the Sun and 16 kiloparsecs from the Galactic Center (i.e. 75 000 and 50 000 light-years respectively),it is the nearest known satellite of the Milky Way. In spite of this proximity, Sgr has been discovered only recently (Ibata, Gilmore & Irwin 1994) because it was hidden to us by foreground Galactic stars. Sgr is now in process of being swallowed by our own Galaxy after complete disruption caused by Galactic tides, showing that at least part of the stellar Halo has formed from accretion of smaller constituents. However, we still lack a clear understanding of this galaxy because the high degree of contamination by foreground Galactic stars and the varying extinction make it almost impossible to get a clean sample of stars.
Figure 1. Location of the Sagittarius dwarf galaxy. The line of sight toward the northern extension of Sgr crosses the Disk and Bulge of our own Galaxy before reaching the satellite galaxy.
Fortunately, Sgr contains a fair amount of RR Lyrae stars. These variable stars have characteristic light curves and can easily be detected and separated from Galactic stars. Indeed, once their type is identified by their light curve, their absolute luminosity is derived, and the measure of their apparent luminosity gives their distance. Using two series of photographic plates, taken at La Silla (ESO) and digitized by the MAMA (operated at the CAI), Patrick Cseresnjes and his collaborators detected about 2000 RR Lyrae stars in Sgr spread over 50 square degrees. The spatial distribution of these stars allows to map the northern extension of Sgr, where the Galactic stars outnumber those of Sgr by a factor up to a thousand. Compared to other satellites of the Milky Way, Sgr seems to be much more massive and extended, even if one considers only the minor axis which is almost insensitive to Galactic tides.
Figure 2. Surface density map of the northern extension of the Sagittarius dwarf galaxy, based on RR Lyrae number counts.
Stellar evolution theory indicates that RR Lyraes are more than 10 Gigayears old. A catalogue of such stars offers therefore an unique opportunity to determine the progenitor of Sgr. The most obvious information available is the period which is very accurate and independent of crowding and extinction, allowing robust comparisons between different systems.
Figure 3 compares the period distribution of RR Lyrae stars in Sgr with those of all other dwarf galaxies with a known RR Lyrae population. The similarity with the Large Magellanic Cloud (LMC) clearly stands out. This similarity is even more striking when one considers that there are no two other couple of distributions showing such a high correlation. Statistical tests show that an identical parent distribution for Sgr and the LMC cannot be ruled out, in spite of the high resolution provided by the large size of the samples in both systems.
Figure 3. Period distribution of RR Lyrae stars in Sgr compared to all other satellites of the Milky Way. The yellow histogram is for Sgr whereas the blue shaded histogram represents the system specified in each panel.
The period of an RR Lyrae star is a complex function of its luminosity, temperature, metallicity and mass, leading to a high degeneracy between these different parameters. It is however possible to restrict the comparison to the mass-metallicity plane using RR Lyrae of a special type (type d, or RRd) which are pulsating simultaneously in the fundamental (P0) and first overtone (P1) radial modes. The position of an RRd star in the Petersen diagram (a plot of P1/P0 as a function of P0) is almost independent of its luminosity and temperature.
Figure 4. Petersen diagram for all the RRd stars detected to date. Heavy red dots represent RRd stars of the system specified in each panel. Small dots recall the position of RRd stars in Sgr.
For most systems, RRd stars are clumped in a specific region of the plot, reflecting an homogeneous population, as expected for simple systems such as globular clusters or dwarf spheroidal galaxies. However, this is not the case for Sgr which presents a large spread, showing that this system is more complex than a typical dwarf spheroidal galaxy. Again, there is a strong similarity between Sgr and the LMC. A debris of the LMC ? The similarity between Sgr and the LMC is not restricted to RR Lyrae stars, but has also been observed through other populations like Carbon stars (Whitelock 1998) or Red Giant Branch stars (Cole 2001). These similarities strongly suggest that both systems have similar stellar populations. Numerical simulations show that a dwarf spheroidal galaxy can not survive more than a couple of Gigayears on such a low orbit, unless the progenitor is given an uncomfortable high concentration, inconsistent with observations. This result is in contradiction with the presence of a substantial number of RR Lyrae stars. This contradiction could be solved if Sgr is a debris pulled out of the LMC after a collision and has been injected on its present orbit only recently. Possible configurations are a collision between the LMC and the Galaxy or the Small Magellanic Cloud. This scenario, though attractive, raises many questions which need to be addressed. When did the collision occur ? What happened to the gas ? How can the present orbital planes of Sgr and the LMC seem to be perpendicular to each other ? Future numerical simulations will assess the feasibility of this scenario.