A sample of RR Lyrae ( RRL ) variables from the Southern Edgeworth-Kuiper Belt Object survey in regions overlapping the expected position of debris from the interaction of the Sagittarius ( Sgr ) dwarf galaxy with the Milky Way ( RA \sim 20 and 21.5 h ; distance = 16–21 kpc ) has been followed up spectroscopically and photometrically .
The 21 photometrically confirmed type ab RRLs in this region have \langle [ Fe/H ] \rangle = -1.79 \pm 0.08 on our system , consistent with the abundances found for RRLs in a different portion of the Sgr tidal debris stream .
The distribution of velocities in the Galactic standard of rest frame ( V _ { \textrm { \scriptsize { GSR } } } ) of the 26 RRLs in the region is not consistent with a smooth halo population .
Upon comparison with the Sgr disruption models of Law et al .
( 23 ) , a prominent group of five stars having highly negative radial velocities ( V _ { \textrm { \scriptsize { GSR } } } \sim - 175 km s ^ { -1 } ) is consistent with predictions for old trailing debris when the Galactic halo potential is modeled as oblate .
In contrast , the prolate model does not predict any significant number of Sgr stars at the locations of the observed sample .
The observations also require that the recent trailing debris stream has a broader spread perpendicular to the Sgr plane than predicted by the models .
We have also investigated the possible association of the Virgo Stellar Stream ( VSS ) with Sgr debris by comparing radial velocities for RRLs in the region with the same models , finding similarities in the velocity-position trends .
As suggested by our earlier work , the stars in the VSS region with large negative V _ { \textrm { \scriptsize { GSR } } } values are likely to be old leading Sgr debris , but we find that while old trailing Sgr debris may well make a contribution at positive V _ { \textrm { \scriptsize { GSR } } } values , it is unlikely to fully account for the VSS feature .
Overall we find that further modeling is needed , as trailing arm data generally favors oblate models while leading arm data favors prolate models , with no single potential fitting all the observed data .