We present a first deep colour-magnitude diagram of the putative central region ( 0.5 \arcdeg \times 0.5 \arcdeg ) of the Canis Major stellar over-density ( l,b ) = ( 240 , -8 ) found by Martin et al .
( 2004 ) , which has been proposed as the remnant of a dwarf satellite accreted onto the Milky Way on a near-equatorial orbit .
We find a narrow ( in apparent magnitude ) main-sequence , extending 6 magnitudes below the turn-off to our limiting magnitude of B \sim 24.5 mag .
This main sequence has very high constrast ( > 3 ) with respect to the thin/thick disk/halo background ; its narrowness at brighter magnitudes clearly implies the presence of a distinct and possibly still bound stellar system .
We derived the line-of-sight size ( r _ { 1 / 2 } ) of this system based on the B-band width of the lower main sequence , obtaining 0.94 \pm 0.18 ( random ) \pm 0.18 ( systematic ) kpc .
That size matches a model prediction for the main body of the parent galaxy of the Monoceros tidal stream .
The high density contrast and limited spatial extent in the radial direction are very hard to reconcile with the alternative explanation put forward to explain the Canis Major stellar-overdensity : a flared or warped Galactic disk viewed in projection ( Momamy et al .
2004 ) .
We also derived a central surface brightness of \mu _ { V, 0 } = 23.3 \pm 0.1 mag arcsec ^ { -2 } and an absolute magnitude of M _ { V } = -14.5 \pm 0.1 mag .
These values place the Canis Major object in the category of dwarf galaxy in the the L _ { V } –size and M _ { V } - \mu _ { V } planes for such objects .
However , like the Sagittarius dwarf , it is an outlier in the [ Fe / H ] - M _ { V } plane in the sense that it is too metal rich for its estimated absolute magnitude .
This suggests that the main mechanism driving its recent and current star formation history ( possibly tidal stripping ) is different to that of isolated dwarfs .