Complementing the data gathered by looking back to high redshift is information gleaned by ``archeological'' studies of objects at z ~ 0. The oldest components of the Milky Way provide evidence that mergers of small galaxies played an important role [40]:
A ``second parameter'' - which may not be age [41] - is required to account for variations in the stellar content of globular clusters.
This second parameter is correlated with orbital direction; clusters with retrograde orbits have Oosterhoff class I variables [42].
Halo stars with [Fe/H] ~ -1 have a large range of
[
/Fe] values
[43,
44].
The outer halo exhibits retrograde rotation with respect to the rest of the galaxy [45].
The halo is not completely well-mixed, as indicated by observations of star streams and moving groups [45, 46, 47].
Items 1-3 indicate that different parts of the halo have different enrichment histories, items 2 & 4 imply that some part of the halo fell in on a retrograde orbit, and item 5 is direct evidence for the gradual dissolution of fragments after merging.
Halo accretion is clearly an ongoing process, as shown by the discovery of the Sgr I dwarf galaxy [48] and by observations of high-latitude A stars [49]. But two different arguments suggest that the bulk of the halo fell into place long ago.
First, halo stars are old. The halo as a whole shows a well-defined
turn-off at B - V ~ 0.4, corresponding to ages 
10
Gyr; only ~ 10% of the stars appear younger
[50].
To be sure, this does not rule out recent accretions of
objects containing only old stars, but most dwarf galaxies in the
local group contain intermediate-age stars as well. Thus, unless the
accreted galaxies were unlike those we observe today, most fell in
more than 10 Gyr ago.
Second, galactic disks are dynamically fragile; accretion of satellite
galaxies can easily ruin a stellar disk. Analytic estimates limit the
mass accreted by the Milky Way to less than 4% in the past 5
Gyr [51].
N-body experiments show less disk heating
than the analytic work predicts; dark halos absorb much of the damage,
and disks may tilt as well as thicken
[52,
53,
54]. Still,
accretion events of any size increase the disk's vertical dispersion,
z. Significant
structure is seen in the z-age relation; most striking is the jump from z
20 to 40 km sec-1
which marks the transition to
the ~ 10 Gyr-old thick disk
[55].
In sum, the Milky Way last suffered a significant merger at least 10 Gyr ago; relics of this event include the outer stellar halo and possibly the thick disk. Presumably, the Milky Way's dark halo was largely in place at this time, since a major merger would have disrupted even the thick disk.