2005MNRAS.361..525D

We present deep Keck spectroscopy, using the Deep Imaging Multi-Object Spectrograph and the Low-Resolution Imaging Spectrometer spectrographs, of a large and representative sample of 67 extremely red objects (EROs) to H= 20.5 in three fields (SSA22, Chandra Deep Field South and NTT Deep Field) drawn from the Las Campanas Infrared Survey (LCIRS). Using the colour cut (I-H) > 3.0 (Vega magnitudes) adopted in earlier papers in this series, we verify the efficiency of this selection for locating and studying distant old sources. Spectroscopic redshifts are determined for 44 sources, of which only two are contaminating low-mass stars. When allowance is made for incompleteness, the spectroscopic redshift distribution closely matches that predicted earlier on the basis of photometric data. Our spectra are of sufficient quality that we can address the important question of the nature and homogeneity of the z > 0.8 ERO population. A dominant old stellar population is inferred for 75 per cent of our spectroscopic sample, a higher fraction than that seen in smaller, less complete samples with broader photometric selection criteria (e.g. R-K). However, only 28 per cent have spectra with no evidence of recent star formation activity, such as would be expected for a strictly passively evolving population. More than ∼30 per cent of our absorption-line spectra are of the `E+A' type with prominent Balmer absorption consistent, on average, with mass growth of 5-15 per cent in the past gigayear. We use our spectroscopic redshifts to improve earlier estimates of the spatial clustering of this population as well as to understand the significant field-to-field variation. Our spectroscopy enables us to pinpoint a filamentary structure at z= 1.22 in the Chandra Deep Field South. Overall, our study suggests that the bulk of the ERO population is an established population of clustered massive galaxies undergoing intermittent activity consistent with continued growth over the redshift interval 0.8 < z < 1.6.