We present the luminosity function and color-redshift relation of a magnitude-limited sample of 145 mostly red field E/S0 galaxies at z \lesssim 1 from the DEEP Groth Strip Survey ( GSS ) .
Using nearby galaxy images as a training set , we develop a quantitative method to classify E/S0 galaxies based on smoothness , symmetry , and bulge-to-total light ratio .
Using this method , we identify 145 E/S0s at 16.5 < I < 22 within the GSS , for which 44 spectroscopic redshifts ( z _ { spec } ) are available .
Most of the galaxies with spectroscopic redshifts ( 86 % ) form a red envelope in the redshift-color diagram , consistent with predictions of spectral synthesis models in which the dominant stellar population is formed at redshifts z \gtrsim 1.5 .
We use the tight correlation between V - I and z _ { spec } for this red subset to estimate redshifts of the remaining E/S0s to an accuracy of \sim 10 % , with the exception of a small number ( 16 % ) of blue interlopers at low redshift that are quantitatively classified as E/S0s but are not contained within the red envelope .
Constructing a luminosity function of the full sample of 145 E/S0s , we find that there is about 1.1–1.9 magnitude brightening in rest-frame B band luminosity back to z \simeq 0.8 from z = 0 , consistent with other studies .
Together with the red colors , this brightening is consistent with models in which the bulk of stars in red field E/S0s formed before z _ { for } \gtrsim 1.5 and have been evolving rather quiescently with few large starbursts since then .
Evolution in the number density of field E/S0 galaxies is harder to measure , and uncertainties in the raw counts and their ratio to local samples might amount to as much as a factor of two .
Within that uncertainty , the number density of red E/S0s to z \simeq 0.8 seems relatively static , being comparable to or perhaps moderately less than that of local E/S0s depending on the assumed cosmology .
A doubling of E/S0 number density since z = 1 can be ruled out with high confidence ( 97 % ) if \Omega _ { m } = 1 .
Taken together , our results are consistent with the hypothesis that the majority of luminous field E/S0s were already in place by z \sim 1 , that the bulk of their stars were already fairly old , and that their number density has not changed by large amounts since then .