Program: GS-2006B-C-5

Using very deep near-infrared imaging we have recently uncovered a substantial population of galaxies at $z>2$ with red rest-frame optical colors, complementary to the well known UV-bright Lyman break galaxies. These red galaxies make up 73\% of the mass at the high mass end (M $>10^{11} \rm M_{\odot}$) at $2<z<3$. However, as they are faint in the UV, these red galaxies are normally missed in spectroscopically confirmed samples of high-redshift galaxies. To obtain a full census of the universe at $z\sim2.5$, we have started in 2005A a public near-infrared spectroscopic survey of mass-selected galaxies at $2.1<z<2.7$. Our first results are very promising and show that GNIRS is uniquely capable of obtaining an unbiased sample of massive galaxies. The 4 first spectra of our mass-selected sample, obtained in May 2005, look very different from previous samples of UV-selected samples. Emission lines like H$\alpha$ an [OIII] are very faint or absent, and the spectra are best fitted by evolved stellar population models. To test if the majority of the massive galaxies at $z\sim2.5$ are already ``red and dead'', and to elucidate the nature of massive galaxies in the young universe, we propose to extent our sample to 30 massive galaxies. Our public survey was selected as a GNIRS Key Science Program in 2005A, and we now request time to complete the project.