Program: GS-2007A-C-9

In the first spectroscopic survey of its kind, we have recently obtained Gemini/GNIRS spectra of 20 $K-$selected galaxies at $z\sim 2.5$. Surprisingly, we found that the spectra of 9 of the 20 galaxies are characterized by strong Balmer/4000\,\AA\ breaks and show no detected H$\alpha$ emission. This demonstrates that star formation in a large fraction of the most massive galaxies was already greatly suppressed when the Universe was only 20\,\% of its present age. It is not clear whether star formation in these galaxies terminated entirely or continues at a much reduced level. The limited S/N of the available spectra, combined with the redshift uncertainty, gives $2\sigma$ upper limits on the H$\alpha$ equivalent width of $6-10$\,\AA, corresponding to values of the Scalo stellar birth rate parameter $b \lesssim 0.1$. Here we propose to obtain very deep follow-up spectra of two of the candidate ``dead'' galaxies. These deep spectra will give redshifts with $<1$\,\% accuracy, and provide much higher S/N in the region where H$\alpha$ is expected. The combination of the smaller redshift uncertainty and higher S/N will yield either direct detections of the elusive H$\alpha$ line, or constrain $b$ to $\lesssim 0.02$. In either case, the proposed observations will provide important new constraints on the mechanism(s) responsible for terminating star formation in the most massive galaxies.