Program: GN-2006A-Q-33

The cosmic evolution of optically-selected QSOs shows a sharp decline in numbers at high redshift (z ~ 4) but it has never been clear whether this reflects a genuine decline in the number density of supermassive black holes or is an artefact of increased dust reddening in the young Universe. Here, we propose a definitive resolution of this question. We have recently shown that a combination of Spitzer 24 micron selection and good-seeing I-band data finds high-redshift Type-I QSOs ~10-times fainter than is possible with Sloan, and without any reddening bias. Gemini spectroscopy is needed to separate lightly-reddened z~3 QSOs from genuine z>4 QSOs. We need spectra of ~30 candidates from 14 square degrees (in Lockman, Elais N1 and Spitzer FLS) to determine the true space density of QSOs at z>4.