We employ a flexible Bayesian technique to estimate the black hole mass and Eddington ratio functions for Type 1 ( i.e. , broad line ) quasars from a uniformly-selected data set of \sim 58 , 000 quasars from the SDSS DR7 .
We find that the SDSS becomes significantly incomplete at M _ { BH } \lesssim 3 \times 10 ^ { 8 } M _ { \odot } or L / L _ { Edd } \lesssim 0.07 , and that the number densities of Type 1 quasars continue to increase down to these limits .
Both the mass and Eddington ratio functions show evidence of downsizing , with the most massive and highest Eddington ratio black holes experiencing Type 1 quasar phases first , although the Eddington ratio number densities are flat at z < 2 .
We estimate the maximum Eddington ratio of Type 1 quasars in the observable Universe to be L / L _ { Edd } \sim 3 .
Consistent with our results in Paper I , we do not find statistical evidence for a so-called “ sub-Eddington boundary ” in the mass-luminosity plane of broad line quasars , and demonstrate that such an apparent boundary in the observed distribution can be caused by selection effect and errors in virial BH mass estimates .
Based on the typical Eddington ratio in a given mass bin , we estimate growth times for the black holes in Type 1 quasars and find that they are comparable to or longer than the age of the universe , implying an earlier phase of accelerated ( i.e. , with higher Eddington ratios ) and possibly obscured growth .
The large masses probed by our sample imply that most of our black holes reside in what are locally early type galaxies , and we interpret our results within the context of models of self-regulated black hole growth .