The baryon fractions and mass-to-light ratios of early-type galaxies

We jointly model 22 early-type gravitational lens galaxies with stellar dynamical measurements using standard CDM halo models. The sample is inhomogeneous in both its mass distributions and the evolution of its stellar populations unless the true uncertainties are significantly larger than the reported measurement errors. In general, an individual system cannot constrain halo models, in the sense that the data poorly constrain the stellar mass fraction of the halo. An ensemble of systems, however, strongly constrains the average stellar mass represented by the visible galaxies to 0.026 +/- 0.006 of the halo mass if we neglect adiabatic compression, rising to 0.056 +/- 0.011 of the halo mass if we include adiabatic compression. Both estimates are significantly smaller than the global baryon fraction, corresponding to a star formation efficiency for early-type galaxies of 10%-30%. In the adiabatically compressed models, we find an average local B-band stellar mass-to-light ratio of (M/L)(0) = 7.2 +/- 0.5 M-circle dot/L-circle dot that evolves as d log (M/L)/dz = -0.72 +/- 0.08 per unit redshift. Adjusting the isotropy of the stellar orbits has little effect on the results. The adiabatically compressed models are strongly favored if we impose either local estimates of the mass-to-light ratios of early-type galaxies or weak-lensing measurements for the lens galaxies on 100 kpc scales as model constraints.