The combination of gravitational lensing and stellar dynamics breaks the mass-anisotropy degeneracy and provides stringent constraints on the distribution of luminous and dark matter in early-type (E/S0) galaxies out to zapproximate to1. We present new observations and models of three lens systems (CFRS 03.1077, HST 14176+5226, HST 15433+5352) and the combined results from the five field E/S0 lens galaxies at zapproximate to0.5-1.0 analyzed as part of the Lenses Structure and Dynamics (LSD) Survey. Our main results are as follows: (1) Constant mass-to-light ratio models are ruled out at greater than 99% CL for all five E/S0 galaxies, consistent with the presence of massive and extended dark matter halos. The range of projected dark matter mass fractions inside the Einstein radius is f(DM)=0.37-0.72, or 0.15-0.65 inside the effective radius R-e for isotropic models. (2) The average effective power-law slope of the total (luminous plus dark; rho(tot) proportional to r(-gamma')) mass distribution is <gamma'>=1.75+/-0.10 (1.57+/-0.16) for Osipkov-Merritt models with anisotropy radius r(i)=infinity (R-e) with an rms scatter of 0.2 (0.35), i.e., marginally flatter than isothermal (gamma'=2). The ratio between the observed central stellar velocity dispersion and that from the best-fit singular isothermal ellipsoid (SIE) lens model is <f(SIE)>=<sigma/sigma(SIE)>=0.87+/-0.04 with 0.08 rms, consistent with flatter-than-isothermal density profiles. Considering that gamma'>2 and f(SIE)>1 have been reported for other systems (i.e., B1608+656 and PG 1115+080), we conclude that there is a significant intrinsic scatter in the slope of the mass-density profile of lens galaxies (rms similar to15%), similar to what is found for local E/S0 galaxies. Hence, the isothermal approximation is not sufficiently accurate for applications that depend critically on the slope of the mass-density profile, such as the measurement of the Hubble constant from time delays. (3) The average inner power-law slope gamma of the dark matter halo is constrained to be <gamma>=1.3(-0.4)(+0.2) (68% CL), if the stellar velocity ellipsoid is isotropic (r(i)=infinity), or an upper limit of gamma<0.6, if the galaxies are radially anisotropic (r(i)=R-e). The observed range of slopes of the inner dark matter distribution is consistent with the results from numerical simulations only for an isotropic velocity ellipsoid and if baryonic collapse and star formation do not steepen dark matter density profiles. (4) The average stellar mass-to-light ratio evolves as d log (M-*/L-B)/dz=-0.72 +/- 0.10, obtained via a fundamental plane analysis. An independent analysis based on lensing and dynamics gives an average < d log (M-*/L-B)/dz >=-0.75 +/- 0.17. Both values indicate that the mass-to-light ratio evolution for our sample of field E/S0 galaxies is slightly faster than those in clusters, consistent with the hypothesis that field E/S0 galaxies experience secondary bursts (similar to 10% in mass) of star formation at z<1. These findings are consistent with pure luminosity evolution of E/S0 galaxies in the past 8 Gyr and would be hard to reconcile with scenarios involving significant structural and dynamical evolution.
