Direct measurements of the fundamental properties of low-mass stars and brown dwarfs

Detailed theoretical models of stars, developed and observationally tested over the last century, now underlie most of modern astronomy. In contrast, models of lower temperature objects, brown dwarfs and gas giant planets, have only recently been developed and remained largely unconstrained by observations. Despite this lack of empirical validation, these models have become entrenched in many active areas of astronomical research, and thus rigorously testing them is imperative. Dynamical masses from visual binaries are central to this effort, but such measurements have been previously been impeded by observational limitations (ultracool binaries are faint, and their orbital separations are very small). This dissertation presents results from our program to test models using precise dynamical masses (as good as 2%) for ultracool binaries, based on infrared parallaxes, nearinfrared spectroscopy, and Keck laser guide star adaptive optics astrometry for a sample of over 30 objects. In just the last 2 years, we have more than tripled the number of ultracool binaries with dynamical masses, extending these measurements to much lower temperatures in previously unexplored areas of parameter space. Our main results are summarized as follows: 1. For most field binaries, based on direct measurements of their luminosities and masses, we find that the temperatures predicted by evolutionary models are discrepant with those derived from fitting the observed spectra with model atmospheres, indicating systematic errors of ~200 K in temperature (or 15%--20% in radius). We have also devised alternative model tests for these systems using their space motion and chromospheric activity to constrain their ages. 2. For the only field binary with an independent age determination from the solartype primary in its hierarchical triple system (from age--activity--rotation relations), we find that evolutionary models systematically underpredict luminosities by a factor of ~2 at a given mass, which means that model-based substellar mass determinations (e.g., for directly imaged extrasolar planets and the low-mass initial mass function) may be systematically overestimating the masses. 3. We have employed our large sample of binary orbits to carry out a novel test of the earliest evolutionary stages, by using the distribution of orbital eccentricities to distinguish between competing models of brown dwarf formation.