Stellar mass plays a central role in our understanding of star formation and aging. Stellar astronomy is largely based on two maps, both dependent on mass, either indirectly or directly: the Hertzprung-Russell Diagram (HRD) and the Mass-Luminosity Re
lation (MLR). The extremes of both maps, while not terra incognita, are characterized by large uncertainties. A precise HRD requires precise distance obtained by direct measurement of parallax. A precise MLR requires precise measurement of binary orbital parameters, with the ultimate goal the critical test of theoretical stellar models. Such tests require mass accuracies of ~1%. Substantial improvement in both maps requires astrometry with microsecond of arc measurement precision. Why? First, the tops of both stellar maps contain relatively rare objects, for which large populations are not found until the observing horizon reaches hundreds or thousands of parsecs. Second, the bottoms and sides of both maps contain stars, either intrinsically faint, or whose rarity guarantees great distance, hence apparent faintness. With an extensive collection of high accuracy masses that can only be provided by astrometry with microsecond of arc measurement precision, astronomers will be able to stress test theoretical models of stars at any mass and at every stage in their aging processes.
