STARS will provide an empirical framework to gain new insights into the physics of stellar interiors, and the connection between the interior and the surface. The mission will probe basic stellar structure. It will also enable us to investigate the physics of stellar interiors: the equation of state, the opacity of stellar material, energy and angular transport by convection and by waves are all areas where greater knowledge is required to build a reliable theory of stellar evolution.
Stellar models are not well constrained by observations. In most cases only the effective temperature and apparent magnitude are known; in some cases we also have values of the distance, mass, surface composition and gravity. For many clusters, it is plausible to assume that all stars have approximately the same age and chemical composition, and are at the same distance. Global properties of stars such as luminosity and radius can be fitted with different assumptions about the chemical evolution of the interior, and do not strongly constrain the internal structure and evolution. Shaviv and Salpeter (1971) demonstrated that one obtains very similar isochrones with different degrees of mixing and different ages, and there are divergences of view as to whether agreement between models and observations of the turn-off point in open clusters is improved by incorporating overshooting from the stellar convective core (Maeder and Meynet 1989; Chiosi et al. 1989; Brocato et al. 1989; Napiwotzki et al. 1991). Convective overshooting can strongly influence main-sequence lifetimes (Roxburgh 1978, 1989; Chiosi et al. 1989) and have important consequences during the dredge-up phase in red giants. Woosely and Weaver (1988) have demonstrated that overshooting is critical in determining the mass of the iron core at the moment of a supernova explosion. This is important partly because it determines the minimum possible mass of a supernova, and partly because it determines the neutrino flux which controls the form of the initial implosion. Determining the degree of overshooting is required also for establishing a better main-sequence mass-luminosity relation. This is needed for predicting the mass-luminosity relation, and ultimately the period-luminosity relation for Cepheids, thereby improving our knowledge of a crucial step in establishing the distance scale of the Universe (Chiosi et al., 1992). Even for the Sun, the mixing length used in the model of convection is treated as a free parameter, and is adjusted to obtain agreement with observation (cf. Ulrich and Cox 1991). However, there is no reason to believe that the value so derived is a universal constant. It has been assumed to be so simply for want of an alternative, and the outcome is inadequate to explain the H-R diagram of some open clusters, and leads to inconsistencies when applied to both members of some binary systems. The multiplicity of models put forward to resolve the solar neutrino problem, for example, is an indication of the limited constraints on the internal structure of the Sun posed by the surface observations of radius and luminosity (cf. Maeder 1989).
For the Sun the situation has improved with the advent of helioseismology. Since different modes of oscillation penetrate to different depths in the interior, the measured frequencies impose severe constraints on models of the present Sun, ruling out most of those that have been put forward to explain the neutrino problem (e.g. Elsworth et al. 1990b). To obtain an understanding of the physical processes in stellar interiors we need similar measurements for a wide range of stars with different properties, and in different stages of evolution; STARS will provide just such an empirical base. Measurements of oscillation frequencies will put severe constraints on the variation of sound speed with radius, especially in the central parts of a star which are of the greatest importance for stellar evolution; they will also enhance our knowledge of convection. This will enable comparison to be made with a range of theoretical models constructed on the basis of different assumptions, leading to an improvement in the theory. Even in cases of distant clusters, from which the data will be of relatively low quality compared with those from the Sun, measurements of oscillation frequencies of a number of stars in a cluster can be used to calibrate, constrain, test and improve models of evolving stars.