6.1. Star Formation Histories of Dwarf Irregular Galaxies
Establishing the detailed star formation history of a galaxy is a very difficult task. Two-point star formation histories, like those determined from H equivalent widths (Kennicutt 1983; Kennicutt, Tamblyn, & Congdon 1994) give us clues to the question of whether the current star formation rate is comparable to the past average rate, but are unable to answer detailed questions (e.g., was there an early dominant burst of star formation). Gallagher, Hunter, & Tutukov (1984) proposed a three-point scheme based on the dynamical mass, the blue luminosity, and the H luminosity, and found that these measures were consistent with roughly constant star formation histories for the irregular galaxies in their sample. Unfortunately, the uncertainties in the conversion of these observables into constraints allows consistency with a large range of star formation histories.
Measures of resolved stars offer another avenue. Indeed, there have been many observational programs that have been very successful at modeling the recent star formation histories of dIs (e.g., Hodge 1980; Aparicio et al. 1987). However, constraining the early star formation histories of galaxies ( 1 Gyr) is a very difficult problem. As an example of what can be done from the ground, one can consider the impressive studies of Sextans B (Tosi et al. 1991) and NGC 3109 (Greggio et al. 1993). These galaxies have distance moduli of about 26.6, so with a V-band limit of roughly 23, stars with absolute magnitudes brighter than -3.5 can be reliably recorded. Their method of comparing synthetic color magnitude diagrams (CMDs) to the observations is successful in re-creating the distribution of the stars, but the comparisons are not very sensitive to the star formation histories (i.e., in the upper CMD, the models for constant star formation look very similar to models of exponentially decreasing star formation and models of two distinct bursts). Gallart et al. (1996a, b, c) have pushed this method to its limits (for ground based data) in their study of NGC 6822, and Tolstoy & Saha (1996) have developed a statistical method to determine the goodness of the fit of Monte Carlo simulations of data to observed CMDs; this allows them to quantify the errors in star formation history models (e.g., Tolstoy 1996).
In the next sections I will describe different projects which (in my opinion) represent new steps forward toward reaching the goal of detailed star formation histories. One advance comes from the fantastic imaging abilities of the Hubble Space Telescope. This has allowed great improvements in the studies of resolved stellar photometry. I will argue that relative chemical abundances can also be used to constrain recent star formation histories. When these two techniques are in agreement, I think that we have achieved "self-consistent star formation histories".