According to hierarchical structure formation scenarios, low-mass systems should have been the first sites of star formation in the Universe. The first star formation events may have occurred as early as at a redshift of z ~ 30, an epoch unobservable for us with our present tools. Larger systems should then have formed through hierarchical merging of smaller systems, leading to the idea of dwarf galaxies as building blocks of more massive galaxies. Hence one important test to carry out is to compare the properties of the old populations in dwarf galaxies with those in massive galaxies to investigate how similar or dissimilar they are. If today's dwarf galaxies are the few survivors of a once much more numerous, since accreted low-mass galaxy population, and if dwarf galaxy accretion is the primary process governing the formation of more massive galaxies, then detailed studies of their old stellar populations should reveal very similar properties.
A number of cosmological models predict that cosmic re-ionization will squelch star formation in low-mass substructures, and that galaxies less massive than 108 to 109 M will lose their star-forming material through photoevaporation during re-ionization (e.g., Ferrara & Tolstoy 2000; Dekel & Woo 2003; Susa & Umemura 2004). As a consequence, low-mass galaxies must form their stars prior to re-ionization (e.g., Tassis et al. 2003) and need to contain ancient populations. Furthermore, one would expect a sharp drop and indeed a complete cessation of star-forming activity after re-ionization is complete. A third consequence is that the oldest populations in high-mass galaxies must be either as old as those in low-mass galaxies, or younger.
Figure 2. Bar diagram indicating the approximate duration of star formation episodes in low-mass galaxies (~ 107 M). The approximate beginning and end of the re-ionization epoch are indicated (based on results from WMAP and from the Sloan Digital Sky Survey). The predicted cessation of star formation in low-mass galaxies, esp. in dSphs, is not observed. For more details, see also Grebel & Gallagher (2004).
These are testable predictions that can be investigated by exploiting the fossil stellar record in nearby galaxies. The LG is an ideal target since here the oldest populations are resolved and can be accessed with HST and increasingly also with large ground-based telescopes operating with high angular resolution. This requires age dating of old populations. The most accurate ages can be obtained for resolved stellar populations. For old populations the most age-sensitive feature is the old main-sequence turn-off (MSTO), where via differential age-dating techniques internal accuracies of less than 1 Gyr can be obtained. Comparison objects are usually ancient Galactic globular clusters of the same metallicity as the target population. Absolute ages are more difficult to determine since here one needs to rely on isochrones, which makes the resulting ages model-dependent. But also differential techniques have a number of drawbacks: They require very deep, high-quality photometry reaching at least 2 mag below the MSTO, which is a challenge in more distant galaxies. They require stellar populations sufficiently numerous to produce a measurable MSTO, which necessarily limits us to Population II stars (note that to date not a single Population III star candidate has ever been detected beyond the Milky Way). Ideally, one wishes to compare populations with the same [/Fe] ratio. (Establishing to what extent this condition is met and what the consequences are will be an important area for large ground-based telescopes in the coming years.) Other effects such as diffusion may further affect relative measurements; see Chaboyer's contribution in these proceedings for details.
For a more detailed discussion of the results of differential age dating as applied to ancient Milky Way populations and nearby dwarfs I refer to Grebel (2000) and Grebel & Gallagher (2004). Here only the results shall be summarized, which are based on Galactic and extragalactic globular clusters and field populations with MSTO photometry: (1) Old populations are ubiquitous, but their fractions vary. There is not a single dwarf galaxy studied in sufficient detail without an old population. (2) There is evidence for a common epoch of star formation. The ancient Population II in the Milky Way, the LMC, and the Galactic dSphs are coeval within ~ 1 Gyr (which is consistent with the building block scenario). (3) In contrast to predictions from cold dark matter models, no cessation of star formation activity is observed during or after re-ionization. (4) Instead, even the least massive galaxies known, the dSphs, show evidence for star formation extending over many Gyr. This holds even for those dSphs entirely dominated by very old populations - their metallicity spread requires enrichment due to star formation extending over several Gyr.
It is certainly correct to point out that the accuracy of age determinations does not permit us yet to confidently state when exactly early star formation began and ended in dSphs. Nevertheless, the presence of ancient populations in dSphs is a fact. The large metallicity spread in old dSphs has been proven spectroscopically. Furthermore, the presence of intermediate-age populations in addition to ancient stars in many though not all dSphs cannot be refuted. In only one of these dSphs (Carina) clearly episodic star formation with well-separated "bursts" has been observed (Smecker-Hane et al. 1994); in all others star formation appears to have proceeded fairly continuously. The characteristics of the star formation and enrichment histories of dSphs contradict the cosmological predictions mentioned earlier, particularly the suggestion of complete photoevaporation of baryonic material not yet turned into stars.
There are several possible ways out of this quandary. Either the above quoted cosmological models are wrong, or do not take other effects that might prevent photoevaporation into account, or the galaxies we observe today as dSphs once were substantially more massive. In case if the latter, today's dSphs would once have needed to be at least a factor of 100 more massive. An oversimplified estimate shows that at the times when dSphs were forming stars, they should have had at least 10 times more baryonic mass if we assume a star formation efficiency of 10%. Their present-day baryonic masses are of the order of 106 M, so an increase by one order of magnitude would not yet change the overall estimated masses of dSphs drastically, but it would considerably alter their mass-to-light ratios. More detailed calculations and more observational data are needed to investigate whether substantially more massive dSph progenitors are plausible. The differences in the metallicity-luminosity relation of dSphs and dIrrs seem to rule out dIrrs as progenitors of dSphs (Grebel et al. 2003).