Although PG searches over the last 20 years have proved disappointing in uncovering a distant population of monolithic PGs, it is precisely this lack of success which has given rise in recent years to renewed interest in galaxy evolution at more modest redshift (z 1-3), where galaxies are detected in numerous quantities by, for example, the Hubble Space Telescope.
These studies of evolution have proved pivital in our understanding the evolution of local galaxies, not by way of uncovering PGs as such - this is one discovery which has alluded the Hubble Telescope so far, but by measuring the rate at which stars are formed in normal spiral galaxies over a look-back time of a few Gyrs and have produced a rapid advance in studies of the Universe in the era of the 10m-class telescopes.
The basic technique used to find galaxies at these redshifts involves broad-band optical imaging in a number of filters to detect the sudden drop in the spectral energy distribution shortward of the redshifted Lyman break ( 900(1 + z) Å). Photons with shorter wavelengths than this ionize hydrogen and are re-processed primarily into Lyman photons. The resultant characteristic spectral ``step'' can be seen in Figure 3 for a zero redshift galaxy. In a series of experiments starting in 1992, Steidel and collaborators (see Steidel et al. 1996) have used this technique along with follow-up deep spectroscopy on the Keck telescope, to confirm the existence of a substantial population of compact (~ 0.4 arcsec) star-forming galaxies at z ~ 3. From the strength of their emission lines and shape of their continua, these galaxies have SFRs 30 M yr-1 and a number density equivalent to between 50%-10% of the space density of present day bright galaxies. Progress in this area has been very rapid and it is now possible for the first time to sketch out the star-forming history of a substantial fraction of the present day galaxy population (Madau et al. 1996): The work of Steidel, together with other samples selected in a similar manner - including galaxies found by the Hubble Space Telescope (e.g. Lilly et al. 1996, Connolly et al. 1997), indicate that the overall SFR of galaxies increases from z = 0 to z = 2, during which time a significant fraction of the heavy elements in the Universe are formed, and then tails off towards higher redshift - see Figure 4 taken from Madau et al. (1998). Although the uncertainties are considerable and many surveys are still underway, Figure 4 is a real landmark in cosmology as it represents the very first attempt to map out the star-forming history of galaxies.
|Figure 4. The implied star-formation rate (SFR) in M yr-1 against redshift inferred from several optically selected galaxy samples. The plot is taken from Madau et al. 1998.|
The Lyman dropout galaxies are known to be young from the shape of their spectral energy distributions, which are quite flat (see Figure 3), and they also show signs of energetic outflows consistent with intense bursts of star formation (Pettini et al. 1998). So in key respects this widespread population of objects at z ~ 3 has some of the essential characteristics which we expect from PGs and it is quite possible that we are seeing directly, for the first time, the formation of present day galaxies. These results certainly demonstrate beyond any reasonable doubt that massive galaxy formation was well underway by z = 3.5. This relatively late epoch of galaxy formation appears to be in general agreement with the popular models of galaxy formation in which the mass density of the Universe is dominated by cold dark matter, discussed in section 3, which predict a peak SFR of a few M yr-1 between 2 z 4 (Baugh et al. 1998).