There is an isotropically distributed population of UV-excess galaxies. Their number density is high: about 300,000 per square degree. There appears to be a leveling off of the N(m) counts in the blue and U wavelength bands, giving most of the contribution to the extragalactic background light at magnitudes between 24 and 26 U mag. The nature of the faint blue galaxies remains unclear, although there is evidence that the visible mix of galaxy type or population changes with redshift, and the FBGs are distributed broadly in redshift. Their high surface number density and low correlation implies a maximal redshift distribution extending up to 3. Not many FBGs can have redshifts above 3, due to the lack of Lyman continuum absorbed FBGs in ultra-deep U band imaging. Their average angular size is maximal and their correlations are minimal by comparison to theories.
Because of their large volume, models with open cosmologies can more easily accommodate the count excess at 26 mag. Models with flat cosmologies must adjust one or more of the other available parameters: LF, SFR scenarios, merging, cosmological constant, etc. Galaxy number non-conservation is not required, even in flat cosmologies. Multiple populations of galaxies and a changing type or luminosity mix as a function of redshift is perhaps the most natural explanation of all the observations taken together. Among the single population models of FBGs those with low spatial correlation and open cosmology, or vacuum dominated flat models are in best agreement with the data. It is very possible that there was a population of fading galaxies at moderately high redshift, and future studies to ultra faint flux levels at wavelengths from radio to x-ray would be required for their characterization (see Treyer & Silk 1993). If the FBGs have faded below detection at z = 0 - 0.1 in photographic surveys, it is puzzling why they are not seen in the large volume now covered by 4 magnitude deeper CCD surveys at z = 0.2 - 0.5.
The optical EBL from summing over the FBGs is close to that required for production of all metallicity, with little room left for an added diffuse EBL. Taken together with the correlation data, the blue color of this EBL and the absence of a comparable density of faint red galaxies at K band implies that we are seeing galaxy formation over a range of redshifts from roughly 1 to 3. Thus, FBGs will be valuable for statistical cosmology; because of their large angular sizes and redshifts these FBGs are a useful tool in statistical gravitational lens studies of foreground mass distributions.
I would like to acknowledge my collaborators in this FBG research over the past decade: Gary Bernstein, Pat Boeshaar, Carol Christian, Raja Guhathakurta, Craig Gullixson, Neal Hartsough, John Jarvis, Greg Kochanski, Steve Majewski, Phil Fischer, Pat Seitzer, Frank Valdes, Ethan Vishniac, Pat Waddell, and Rick Wenk. The staff and telescope operators at CFHT, CTIO and KPNO have been very helpful, sometimes beyond the call of duty. Bill Baum, Art Hoag, and the staff at Lowell Observatory greatly aided the calibration of the BjRI filter system. I have also had helpful discussions recently with Richard Ellis, Mike Fitchett, Jim Gunn, Richard Kron, Carol Lonsdale, Adrian Melott, Jordi Miralda-Escudé, Gus Oemler, Marc Postman, Chris Pritchet, and Simon White.