Several years ago I was exploring the venerable timing argument as applied to the Local Group of galaxies (including the small satellites: `Observatory', 101, 111, 1981); the Local Group is the collection of two or three dozen centered on the Milky Way and Andromeda. Dissatisfied with the plot I had, I asked Dr. Mike Irwin about the completeness and sensitivity of searches for Local Group dwarfs. There was no answer because apparently there hadn't been any dedicated searches. So Mike, George Hau and I started one.
Bright, easy-to-find galaxies have all been found and catalogued, and any of those belonging to the Local Group are well known; so any remaining ones must be faint. Since we're looking from inside the Group, they could be anywhere on the sky. That means we needed a full-sky survey to the faintest available limit, which in turn meant the ESO/SRC photographic survey of the southern sky and the complementary second Palomar Observatory Sky Survey (POSS-II) in the north. And since we would be searching for faint signals among lots of noise, neither easy to characterize mathematically, we decided to search by eye. (The human eye is very good at extracting information from complicated scenes, a very useful skill in the primeval jungle, though sometimes the information extracted isn't actually there.)
I scanned all 894 fields in the south and the same number in the north, using glass copies of the original plates (which have lower noise at the faintest levels than film copies). Looking for faint objects of reasonable size (one arc minute or above), which had not already been determined to be Local Group galaxies (or not), we ended up with a list of almost 300. Each of these was observed with either the CTIO 1.5m or the KPNO 2.1m to check for resolution into stars; if it happened to be within a megaparsec we should have seen the Red Giant Branch. Most of the time we didn't. Sometimes we did, and two candidates, Antlia and Cetus, are in fact previously undetermined Local Group dwarf galaxies. (Someone had suggested that Antlia was nearby, based on a radio observation of its redshift, in a paper a few years before. But no one had followed it up, and it found its way onto our list independently. We also looked at about 20 objects which others had suggested were Local Group dwarfs; none is.)
The original purpose of the Dwarf Hunt was partly fulfilled: we found two new galaxies. Ironically, for various reasons neither has been useful to fill out the timing-argument plot. But the fact that we only found two, even though we were looking hard, is significant. Simulations of galaxy formation tend to show hundreds of satellites around things like the Milky Way, not dozens; now we can say with certainty that they aren't there.
Our summary of the Local Group Dwarf Hunt appeared in the Astronomical Journal. The hard part was figuring out just how sensitive our search was, and how reliable. We estimate that, of the sky not covered by the Milky Way, our search was at least 77% complete; and since others have also looked for similar objects, there can be no more than one or two lurking where they could be seen. Something like a dozen may be hidden. And we could reliably see objects to 25 magnitudes per square arc second in R, probably 25.5, possibly 26; which is one percent of the night sky brightness, or less.
Here are a few of the objects we found along the way. (Click on the pictures to get a higher-resolution version.)