Hunting oxymorons

We recall an example of cutting-edge research using obsolete means.

Our astronomer was recently reminded of a research project he conducted some years ago.  He started it as a graduate student, when one is just shaping one’s research career.  Of course one then wants to use the latest, biggest instruments, fastest computers, and newest techniques.  Partly this is to get attention, something necessary for postdoc positions and grant money; partly it’s because it’s often necessary in order to do something that hasn’t been done before.  But the project had other plans.

Without going into the motivation, he wanted to conduct a deep, systematic survey for Local Group dwarf galaxies. The Local Group is a concentration of a few dozen galaxies of which our own, the Milky Way, is a member.  The dwarf galaxies he was looking for are much smaller and fainter than the Milky Way, however, and very much harder to see.  [Thus these are small, nearby galaxies, something of an oxymoron: any galaxy will have at least tens of millions of stars and lie at least thousands of light-years away.]

Since we are inside the Local Group, another member of it could lie in any direction, so he needed data covering the whole sky.  Now, the best detectors at the time were already electronic; astronomers had been using CCDs for years before the first consumer digital cameras appeared.  But CCDs did, and do, have a major drawback: it’s very expensive to make big ones.  And the sky is a very big place.  A CCD on a professional telescope at the time could image a small fraction of a degree, and there are over forty thousand square degrees in the sky.  Obviously using one for a survey would take a very long time.

But deep, all-sky data already existed.  About that time the last of the great photographic surveys was nearing completion.  Silver-halide emulsions coated on glass plates had been used to image the entire sky.  Each plate was 14 inches square (196 square inches) and covered a field about 6 1/2 degrees on a side.  With some overlap, it took 894 plates to survey the entire sky north of the equator and the same for the south.  Since each field was imaged in blue, red and infrared, the effort ran to thousands of plates and took a decade or more to complete.  But this was the last hurrah of silver-halide astronomy.  CCDs were so obviously superior for scientific purposes that even the survey telescopes were converted from analog to digital as soon as it was finished.

Then arose the question of detecting a dwarf.  Our astronomer knew what he was looking for: faint fuzzy things, looking something like the barest smudged fingerprint on the glass plate.  The new-technology way to find them would be to instruct a computer to look for just these things.  The survey data would have to be in digital form for that, but it was rapidly being digitized on laboratory machines, and was either already available or would be soon.

But putting “faint, fuzzy” into something a computer would understand was hard.  And worse, there are so many faint, fuzzy things that aren’t Local Group dwarfs, or galaxies at all. that a computer would come up with mostly false positives, each of which would have to be examined by eye.  So our astronomer gave in and examined his survey plates by eye from the beginning.

In the end, our astronomer employed, as the best tools for the purpose, an obsolescent imaging technique that was even then being dismantled, along with a detection algorithm (looking at things with his eyes) that had been abandoned in most of astronomy a century before.