How I Killed Pluto and Why It Had It Coming - Mike Brown [21]
Palomar Observatory had had a similar apparatus to Tombaugh’s in its early years, but it had been disassembled a couple of decades earlier. But even if such a thing did still exist, it would have done me no good. Because the telescope that I was using was so much more powerful than the one Tombaugh had used to find Pluto, each of my images showed one hundred times more stars, and thus would have taken one hundred times longer to have gone over by eye. Early on in the project, I calculated that to look at every star on every photographic plate by eye would have taken me forty straight years of staring into the blink comparator and slowly watching pictures of the sky go by.
Not wanting to wait forty years, and it being 1998 instead of 1930, I put the computers to work instead. First, we needed to scan the photographic plates to get them into digital form, and then the computer could do the rest. The scanning was quickly done on a big machine that already existed. Getting the computer to do the rest, though, took longer. There is no software package that looks for planets. I would have to write it myself. Though I knew nothing about emulsion and developer and fixer, this I could do. This I was good at. I had been writing little computer programs to analyze and predict and follow the stars and moons and planets in the night sky since high school. This would be the first program that actually mattered.
I spent most of that year slouched in front of a computer screen in my office, testing, scowling, starting over, typing furiously, and pondering. For someone looking for planets, I spent an awful lot of my time looking at computer code and numeric outputs instead. My nights were spent not outside staring at the sky but inside staring at numbers and computer programs and doing every test conceivable. I needed to make sure the software wasn’t going to make any mistakes. I wanted to make sure that I didn’t do anything stupid that made me miss planets that were right in front of me.
I made the computer begin by looking at a triplet of scanned photographic plates. It examined each of the little blips of light on each of the three images taken over the three nights. All of the stars in the sky, all of the galaxies, all of the nebulae, had the same coordinates on each of the three photographic plates, so the computer quickly identified them as not moving and tossed them aside. Sometimes, though, something appeared at a spot in one image where the other two images showed only blank sky. The computer took note. It could be many things. Sometimes stars in the sky get brighter and suddenly show up where they weren’t seen before. Sometimes satellites in orbit around the earth give a sudden glint that looks like a star. Sometimes dust blowing around at night sifts through the open shutter of the telescope and settles down on the photographic plate, disturbing the precarious emulsion and making something that looks vaguely like a star. But sometimes something appears where it has never appeared before because it is slowly wandering across the sky and that single picture happened to catch it momentarily in one spot. In that case, an image the next night would find it again, only a little displaced from the previous night. I used the third picture as a final check. When the computer found a third object that looked as though it could be connected to the first two, it put that object on a list of potential new wanderers and moved on to the next spot in the sky. All of this takes, of course, about a millisecond. To process our two years’ worth of images took under two hours.
So after Kevin and Jean had spent all of those nights loading and developing plates, and I had spent a year programming the computer, and the computer had spent two hours processing all of the final data, I finally had a list of all of the potential new planets to look at. I had