A few days ago, a colleague and I were working on some data we had published in a paper several months ago. But something wasn't adding up right. After some research, I found that I had made a mistake in analyzing the original data -- in matching two sets of data on the same galaxies, things got shuffled. Luckily, it only affected one of the twelve objects we were looking at, but it also meant that one group of galaxies we claimed to have discovered actually doesn't exist. Oops.
We find small errors in our work all the time. Most typos are just allowed to exist -- they don't hurt anybody. But if a typo messes up an equation, or if the publisher mixes up the figures, or if we claim to find something that doesn't exist, we have to do something about it.
For this case, I wrote a short article called an "erratum." This article just says that we made a mistake, details what the mistake was, and what changes need to be made to the conclusions because of the mistake. Thankfully, although my mistake was, in some ways, a pretty dumb one, it didn't change our conclusions much. In short, an erratum says, "Oops -- we messed up, here's the right answer."
If my mistake had been much more serious (like it had affected all twelve of the objects we talked about in our original paper), an erratum would not be enough. In a case like that, we would need to print a retraction, a short blurb saying, "Completely ignore this paper, we screwed up big time." Retractions look bad, but they do happen.
Admitting mistakes is an important part of science, as I discussed a few entries ago. Everyone makes mistakes, even in science. And, hopefully, people will forgive me for messing up here. You can bet I will be a bit more careful in the future to avoid the stupid error!